. a:
‘ 33.74%... .
. __ . I

kn.) .
1:

. u. .
Nut", 7r
'3'!»
”)1 1

31325.38. 10.2..

L "w.

 

 

1,. win-u >

my 2.3“ Ravi ..

 

l.

mass:

STATEU NIVERSITYL

IIIIIIIIIIIIIIIIIII IIIII IIIIIIIIIIIIIIIII

3 1293 O1I5I72 8698

 

 

 

 

 

 

 

 

 

 

This is to certify that the

thesis entitled

Hygrothermal Effects on the Epoxy Resins

Influenced by the Network Topology
presented by

JinKyu Choi

has been accepted towards fulfillment
of the requirements for

Master's Materials Science

degree in

Aim/W

Mahpéf$sor

May 7, 1996 //

 

 

Date

 

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

LIBRARY
Michigan State
Unlvorslty

 

 

 

PLACE ll RETURN BOX to roman this checkout from your record.
TO AVOID FINES Mum on or before duo duo.

DATE DUE DATE DUE DATE DUE

     

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

!_ ‘ ————
—_
i _———
.

Msu Is An Affirmative ActlotVEquul OppodUnlty Institution
Walla-m

 

7!

 

 

HYGROTHERMAL EFFECTS ON THE EPOXY RESINS
INFLUENCED BY THE NETWORK TOPOLOGY

By

JinKyu Choi

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

1996

ABSTRACT

The hygrothermal effects on the epoxy resins influenced by the network
topology were investigated. In this study, the 3-dimensional epoxy network
structures were altered by using di-functional, tri-functional, and tetra-functional
epoxy resins cured with a di-functional amine, respectively. The samples were
immersed in distilled water at isothermal temperature ranging from 45° C to 90° C
for over 2000 hours. Moisture absorption was monitored gravimetrically as a
function of time. Weight gain percentage was curve fitted against time by using
non-linear least-squared curve fitting from which the diffusivity and weight gain
percentage at saturation were obtained. The diffusion coefficients were found to
be inversely related to the difference between the glass transition temperature (T8)
and the conditioned temperature. The glass transition temperatures were
measured by a length dilatometer. Plasticization of the polymer was evident.
Positron Annihilation Lifetime Spectroscopy (PALS) provided a great deal of
information regarding the nature of water molecules occupying free volume
between the polymer molecules. Available hydroxyl sites in the epoxy network,
however, proved to be the dominant factor controlling the moisture absorption.
Tensile testing results showed that hygrothermal degradation significantly reduced
mechanical properties of the epoxy glass. Results from the cross-polarization lH
solid state NMR in rotating frame and quadrupole echo deuterium 1H NMR
showed that there were no physical evidence of chemical bonding between the

water molecules and the crosslink network.

 

 

DEDICATION

To

my parents

 

 

8C

the

ina

I’m“
Sper
num
WIIII
Rool
Who

TeSea

and II.

bECam

ACKNOWLEDGMENTS

Financial support for this research was provided by the Research Excellence
Fund of the State of Michigan.

Upon completion of this research, many people deserve to be thanked and
acknowledged.

I would like to thank my advisor, Dr. Andre Lee, for his guidance, support,
and most of all, abundant patience throughout this research. I also would like to thank
the members of my committee, Dr. Kalinath Mukherjee and Dr. Roger J. Morgan for
their input and guidance. I am also grateful to Dr. James P. Lucas for his generously
in allowing me to use the moisture chambers, and for his input on this research.

There are some other members of the College of Engineering at Michigan State
University whose help I deeply appreciate. Those include Dr. Kermit Johnson, who
spent a number of hours to assist me on operating and analyzing NMR spectra, and
numerous discussions with him on NMR are invaluable; Mr. Mike Rich, who was
willing to answer any questions I had regarding experimental techniques; Mr. Brian
Rook, who patiently instructed me on operating various equipments; Mr. J iming Zhou,
who shared many articles, ideas and friendship with me during the course of this
research.

I deeply thank my parents and family for their emotional and financial support,
and their never-ending encouragement throughout my education.

Last, but not least, I would like to thank my wife, Sora who “officially”

became a part of my life during this research, for her patience, love and support.

iv

 

 

 

 

2.II

Table of Content

LIST OF TABLES ...................................................................................... ix
LIST OF FIGURES ..................................................................................... x
1. INTRODUCTION ................................................................................. 1
1.1 Problems .............................................................................................. l
1.2 Outline ................................................................................................. 3
1.3 General Background ............................................................................ 5
1.3.1 “Volumetric” and “Interaction” Approach .......................................... 7
1.3.2 Spectroscopy ................................................................................... 9

1.4 Overview ............................................................................................ 10
2. MATERIALS AND SAMPLE PREPARATIONS ................................... 12
2.1 Materials ............................................................................................ 12
2.2 Sample Mixing ................................................................................... 17
2.2.1 Weight fraction of DGEBA, MY510, MY721, and D230 ................... 17
2.2.2 Mixing .......................................................................................... 17

2.3 Curing Cycle ................................................................................... 19
2.3.1 DGEBA/D230 ............................................................................... 19

 

3.!

 

3.5

4.2

4.3

2.3.2 MYSlO/D230 ................................................................................ l9

 

2.3.3 MY721/D230 ................................................................................ 20

2.4 Sample Storage .............................. 20
3. MOISTURE ABSORPTION ................................................................ 21
3.1 Diffusion ............................................................................................ 21
3.2 Sample Preparations .......................................................................... 23
3.3 Experimental Procedure ..................................................................... 23
3.4 Results and Discussion ....................................................................... 26
3.4.1 Sorption behaviors ......................................................................... 26
3.4.2 Thickness effects on the sorption behavior ....................................... 36
3.4.3 Hydroxyl Sites Dependency ............................................................ 42

3.5 Summary ............................................................................................ 44
4. MOISTURE EFFECTS ON PHYSICAL PROPERTIES ......................... 45
4.1 Dilatometry ........................................................................................ 45
4.2 Experimental Procedure ..................................................................... 46
4.3 Results and Discussion ....................................................................... 52
4.3.1 Glass Transition Temperature (Ts) .................................................. 52

vi

4.3.2 Coefficients of Thermal Expansion (CTE) ........................................ 62

4.4 Positron Annihilation Lifetime Spectroscopy (PALS) ......................... 67
4.4.1 Background ................................................................................... 67
4.4.2 PALS ............................................................................................ 69
4.4.3 Experimental Procedures ................................................................ 71
4.4.4 Results and Discussions .................................................................. 71

4.5 Summary ............................................................................................ 76

5. MOISTURE EFFECTS ON MECHANICAL PROPERTIES ................... 78

5.1 Experimental Procedures .................................................................... 78

5.2 Samples .............................................................................................. 78

5.3 Results and Discussions ...................................................................... 79
5.3.1 Oxidation and Degradation ............................................................. 79

5.4 Summary ............................................................................................ 92

6. NUCLEAR MAGNETIC RESONANCE SPECTROSCOPY ................... 93

6.1 Introduction ....................................................................................... 93

6.2 The Spin-Lattice Relaxation Time (T 1) ............................................. 104

6.3 Cross-Polarization (CP) .................................................................... 106

vii

In

 

6.!

7.0

 

8. R“

6.4 Proton Spin-Lattice Relaxation Times in the Rotating Frame ........... 111

 

6.4.1 The rotating frame ....................................................................... 111
6.4.2 Proton spin-lattice relaxation times in the rotating frame ................. 116

6.5 Sample Preparation for the T c” and H T 1,, Experiment ....................... 117
6.6 Experimental Procedure ................................................................... 117
6.7 Results and Discussion ..................................................................... 118
6.7.1 lH Spin-Lattice Relaxation Times in the Rotating Frame ................. 118

6.8 Quadrupole Echo Deuterium NMR Experiment ................................ 124
6.8.1 Results and Discussion ................................................................. 125

6.9 Summary .......................................................................................... 129
7. CONCLUSIONS .............................................................................. 130
8. REFERENCES ................................................................................ 133

viii

LIST OF TABLES

Table 2-1: The stoichiometric mixing ratio of epoxies and a diamine. ................ 18

Table 3-1: Changes in M... and diffusivity due to the functionality difference and
moisture absorption. ............................................................................... 30

Table 3-2: Activation energy and diffusion coefficients of the epoxy systems ..... 35
Table 3-3: Thickness effects on M... and D. ...................................................... 41
Table 3-4: Number of hydroxyl groups in each epoxy ....................................... 43

Table 4-1: Change of the glass transition temperatures due to moisture saturation.53

 

Table 4-2: Correlationship between AT = T, — T" and the diffusivities of the epoxy

glasses. .................................................................................................. 55

Table 4-3: Thermal Expansion Coefficients of DGEBA/D230, MY510/D230, and
MY721/D230. ........................................................................................ 64

Table 4-4: Thermal Expansion Coefficient of DGEBA/D230, MYS 10/D230, and
MY721/D230 after the moisture has been desorbed at 100°C for 3 days. 66

Table 4-5: 2'3 and 13 results on epoxy systems. ................................................. 74

Table 4-6: Conditioned-temperature effects on the free volume. Presented here

are data obtained on DGEBA/D230 epoxy network ................................... 75
Table 5-1: Tensile testing results on DGEBA/D230 epoxy systems. .................. 84
Table 5-2: Tensile testing results on MYSlO/D23O epoxy systems. ................... 85
Table 5-3: Tensile testing results on MY721/D230 epoxy systems. ................... 86
Table 6-1: Tc” and ”LP of DGEBA/D230 ..................................................... 119
Table 6-2: Ta. and ”T1,, of MYSIO/D230 ...................................................... 120
Table 6-3: Tc” and ”2",, of MY721/D230 ...................................................... 121

ix

 

 

 

Figt
FigL

Figt

Figu

LIST OF FIGURES

Figure 2-1: Diglycidyl ether of bisphenol A (DER 332) .................................... 13
Figure 2-2: Diglycidyl ether of para-aminophenol (Araldite MYSlO) ................. 14
Figure 2-3: N,N,N’,N’,-Tetraglycidyl—4,4’-methylenebisbenzenamine

(Araldite MY721) ........................................................................ 15
Figure 2-4: The chemical structure of diamine. ................................................ 16

Figure 3-1: The variation of M/M... with time. Exact (solid line): Eq. 3-4;
approximate (dotted line): Eq. 3-5. ............................................... 24

Figure 3-2 (a): Moisture absorption curve of DGEBA/D230. Samples were
saturated with water at 45°C, 60°C, 75°C, and 90°C for 2254 hours.
Lines represent theoretical curves. Symbols represent experimental
data at different temperatures. ...................................................... 27

Figure 3-2 (b): Moisture absorption curve of DGEBA/D230. Samples were
saturated with water at 45°C, 60°C, 75°C, and 90°C for 2254 hours.
Lines represent theoretical curves. Symbols represent experimental
data at different temperatures. ...................................................... 28

Figure 3-2 (c): Moisture absorption curve of DGEBA/D230. Samples were
saturated with water at 45°C, 60°C, 75°C, and 90°C for 2254 hours.
Lines represent theoretical curves. Symbols represent experimental
data at different temperatures. ...................................................... 29

Figure 3-3 (a): InD vs. l/T curve of DGEBA/D230, where D is the diffusivity
(cmZ/sec) and T is the temperature (K). ......................................... 32

Figure 3-3 (b): lnD vs. l/T curve of MYSlO/DZBO, where D is the diffusivity
(cmZ/sec) and T is the temperature (K). ......................................... 33

Figure 3-3 (C): MD vs. 1/T curve of MY721/D230, where D is the diffusivity
(cmz/sec) and T is the temperature (K). ......................................... 34

Figure 3-4 (a): Sorption curves for DGEBA/D230 with different thickness; 81 and
82. Hollow symbols represent thick samples (S1). Filled symbols
represent thin samples (S2). ......................................................... 38

 

Figure 3-4 (b): Sorption curves for MYS 10/D230 with different thickness; 81 and
S2. Hollow symbols represent thick samples (S1). Filled symbols
represent thin samples (82). ......................................................... 39

Figure 3-4 (c): Sorption curves for MY721/D230 with different thickness; 81 and
82. Hollow symbols represent thick samples (SI). Filled symbols
represent thin samples (82). ......................................................... 40

Figure 4-1: A length dilatometer built by Omega Co. ....................................... 48

Figure 4-2: Strain versus Temperature (°C) of the DGEBA/D230 epoxy resin.
Samples have been saturated with distilled water at 45°C, 60°C, 75°C,
and 90°C for 2254 hours. ............................................................. 49

 

Figure 4-3: Strain versus Temperature (°C) of the MY510/D230 epoxy resin.
Samples have been saturated with distilled water at 45°C, 60°C, 75°C,
and 90°C for 2254 hours. ............................................................. 50

Figure 4-4: Strain versus Temperature (°C) of the MY721/D230 epoxy resin.
Samples have been saturated with distilled water at 45°C, 60°C, 75°C,
and 90°C for 2254 hours. ............................................................. 51

Figure 4-5: Change of the strain between the dry and moisture desorbed samples
of the DGEBA/D230 epoxy resin were plotted against temperature
(°C) ............................................................................................ 59

Figure 4-6: Change of the strain between the dry and moisture desorbed samples
of the MYSIO/D230 epoxy resin were plotted against temperature
(°C) ............................................................................................ 60

Figure 4-7: Change of the strain between the dry and moisture desorbed samples
of the MY721/D230 epoxy resin were plotted against temperature
(°C) ............................................................................................ 61

Figure 4-8: Schematic illustration of the variation of the specific volume, V, of a
polymer with temperature, T. The free volume is represented by the
shaded area. (Adapted from R. J. Young, Introduction to Polymers,
1st ed., 1981.) ............................................................................. 68

Figure S-l: Comparison of the color change between dry and moisture saturated
samples. As the soaking temperature increases, the color of the
sample becomes darker. Shown here are DGEBA/D230 epoxy resins.80

xi

 

 

 

 

Figure
Figure

Figure
Figure
Figure

Figure

Figure I

Figure 6

Figure 6

Figure 5-2

Figure 5-3:

Figure 5-4:
Figure 5'-5:
Figure 5-6:

Figure 6-1:

Figure 6-2:

Figure 6—3:

Figure 6-4:

Figure 6-5:

Figure 6-6:

Comparison of the color change between dry and moisture saturated

samples on MYS 10/D230 epoxy resins. ......................................... 81
Comparison of the color change between dry and moisture saturated

samples on MY721/D230 epoxy resins. ......................................... 82
Change of °/o strain at break (%) are shown here. ........................... 89
Change of load at break (kN) are shown here. ................................ 90
Change of Young’s modulus (MPa) are shown here. ....................... 91

A spinning proton generates a magnetic field, called its magnetic
moment. This magnetic field (H) resembles that of a small bar
magnet. Adapted from [43]. ........................................................ 94

An external magnetic field (11.) applies a force to a small bar magnet,
twisting the bar magnet to align it with the external field. The
arrangement of the bar magnet aligned with the field is lower in
energy than the arrangement aligned against the field. Adapted from
[43]. .......................................................................................... 95

In the presence of an external magnetic field (Ho), a proton has a spin
quantum number (I) of either +1/2 or -1/2. The state with spin I =
+1/2 is called the a-spin state, aligned with the external magnetic
field. The state with spin I = -1/2 is called the B-spin state, aligned
against the external field. The B-spin state is higher in energy than
the a-spin state. Adapted from [43]. ............................................ 97

The nuclear spin generates a small magnetic field, and in the absence
of an applied magnetic field, the orientation of these dipoles is random
(a). When a sample is placed in a homogenous magnetic field, the
dipoles will align themselves with the lines of induction or force of the
applied magnetic field (b). However, thermal motion does not allow
the alignment to remain, and the average alignment is at a small angle
to the magnetic field (c). The magnetic moment precesses about the
magnetic field at the Larmor frequency (d). Adapted from [43]. 100

Quantized energies of nuclei in a magnetic field. Adapted from [43]. 102

A nucleus in “in resonance” when it is irradiated with radio-frequency
photons having energy equal to the energy difference between the spin
states. Under these conditions, a proton in the a-spin state can

absorb a photon and change to the B-spin state. Adapted from [44 ]. 103

xii

“UNI-i“:

 

 

 

 

Figure

Figure

Figure

Figure

Figure

Figure

Figure i

Figure 6-7: The timing sequence for the cross-polarization experiment. Adapted
from [43]. ................................................................................. 109

Figure 6-8: Diagram to show the motion of a spinning nucleus in a magnetic field
of direction z. Adapted from [48]. ............................................. 113

Figure 6-9: Changing coordinates to the rotating frame: (a) rotation of magnetic
moment it about laboratory coordinates; (b) change of coordinates to
rotating frame and application of a radio-frequency induced magnetic
field along x’; (c) after application of field H1 along x’ (note the
vector M moves through the z’y’ plane by 90° to become colinear
with y’); (d) after changing the phase of the irradiation field so that
H1 is colinear with M and y’. Adapted from [48]. ....................... 115

Figure 6-10: The change in carbon magnetization with contact time for the cross-
polarization experiment. The initial rise is due to the cross-
polarization contact time, T CH, and the relaxation decrease is
governed by the H T ,p .................................................................. 122

Figure 6-11: Quadrupole echo deuterium 1H NMR spectrum of D20. The sharp
peak at 616000 is indicating the chemical shift of pure deuterium. 126

Figure 6-12: Quadrupole echo deuterium 1H NMR spectra of samples saturated
with D20 at 90°C; (a) DGEBA/D230; (b) MY510/D230; and (c)
MY721/D230. ........................................................................... 127

Figure 6-13: Quadrupole echo deuterium 1H NMR spectra of D20 desorbed

samples (100°C for 4 days); (a) DGEBA/D230; (b) MYSlO/D230; and
(c) MY721/D230. ...................................................................... 128

xiii

'l

 

 

inatn
[0 ex
mech
to nu
envuc

behav

nuustt
and pa
efiécr

interac
behavi
conten

I0 the

1.1 Pr

1. INTRODUCTION

Epoxy resins are widely used as structural adhesives and composite
matrices in both the aerospace and automotive industries, and are ofien exposed
to extreme environments. Such service environment raises some concern that the
mechanical properties of such materials may suffer when the material is exposed
to moisture for long periods of time. The durability of epoxies in such service
environments is difficult to predict without a knowledge of the structure, and

behaviors of deformation and failure.

Many investigation have been conducted to address various aspects of
moisture induced behavior in part to understand the structure [1, 4~40, 61, 62]
and partly to understand the behaviors of mechanical response due to the moisture
effect [56, 58, 59, 60]. Until now, the explanations on (i) how the moisture
interact with the epoxy network and (ii) how its structure and mechanical
behavior were influenced by the moisture were based upon varying the amine
content. Moisture uptake was monitored as a function of the amine’s mole ratio

to the epoxy, of the different amine types, or of the curing time.

1.1 Problems

Most of the sorption kinetics studies have been performed either by varying
the amine content, or by changing cure conditions. Synthesizing epoxides and

hardeners with off-stoichiometric ratio will produce excess amount of either

 

 

epoxh
their 5
group:
until t
temper
other I
absorb
These

bondin

0f the
bISphei
diamini
Slates l
PDA is

Would .

extent.
they Orj

MOrel .

epoxides or hardeners in the system. Carfagna et a1. discussed this problem in
their studies [11]. If excess amount of epoxide is present, the secondary hydroxyl
groups, formed in the previous reaction, could add to the remaining epoxide rings
until the glass transition temperature of the forming network exceeds the curing
temperature and the unfavored reaction becomes diffusion controlled. On the
other hand, when the resin is crosslinked with a large excess of hardener, the resin
absorbs more water than that crosslinked with stoichiometric amount of hardener.
These findings suggest that the presence of unreacted amines provide hydrogen

bonding sites, resulting in increased moisture absorption.

Another effect discussed by Diamant et a1. [4] is due to the resin polarity
of the hardener. In order to increase crosslink density, diglycidyl ether of
bisphenol A (DGEBA) was cured with m-phenylenediamine (m-PDA) and parts of
diamine was replaced with aniline. The m-PDA molecule has more resonative
states which allocate a higher negative charge on the benzene ring such that m-
PDA is highly soluble in water whereas aniline is almost insoluble. Thus, m-PDA

would attract moisture more than aniline.

Curing condition is another contributing factor in moisture absorption
behavior. Danieley et a]. [10] found that the water uptake increases with the
extent of cure, and concluded that hydroxyl groups play a predominant role since
they originate from epoxide-amine reactions. However, a question was raised by

Morel et a]. [7] that epoxide-epoxide and epoxide-alcohol side reactions in

 

 

TGNI
obtair

CXlClll

the dc
crossl
polyrr
affect
sorpti
water
crossl
shes
amour

COME

1.2

maIEr

hydro
differ:

h. 1 <<

TGMDA-DDS system are also important, because the higher degree of cure is
obtained with a relatively low hardener concentration. Thus, it is to a certain
extent difficult to speculate on the hydroxyl build-up in the final stage of cure [7].

Moy and Karasz [12] also argued the moisture absorption dependency on
the degree of curing. Curing at higher temperatures is expected to increase the
crosslink density as well as increasing the number of hydroxyl groups in the
polymer such that the equilibrium sorption of water in the resin is markedly
affected by the curing conditions. It was found that a decrease in the rate of
sorption with increasing cure temperature suggests that sorption and transport of
water vapor in the glassy state of the epoxy is predominantly controlled by the
crosslink density rather than by the increase in the number of possible sorption
sites. These arguments, however, ignore a possibility that uncured or excess
amount of hardener would react with water resulting in an increased moisture

content in the epoxy resin.

1.2 Outline

Chemical structure, processing conditions, and sample storage of the

materials were presented in Chapter 2.

In Chapter 3, the moisture sorption behaviors in relation to the available
hydroxyl sites were investigated. Moisture uptakes of the materials under
different hygrothermal conditions were monitored as a function of time. Since the

h/I << 1 and h/w << 1 (h is the thickness, 1 is the length, and w is the width of the

 

 

smnph
on Ihl
vanou

vanau

(PAd.S
mecha
intensi
PIJ.S_

expans

audnc
Prepare
fishnet
temper
WOISEn

dncusg

"Won

Ihk Stu

sample), l-dimensional diffusion was assumed, neglecting “edge effects”. Based
on the solution of Fick’s second law, diffusion coefficients of the samples at
various conditioning temperatures were obtained. The effect of the thickness

variation on the sorption behavior was also discussed.

In Chapter 4, by using the positron annihilation lifetime spectroscopy
(PALS) and length dilatometry, moisture effects on the change of physical and
mechanical properties were discussed. Measurements of the free volume size and
intensity change of the free volume upon moisture absorption were measured by
PALS. Change of glass transition temperatures and coefficient of thermal

expansion were measured by the dilatometer.

In Chapter 5, the mechanical response due to the moisture absorption was
studied by using a screw-driven Instron tensile testing machine. Few tables were
prepared to report tensile testing results. After prolonged immersion in the
distilled water at elevated temperatures, surface discoloration was evident. As the
temperature of the immersion increased, the degree of discoloration became
worsen. Reasonings in conjunction with the effect of chemical degradation were

discussed.

To conduct a microscopic investigation on the moisture in the epoxy
networks, nuclear magnetic resonance spectroscopy was employed and its results
were presented in Chapter 6. It is noteworthy that the NMR technique used in

this study was 13C NMR spin-lattice relaxation times supplemented by cross-

 

poIariz.
moistur
echo dc

on the r

results (

1.3 Get

perform:
int'oIve ‘
interacti.
rthponse
be safe t

IUndredg

GI
mOISIUre
different
W with .
ltVeaICd ‘

polarization. Delay of the relaxation time, H T ,p, and its relationship to the
moisture mobility in the epoxy network were discussed. In addition, a quadrupole
echo deuterium NMR study was also conducted to provide additional information

on the nature of the moisture in the epoxy networks.

The last chapter, Chapter 7, contains conclusions of the experimental

results of this study.

1.3 General Background

The sorption of solvent in polymeric composites and their effects on the
performance of the composites are highly complicated issues. Their studies
involve the implications of (i) the polymer science; focusing on molecular-level
interactions, and (ii) the applied mechanics; mainly dealing with mechanical
response. An extensive review of this subject is a profound task such that it may
be safe to assume that the total number of papers regarding this subject run into

hundreds, maybe even thousands.

Gupta, Drzal, and Rich [5] conducted an experiment characterizing the
moisture transport through a di-functional epoxy resin (DGEBA) cured with
different amounts of metaphenylene diamine (m-PDA), using two cure cycles i.e.,
one with standard cure and the other with additional post cure. The sorption data
revealed that moisture sorption proceeded with having two stages. The initial

diffusion-controlled linear region showed considerable temperature dependence.

 

 

Thet
hghh
sauufl
less d
appea
taupe
inohu
that l
deterr
i.e., at
in ton
on th.

dufere

condu.
bhphe
fihers
nonhn.
Pout.
Coeffic

The temperature dependence of the initial region was relatively small for the most
highly crosslinked, high T 3 samples and was high for the less crosslinked, low T8
samples. At longer times, however, the moisture uptake appeared to be rather
less dependent on temperature. The temperature dependence in this region also
appeared to be large for samples of low crosslink density, low T3. At room
temperature, the most highly crosslinked samples absorbed the maximum
moisture, while at 100°C they absorbed the least. From these findings, they noted
that the specific volume and the free volume played a. important role in
determining the moisture absorption characteristics of the post-cured samples,
i.e., at room temperature, there were small differences in specific volume resulting
in considerable differences in their diffusion coefficient. At higher temperature,
on the other hand, large differences in specific volume lead to only small

differences in diffusion coefficient.

Similar results can be found in other papers. De’Néve and Shanahan [19]
conducted an experiment of water sorption based on the diglycidyl ether of
bisphenol A (DGEBA), crosslinked with dicyandiamide (DDA), and contained
fillers (Ciba-Geigy X83131). At the start of water uptake, the curve was
nonlinear and rather it was more like sigmoidal in shape with a single inflection
point. This type of curve was explained by using a history-dependent diffusion
coefficient. Rising penetrant concentration was assumed to lead to an immediate

increase in diffusion coefficient, followed by a slow drift towards an equilibrium

 

 

nhea
noted t
dependt

the sigr.

an ofis
insuffici
undergc

oxidauc

showed
“etertl
Weight I
belWeer
0neis d
and the

"OIUmii

value as a result of a relaxation process characteristic of the glassy state. They
noted that the value of the diffusion coefficient in any element of the system
depended on the concentration history of the element. They also concluded that

the sigmoidal behavior was evident in the case of an excess of hardener.

Wong and Broutman [45, 46] also discussed in their papers that there was
an offset between the theoretical and experimental diffusion curves for an
insufficiently cured epoxy network (DGEBA/m-PDA) and was caused by the resin
undergoing firrther postcuring in the sorption process. They also noted that

oxidation of the resin during sorption also caused the offset between the curves.

Moy and Karasz [12] found that the sorption curve they have constructed
showed a non-linearity during the initial sorption process. The diffusion of the
water through the resin is found to increase with the concentration of water from
weight gain rates at various partial pressures. Their reasoning for the mismatch
between the theoretical and experimental diffusion curves consists of two effects;
one is due to the irreversible chemical reaction taking place between the polymer
and the water, forming hydrogen bonds. The other is due to decoupling of the

diffusion to a relaxation process at T g.

1.3.1 “Volumetric” and “Interaction” Approach

Besides the arguments on the kinetics of fluid sorption, how the moisture
resides within the epoxy network is another controversial issue; one is the

“volumic” approach and the other is the “interaction” approach.

 

which

molecu
that it
microvt
free Vt
interact
has beer
the avai
microvc

ofiff pft

are nei
morphO]
hidmph
lites ot
that the

OIIhe e;

stoichim

The free volume, by definition, is the space in a solid or liquid sample
which is not occupied by polymer molecules, i.e. the ‘empty-space’ between
molecules [50]. Researchers based upon the free volume approach are claiming
that the absorbed moisture takes up the free volume, ‘empty-space’ and
microvoids, and is relatively free to travel through the free-volume voids [6]. The
free volume approaches, however, often neglect the existence of specific
interactions between water and hydrophilic sites of the network. For instance, it
has been suggested that the water equilibrium concentration is mainly governed by
the available free volume [6, 8], or that water molecules occupy essentially the
microvoids, internodular areas and other morphological defects [4, 5] which can

offer preferential ways for their diffusion.

The “interaction” approaches, on the other hand, state that water molecules
are neither distributed randomly in the network nor concentrated in its
morphological defects, but rather linked by strong hydrogen bonds to some
hydrophilic loci, mainly hydroxyls [10] or amines [11]. The contribution of both
types of groups could be interdependent [7]. Some researchers also suggested
that the degree of curing takes an important role in the morphology and chemistry

of the epoxy-water system.

Carfagna et a1. [11] conducted a sorption experiment by altering
stoichiometric ratio of amine and found that the epoxies with the excess amount

of amine had higher equilibrium water uptake level than the epoxies with less

 

amount of amine. It was proposed that the presence of unreacted amines
contributed to increasing the equilibrium water uptake level by providing more

hydrogen bonding sites.

It is noteworthy that in the volumic approach, the behavior of the water-
epoxy system is mainly controlled by the physical state of the polymer, while in
the interaction approach, the chemical structure plays a dominant role and often
this approach was accounted for by spectrometric observations such as FT-IR [13,

4o, 64] and NMR[9, 12, 41, so, 51].

1.3.2 Spectroscopy

Using FT-IR spectroscopy, Ghorbel and Valentin [13] revealed that there
was a modification in the intensities and shape of the peaks in polyester and
vinylester resins due to reactions such as the decrease in the styrene content and
the hydrolysis. In addition, Koenig [43] showed that the frequency of the in-plane
bending mode of water in epoxy resins lies between the frequencies of liquid and
mobile water proposing that the sorbed water was clutched within the epoxy resin

by hydrogen bonding.

Nuclear magnetic resonance is another sensitive technique of probing the
molecular interaction between two components. Moy and Karasz [12], in their
results, suggested that at low concentration, the water may be interacting with
adjacent bond sites so that not only translational but rotational movement

becomes hindered as well. At higher concentrations, however, secondary layers

 

 

of wat
fimflu
wmon
became
nomdt
whet
mednt
thata t
than ag
present
Karasz
interact
that (1]
Water,

IhWat

1-4 ()1

10

of water molecules were able to rotate freely and line narrowing was observed.
Similar results were reported by Schadt and Vanderhart [41] by using solid state
proton NMR. It was noted that the at higher water concentrations, linewidth
became narrower, indicating that the water mobility was increased. It was also
noted that they found no evidence of a univorm distribution of crosslinds on a
scale, greater than 4 nm, which possessed a mobility gradient. This suggests that
the diameter of a void is no larger than 40 angstroms. In conclusion, they stated
that a majority of water molecules were molecularly dispersed in the epoxy rather
than aggregated in a water like state within voids, suggesting that if voids are

present, they can only accommodate one or two molecules. In different papers,

Karasz and co-workers [50, 51] reported the nature of the epoxy-water molecule
interactions using quadrupole echo deuterium NMR spectroscopy. They stated
that (i) the movement of water is impeded in epoxy resin; (ii) there is no free
water; (iii) there is no evidence for tightly bound water; and (iv) it is unlikely that

the water disrupts the hydrogen-bond network in the epoxy resin.

1.4 Overview

In this study, instead of varying amine contents or tempering with the
curing conditions, the entire experiments were carried out based upon varying the
functionalities of the epoxy resins only, such that di-, tri-, and tetra-functional
epoxides were cured with an amine terminated PPO (Jeffamine D230, designation

of Texaco Chemical Company). Also, by having the mole ratio of the epoxide to

 

annne

quy c
tno cc
concer
(n) th
monon
inconu
enable

the ep.
dhfusic

morphc

ll

amine to that of the stoichiometric ratio, it was assumed that the epoxies were
fully cured pertaining the highest crosslink density as much as possible. These
two conditions, as well as using more flexible hardener like PPO, eliminated
concerns evolved from (i) the further curing during the hygrothermal treatment,
(ii) the interaction between the water molecules and an excess amount of
monomers, (iii) diffusion controlled secondary amine reaction, and (iv) an
inconsistent free volume distributions. It is believed that these conditions will
enable a thorough investigation on the physical nature of the water absorption in
the epoxy networks. The moisture absorption behavior, characterization of
diffusion process, moisture effects on physical and mechanical properties, and

morphological characterization by using solid state NMR were also presented.

2.]

dia:
are
Che
phe
epo:
etos
desil

Epot

2. MATERIALS AND SAMPLE PREPARATIONS

2.1 Materials

Sample having different polymer network topologies cured with PPO
diamines have been investigated. The commercially available epoxy resins studied
are based on di-functional pure diglycidyl ether of bisphenol A (DER 332, Dow
Chemical, USA); tri-functional low viscosity epoxy resin based on para amino
phenol (Araldite MYSlO, Ciba-Geigy Corporation); and tetra-functional liquid
epoxy based on methylene dianiline (Araldite MY721, Ciba-Geigy Corporation)
crosslinked with an amine terminated poly(propylene oxide) (Jeffamine D230,
designation of Texaco Chemical Company). The chemical structures of the three

epoxies mentioned on the above are described in Figure 2-1 through Figure 2-4.

12

l3

 

 

H26

\.._.._.ti._.. _../ \..

Figure 2-1: Diglycidyl ether of bisphenol A (DER 332)

Fig:

l4

 

 

 

o ——cr+,. —CH\ /cr12
O
"\
CH; ca -——CH2/ CH, —CH on,

V \0/

Figure 2-2: Diglycidyl ether of para-aminophenol (Araldite MYSIO)

15

\/
/\

O

/\

CH2

 

30
ll

Hzc—CI'I

Figure 2-3: N,N,N’,N’,-Tetraglycidyl-4,4’-methylenebisbenzenamine
(Araldite MY721)

l6

 

 

 

HzN—CH

CH2 +0

CH2

X
CH3 CH3 X = 2.6

Figure 2-4: The chemical structure of diamine.

J effamine D-230 Polyoxypropylenediamine

LZSam

L;
DI
If
If
DZ
22
Em
Udhmme
stoichiome
ungagm
and lIlC mi:
TM mnw
I”literature

euhnmme

17

2.2 Sample Mixing

2.2.1 Weight fraction of DGEBA, MY510, MY721, and D230

DGEBA: 170 g / epoxy group (e.g.)

MY510: 100 g / e.g.
MY7212110 g / e.g.
D230: 57.5 g / e.g.

2.2.2 Mixing

Epoxies (DGEBA, MY510, and MY721) were mixed with diamine
(Jeffamine D230) and the mole ratio of epoxide to amine was that of
stoichiometric ratio, as shown in Table 2-1. After the weights were measured by
using a gravimetric balance with 0.1 mg resolution, diamine was added to epoxies
and the mixtures were stirred with wooden stirrers until they became clear liquid.
The mixtures were then placed in a vacuum oven for degassing at room
temperature for 30 to 40 min. Weighing, mixing, degassing were performed on

each sample, separately.

m

==§
DGEE

 

18

Table 2-1: The stoichiometric mixing ratio of epoxies and a diamine.

 

 

 

 

 

Epoxies g/epoxy group Total weight (g) Wt. fraction of Wt. fraction of
(epoxy+diamine) D230 epoxies
DGEBA 170 170 + 57.5 57.5 / 227.5 170 / 227.5
= 227.5 = 0.2527 = 0.7473
MYSIO 101 101+ 57.5 57.5/ 158.5 101/158.5
= 158.5 = 0.3628 = 0.6372
MY721 110 1]0+57.5 57.5/ 110 110/ 167.5
= 167.5 = 0.3433 = 0.6567

 

 

 

 

 

 

hou

the

moh
nnxt
degn
tnno
nhco
and 6
24 ho

the“

hours I
the dia
SIOIChic
homOge

30to 4(

19

2.3 Curing Cycle

2.3.1 DGEBA/D230

The DGEBA, di-functional epoxide monomer, was preheated at 60°C for 2
hours to melt any crystals present. The amine was then added to the epoxide and
the mixture was stirred by a wooden tongue until it becomes a clear mixture. The
mole ratio of the epoxide to amine was that of the stoichiometric ratio. The
mixture was degassed at room temperature for 30 to 40 minutes, depending on the
degree of the bubbles present on top of the mixture’s surface, under vacuum to
remove all the air bubbles as much as possible. Then, the mixture was cast into
silicon molds with dimensions of 7.62 cm x 0.32 cm x 1.27 cm rectangular strips
and 6.0 cm x 0.16 cm x 1.0 cm tensile testing coupons, and cured at 100°C for
24 hours. After the drying cycle, the molds were removed from the oven and

allowed to cool slowly at room temperature over night.

2.3.2 MY510/D230

The MY510, tri-functional epoxide monomer, was preheated at 60° C for 2
hours to melt any crystals left in the epoxide. Then the epoxide was mixed with
the diamine, D230, where the mole ratio of the epoxide to amine was that of the
stoichiometric ratio. The mixture was stirred by a wooden stirrer until it became
homogeneous viscous liquid. The mixture was degassed at room temperature for

30 to 40 minutes under vacuum to allow all the bubbles to escape. The mixture

was
of t
curi

tern

2 ht
epo:
epoa
at l'(
for t

alibi

2.4

20

was then cast into a silicon mold and cured at 130° C for 24 hours. Dimensions
of the specimens were identical to that of the DGEBA/D230 specimens. After the
curing, the mold was removed from the oven and allowed to cool slowly at room

temperature over night.

2.3.3 MY721/D230

The MY721, tetra-functional epoxide monomer, was preheated at 60°C for
2 hours to melt any crystals that might be trapped in the epoxide. The MY721
epoxide was then mixed with diamine (Jeffamine D230). The mole ratio of the
epoxide to amine was that of the stoichiometric ratio. The mixture was degassed
at room temperature for 30 to 40 minutes under vacuum, then cast into the mold
for curing at 157° C for 24 hours. The molds were removed from the oven and

allowed to cool slowly at room temperature over night.

2.4 Sample Storage

After the curing cycle, the samples were stored in zipploc bags at room
temperature until the experiments were begun. Before the moisture experiment

began, the samples had been dried in a vacuum oven at 60° C for 17 days.

3.1 Dif

the hype
it sectior

section, '

Where J
concentra

comPOrtet

As
volume I
Therefme
cOmPOUCn

Mass Con,

3. MOISTURE ABSORPTION

3.1 Diffusion

The mathematical theory of diffusion in isotropic substance is based upon
the hypotheses that the rate of transfer of diffusing substance through unit area of
a section is proportional to the concentration gradient measured normal to the

section, i.e.

J, = -D,VC,.

Eq. 3-1

where J is the rate of transfer per unit area of section by component i, C the
concentration of diffusing substance, and D is called the diffusion coefficient of

component i [47].

Assume that the surface is a closed, continuous surface that encloses a
volume V. Classically, we assume that mass is not created or destroyed.
Therefore, mass flowing out of the surface must result in a decrease in the mass of
component i, in the volume V enclosed by the surface. This is called the Law of

Mass Conservation. This assumption leads to a mass continuity equation, i.e.

div] = -§—'-
0'1
Eq. 3-2

21

wh

can

Ian

squt

22

where t is time.

If the diffusion coefficient is constant, D at D(C), and if there is a gradient

of concentration only along the x-axis, by substituting Eq. 3-2 into Eq. 3-1, one
can discuss the diffusion phenomena by using a simple solution of Fick’s second

law as

 

2
gzDaC
a 63:2

Eq. 3-3.

For diffusion occurs through both sides of surface, as depicted below,

solution of Eq. 3-3 can be written as

M, 8 °° I_(2i+1)2 3%)]: 00)

no IZ" i=0 (21+1)2

 

 

 

Eq. 3-4

where M

saturated

Eq

and the C(
the apptot

non-linear

3.2 Samp:

Afte
obtain Ore:

for 17 days

33 EXper

The
chambers 0
MOISIUre Up
mileufluu‘

WeIinl 831 [I

23

where M. is the initial diffusant content absorbed in the system at t=t, M... the

saturated diffusant content at t=ao, and h is the thickness of the sample.

Eq. 3-4 can be approximated [l] by the expression

0.75
M‘ = 1— exp —7.3(221)
M h

d)

 

= th)

 

Eq. 3-5

and the comparison between Eq. 3-4 and Eq. 3-5 is shown in Figure 3-1. Since
the approximate curve is in well agreement with the theoretical curve, by using

non-linear least-square curve fitting, the diffusivity can be obtained.

3.2 Sample Preparations

Alter being discharged from the molds, samples were carefully polished to
obtain overall flat surface. Samples then dried in an oven at 60°C under vacuum

for 17 days to achieve dry states from which reference points were collected.

3.3 Experimental Procedure

The specimens from each epoxy system were immersed in distilled water
chambers of temperatures set at 45°C, 60°C, 75°C, and 90°C for 2254 hours. The
moisture uptake was measured periodically by using an analytical balance with 0.1
mg resolution. Assuming that weight gain was due to the moisture uptake, the

weight gain percentage was obtained by Eq. 3-6,

Git)

HItI
-- 0

Figure 3.1

24

 

0.8 ”
Approximation, H(t)

 

    
  
 

 

 

0.6 2
(9(1)
H(t)
-- 0.4 —
Exact, (3(1)
0.2 —
0 — I I
0.001 001 0.1 l
D-t
l12

Figure 3-1: The variation of M/M... with time.

approximate (dotted line): Eq. 3-5.

Exact (solid line): Eq. 3-4;

nhere

isthe:

flucknt

Thethi

(sec) n
hnearlt
3~5 to

tabulate
fithng v
and to 1

gUESSIng

IPPTOxin
Slopes OI

Was Dion

25

 

M=m'mxwo

Eq. 3-6

where M is percentage weight gain (%), W, weight of the sample at time t, and W,

is the initial weight of the sample obtained from dry samples.

In addition, during sorption experiment, samples with two different
thicknesses were used to investigate the thickness effect on the sorption process.

The thicknesses were 0.32 cm and 0.159 cm, denoted as S] and 82, respectively.

After sufficient amount of data were collected, % weight gain versus time
(sec) moisture absorption profiles were plotted, as shown in Figure 3-2. Non-
linear least-squared curve fitting was performed on experimental data by using Eq.
3-5 to obtain Mas and diffusion coefficients of the samples. The results were
tabulated in Table 3-1. The reason for using the non-linear least-squared curve
fitting was to use the actual data as much as possible, to obtain the diffusivities
and to let M... to be decided mathematically by the curve fit value, rather than

guessing from curves.

Conventional method of obtaining diffusivities was to use a short-time
approximated diffusion equation from which diffusivities were obtained by taking
slopes of the linear region of the sorption curves where weight gain percentage

was plotted against t”). This method, however, tends to ingore data points afier

the lit

the so

and di
spread
the en

paramt

3.4 R.

enntet
reahstn
TabIe 3
Interse:
seem,
pIOCeSS
crOSSIIn
soaking

26

the linear region, using only a portion of data. Also, one must assume M... from

the sorption curves in order to obtain the diffusivity.

Non-linear least-squared curve fitting, however, allows one to obtain M...
and diffusivity right out of the approximation by solving the equation, aided by a
spread sheet software program. Therefore, this method uses not a portion of but
the entire range of data. One must be careful, however, choosing the correct

parameter for calculating the equation.

3.4 Results and Discussion

3.4.1 Sorption behaviors

Non-linear least-square curve fitting by using Eq. 3-5 allows to use the
entire range of data to obtain the diffusivities such that it provides much more
realistic results. Results from the moisture sorption process, as illustrated in
Table 3-1, suggests that the magnitude of the diffusivity of the epoxy network is
inversely related to its functionality such that the tetra-functional epoxy resin
system, MY721/D230, revealed the lowest diffusivity. It is clear that the diffusion
process of the crosslinked epoxy of a higher functionality is hindered by its higher
crosslink density. In addition, the diffusivity of the epoxy glass increased as the
soaking temperature increased, showing the temperature dependence of the

diffusivity, as expected.

ain
ight g

e

°/o w

0.5I

O.(

”Slim 3
SIIUrared

27

DGEBA/D230
3.50 mm... .1 . mu... .
G 45.0 .. ..
——a cot:
— Ar— -7SC
-----ooc

 

 

   
   

 

3.00

I"
or
o

 

I“
o
o

._L
in
O

°/o weight gain

1.00

0.500

jllIlIrlllllIIITIIIIIIIIIIIIIIIIII

 

 

 

lllllllli IJIII

 

0.00 11111
1000 104 105 106 107

Time (s)

Figure 3-2 (a): Moisture absorption curve of DGEBA/D230. Samples were
saturated with water at 45°C, 60°C, 75°C, and 90°C for 2254 hours. Lines were
obtained from non-linear least-squared curve fitting, by using Eq. 3-5.

in
eight ga
% w

Figu
Satur
0btai

28

MY510/0230
8.00 I lllTlTll I llllllll l llllllll l llflllL

 

 

 

«.0 450
——B -600 I
—A—-7SC I 5
".“90C ll AA

A

ODD t> ‘
!l
I
s t> o
'5
D
II
D
91 J"

llllIllllIIIIIIIIIIIIIIIIIIJIIIII

 

3.00

2.00

1.00

 

°/o weight gain
4:.
O
O
TIIT'TITIllTTrlllTIIIljIIIIIIITIIIIITTI

 

 

 

000 11111 lllllllJI llllllllI _i llllLL
1000 104 105 106 107
Time(s)

Figure 3-2 (h): Moisture absorption curve of MY510/D230. Samples were
saturated with water at 45°C, 60°C, 75°C, and 90°C for 2254 hours. Lines were
obtained from non-linear least-squared curve fitting, by using Eq. 3-5

irt
eight 98
‘>é> var

figure
SalUrat

29

MY721/0230

 

 

      
   
   
     

 

 

6.00 l l llllll] i l lllllll l l lllllll lj IIITTL
,
- G 45 C
5.00 :- —B _ 60 C -q
"I — Ar- -75 C '_
400’— -----eoc -
.s I I
m u:
o, .
E ‘ 1
.9 3.00 - _
a) .—
3 - ..
,\° - -
2.00 :- T.
1.00 f -_‘
I. llllllli l lllllllI l lIllI‘

 

 

 

O 00 l t l ll
6 7
to3 104 105 10 10
Time (s)
Figure 3-2 (c): Moisture absorption curve of MY721/D230. Samples were

saturated with water at 45°C, 60°C, 75°C, and 90°C for 2254 hours. Lines were
obtained from non-linear least-squared curve fitting, by using Eq. 3-5

Table 3-1

r...
M:

TEE
(di-ft

MYS
(tri-ft

MY?

fun:

 

30

Table 3-1: Changes in M... and diffusivity due to the functionality difference and
moisture absorption.

 

 

 

 

Materials T, (°C) T (°C) M... M... 0x10"
(dry) (cond. temp.) (g) (%) (cmZ/sec)
DGEBA/D230 80 Dry (as cast) ----- 0.00 -----
(di-functional) 45 0.0997 2.88 0.39
60 0.1127 3.01 1.32
75 0.1111 3.06 2.63
90 0.1157 3.16 8.21
MY510/D230 98 Dry (as cast) ----- 0.00 -----
(tri-functional) 45 0.2323 8.50 0.206
60 0.2275 7.37 0.829
75 0.2228 7.05 2.165
90 0.2231 7.28 6.534
MY721/D230 136 Dry (as cast) ----- 0.00 -----
(tetra- 45 0.1398 5.76 0.19
functmml) 60 0.1420 5.27 0.62
75 0.1351 4.98 1.33
90 0.1211 4.65 3.84

 

 

 

 

 

 

 

 

D (x 10'8 cm’lsec) = Diffusivity.
Mm, = Maximum weight gain in gram.

M... = Maximum weight gain in percentage.

tempera
regressn
obtained
Shown i
agreeme:

behavior

31

The diffusivities obtained by using Eq. 3-5 at different sorption
temperatures were plotted as lnD vs. I/T in Figure 3-3. By extrapolating the
regression line to the y-axis, D. was obtained. The activation energy, Q, was also
obtained by taking the slope of the regression line. The values of D. and Q are
shown in Table 3-2. It can be seen from Figure 3-3 that there is a good
agreement between the variation of D with temperature and the Arrhenius
behavior,

D = D, exp(— %)

Eq. 3-7

where D. is the diffusivity at a temperature (cmz/sec), Q is the activation energy

of the material (J/mol), R is the gas constant (J/mol ' K), and T is the temperature

(K)-

Activation energy is an energy required for a diffusant to diffuse through
the material from one [site to another. If the system is a highly crosslinked
polymer network, the diffusion sites (either hydroxyl sites or free volume) will be
closer to each other so that the water molecules can easily hop around (requiring
less energy) and occupy the site. Table 3-2 illustrated the activation energy (Q)
and the diffusion constant (D0) of the each epoxy networks. It is shown that

MY721/D230 has the lowest activation energy and the highest diffusion constant.

~160

~17.0

InD

-18.0

-19.0

20.0

Figure 3.
(Cm: Sec)

32

DGEBA/D230

 

-15.0

-16.0

-17.0

-18.0

-l9.0

lnD
IIII'WIII'IITI'IIII'ITFI

 

 

LIIIIIIIIIIIIIIIIIIIIIJ

 

-20.0
0.0026 0.0027 0.0028 0.0029 0.0030 0.0031 0.0032

1/r (1/C)

Figure 3-3 (a): InD vs. 1/T curve of DGEBA/D230, where D is the diffusivity
(cmz/sec) and T is the temperature (K).

33

MY510/D230

 

-15.0

-l6.0

-17.0

-18.0

-19.0

lnD
IIII!IIII'IITI'IIII'IIII

 

 

I.-

 

_200 IIiILIJILILIIIIIIILIII
0.0026 0.0027 0.0028 0.0029 0.0030 0.0031 0.0032

1/1‘ (l/°C)

Figure 3-3 (b): InD vs. l/T curve of MY510/D230, where D is the diffusivity
(cmz/sec) and T is the temperature (K).

34

MY721/D230

 

-16.0
-16.5

-l7.0

-17.5

2 -18.0

-18.5

-l9.0

'IIII!IIITliiii'IIII'IIII'IIII'IIII

-19.5

I”!

 

 

-200JIIIIIIIIIIIIIIIIJJIIII
0.0026 0.0027 0.0028 0.0029 0.0030 0.0031 0.0032

lfl‘ (l/C)

 

Figure 3-3 (0): InD vs. l/T curve of MY721/D230, where D is the diffusivity
(cmz/sec) and T is the temperature (K).

 

35

Table 3-2: Activation energy and diffusion coefficients of the epoxy systems

 

 

 

 

 

Materials Q (kJ/mol) D. (cmz/sec)
DGEBA/D230 69.5 1.05 x 10"
MY510/D230 72.5 1.72 x 10'3
MY721/D230 62.8 0.39 x lo2

 

 

 

36

3.4.2 Thickness effects on the sorption behavior

In this experiment, in order to investigate the thickness effects on the
sorption behavior, samples with two different thicknesses were used; 0.32 cm (S1)
and 0.159 cm (S2). The result, as depicted in Figure 3-4, indicates that thickness
variation affects the sorption such that during the initial sorption process, thinner
samples (82) absorbed moisture at a rate faster than for the thicker samples (S1).
This is shown in Table 3-3 where the diffusivity of the thin specimens are higher
that that of thick specimens. In addition, the magnitude of the saturation point
(M...) for the thinner specimens is smaller than that for the thicker samples
suggesting that in a epoxy glass, M... as well as duffusivity are depended upon the

thickness of the specimen.

Consider the concentration profiles in the thickness direction of thin and
thick samples, since the length and the width of the two samples are asssumed to
be the same. As moisture absorption proceeds from the surface towards the
center of specimen, at a given time, water concentration in the thickness direction
of the thin samples will be higher than that of the thick samples since the distance
between the surfaces are shorter for thin samples such that the distance that water
molecules have to travel through the matrix to the center is shorter. However, for
thick samples, as opposed to thin samples, the distance between surfaces are
longer such that it requires longer time for water molecules to diffuse through the

matrix. In addition, assume that at a given time such that the thin samples are

37

completely‘saturated, only the outer surface and some distance towards the center
of the thick samples will be saturated. Consequently, some degree of swelling will
be observed on the surface of the thick samples, as a result, diffusion process will
be hindered by the swelled structure and while being saturated, due to the swelling
effect, equilibrium moisture content, M.., will be higher than that of thin samples.
This is exactly what happened as shown in Figure 3-4 where weight gain
percentages were plotted against time, normalized by the thickness. The results,
Table 3-3, showed that the diffusion coefficients of thick samples were lower than
those of thin samples and the equilibrium moisture content, M..., of thick samples,
on the other hand, were higher than that of thin samples. As discussed on the
above, those results can be said of due to swelling effect. Similar results were
also observed elsewhere [6, 45, 46] that the distribution of water between the less

dense and dense specimens were explained by the swelling phenomena.

38

 

 

 

 

DGEBA/0230
4.0 lllllllll I lillttll llllltlll 11111111] 1 111111
3.0 -
.E - _
(U 0
2° - . -
.C
.9 2.0 - -
Q)
3 ~ -
e\° - -
O
' oom ‘
1.0 r O 15.0 -
I' A o ..
_ O A _
o” It]
- A0 D ..
- 0 am -
00 IIIHDIII llllllllI llllllllI lllllllli llllllll

 

104 105 106 107 108 109
Reduced Time (s/cmz)
Figure 3-4 (a): Sorption curves for DGEBA/D230 with different thickness; SI and

S2. Hollow symbols represent thick samples (S1). Filled symbols represent thin
samples (S2). (13) 45°C, (0) 60°C, (A) 75°C, (0) 90°C.

39

 

 

 

 

 

 

MY510/0230
8 _ 1 111111.] 1 lllllll rrtltm] l iTllltll 1 111111_
I 1
7 [ q
I- -I
C i
I
i’ -
5 I l
.s I- -
CU | -
(D I -
E 4 t ‘
.9 I- I
g ._ a
e\° I j
3 —- —I
: O A o I! :
1 ~ <* a —
- A O b 1
I 0 D 7.
O b l IIJJIIIP J llJlJtII l lltllllI l llIllIII L ltlllli

to4 to5 106 107 108 10°

Reduced Time (s/cmz)

Figure 3-4 (h): Sorption curves for MY510/D230 with different thickness; 81 and
$2. Hollow symbols represent thick samples (5]). Filled symbols represent thin
samples (S2). (D) 45°C, (0) 60°C, (A) 75°C, (0) 90°C.

40

MY721/D230

 

6 1 1111111] 1 1111111] 1 1111111] 1 11111“

d

I
I

l

I W I I 1 U I T

.15
'fi

T I

°/o weight gain
00

 

 

 

 

0 111111111 ILllllIlI 111111111 1111111

105 1o6 107 108 10°

Reduced Time (s/cmz)

Figure 34 (c): Sorption curves for MY721/D230 with different thickness; SI and
S2. Hollow symbols represent thick samples (SI). Filled symbols represent thin
samples (52). (D) 45°C, (0) 60°C, (A) 75°C, (0) 90°C.

41

Table 3-3: Thickness effects on M... and D.

 

 

 

 

 

 

 

 

 

 

 

 

Epoxy system Cond. temp. Specimens M... D
(°C) 1%) (x 10“
cmZ/sec)
DGEBA/D230 45 S1 (thick) 2.89 0.39
82 (thin) 2.87 0.31
60 S] 3.01 1.32
82 3.01 1.38
75 S] 3.06 2.63
82 3.09 3.67
90 S1 3.16 8.21
S2 3.16 8.23
MY510/D230 45 51 8.5 0.206
82 7.58 0.209
60 81 7.37 0.829
52 7.06 0.852
75 51 7.05 2.120
52 7.02 2.123
90 S] 7.28 6.53
82 7.24 5.57
MY721/D230 45 S1 5.76 0.19
S2 5.45 0.21
60 S1 5.27 0.62
52 5.07 0.66
75 81 4.98 1.33
82 4.83 1.34
90 S1 4.65 3.54
52 4.44 3.84

 

 

 

 

42

3.4.3 Hydroxyl Sites Dependency

Non-linear least-squared curve fitting, by using Eq. 3-5 deduced M... for the
epoxy glasses under study. The sorption curves showed that a tri-functional
epoxy network manifested a saturation level of approximately 7.5 %, followed by
a tetra-fianctional (z 5.2 %), and a di-functional epoxy resin absorbing the least (z
3.2 %), To provide an analytical results, hydroxyl group affinity of the epoxy
network was examined by calculating the numbers of hydroxyl groups present in

each epoxy network. The results, as presented in Table 3-4 show that the tri-

functional epoxy resin (MY510) having 6.309 x10’3 mole of [OH]/g contains the

highest hydroxyl groups per mole and the di-fimctional epoxy resin (DGEBA)

having 4.396 x10’3 mole of [OH]/g contains the lowest. The chemical nature of
the penetrant versus that of the polymer determines the penetrant-polymer
affinity. The relatively high water absorption capacity of epoxy resins derives
from the presence in the epoxy chains of [OH] groups attracting the polar water

molecules.

Therefore, it is suffice to say that the epoxy network having the highest
number of hydroxyl sites available in its network will attract and accommodate
water more readily than the epoxy network having a lower number of hydroxyl

sites available in its network.

43

Table 3—4: Number of hydroxyl groups in each epoxy

 

 

 

 

 

Epoxy System g / mole of [OH] mole of [OH] / g
MY510/D230 158.5 6.309x10'3
MY721/D230 167.5 5.970 >110—3
DGEBA/D230 227.5 4.396 x10'3

 

 

 

 

44

3.5 Summary

Diffusion process for an epoxy glass is directly related to the functionality
of the epoxy resin such that an epoxy glass formed by a resin having a higher
functionality will produce a highly crosslinked epoxy network, resulting in
hindrance in diffusion process. The results showed that the diffusivity of
MY721/D230 epoxy glass was the lowest while DGEBA/D230 having the highest
diffusivity. As expected, diffusion process in polymer network was also sensitive
to the temperature of the environment such that the diffusion rate increased with

the soaking temperature.

Thickness variation also affects the diffusion process. Swelling appeares to
be the main cause for the thickness variation effects where thin specimens showed
higher diffusion coefficients than thick specimens and thick specimens retained
higher equilibrium moisture content. As for the thick samples, due to swelling
effect at the surface, diffusion process was disturbed and overall equilibrium
moisture content was raised by the addition of moisture in swelled structure at the

surface of the specimen.

Finally, the epoxy resin containing the highest number of hydroxyl groups
per mole creates more hydroxyl sites for water molecules to react, resulting in a
higher moisture weight gain. As a result, MY510/D230 epoxy system absorbed

more moisture than any other epoxy systems studied.

4. MOISTURE EFFECTS ON PHYSICAL PROPERTIES

4.1 Dilatometry

Dilatometry is used to study and to measure dimensional changes of matter.
Dimensional changes in polymers occur as a result of a variety of stimuli, e.g.
absorption of fluids; mechanical stresses; chemical reactions (polymerization and
cross-linking); and effects of time (aging of glasses and crystallization). For
processing and engineering applications of polymeric materials, the study of
dimensional changes bears great technical significance. In addition, dilatometry is
also used to study the physical and chemical phenomena in conjunction with
dimensional changes, such as melting, crystallization, secondary transitions, glass
formation, polymerization, cross-linking reactions, fluid sorption, formation of
voids and crazes during mechanical deformation, and many others. Linear
dimensional changes are most useful in situations where it may be assumed that

the shape of the test specimen remains unchanged.

In this study, dilatometry was used on dry and moisture saturated samples
to monitor the change of the glass transition temperatures such that one can
examine the deviation of the glass transition temperature with respect to different
conditioning temperatures. Moisture desorbed samples were also examined by a
dilatometer; (i) to investigate if there were any changes in the glass transition
temperatures among the dry, saturated, and moisture desorbed samples; (ii) to

examine any change in the network conformation affected by the moisture.

45

46

By definition, the temperature at which the polymer undergoes the
transformation from a rubber to a glass is known as the glass transition

temperature, T g.

 

 

 

The linear thermal expansion coefficient

1 AI
aL =——
1, AT

Eq. 4-1

for either 0:8 or a; is obtained by taking the slope of the glassy or rubbery region,

respectively.

4.2 Experimental Procedure

The glass transition temperatures of the moisture saturated and desorbed

samples were measured by a length dilatometer manufactured by Omega shown in

47

Figure 4-1. The heating rate during the experiment was 2°C/min. Samples
saturated with distilled water were cut from their initial dimension of 7.62 cm x
0.32 cm x 1.27 cm into 2.54 cm long specimens, and the change of the sample
length with respect to the temperature increase was measured by a LVDT (Linear
Variable Differential Transducer). From the output voltage collected from the

dilatometer, the strain, a, was calculated by Eq. 4-2,

 

a = 0.023 x ("y/J

0

Eq. 4-2
where a = Strain
mV = Output voltage from the dilatometer.
1,. = Initial length of the sample (cm).
0.023 = A calibration constant determined by the deformation of

LVDT with a known length change (cm/m V).

Strain versus temperature curves are shown in Figure 4-2 through Figure 4-7.

The T g was determined as an onset point of the strain versus temperature curve.

48

 

Figure 4-1: A length dilatometer built by Omega Co.

49

 

   
  
  
  
 

 

 

 

DGEBA/D230
0.014 _
0.012 :- 3 my
0010i A Wetat45C
- l-
E o Wetat60C
0.008 _—
t 0 Wetat75C
.= :
§ 0.006 _— x Wet at 900
U) I
0.004:-
o.ooz :-
0.000,E
_0002:.l.l.1.l.1.l.
20 40 60 80 100 120 140 160

Temperature (C)

Figure 4-2: Strain versus Temperature (°C) of the DGEBA/D230 epoxy resin.
Samples have been saturated with distilled water at 45°C, 60°C, 75°C, and 90°C
for 2254 hours.

0.012

0.01'

0.00

0.00

Strain

0.00

0.00

0.00

~0,0C

FIgUI'e
Simplel
for 225,

50

 

    
  
   

 

 

 

MY510/D230
0.012 _
I :1
" 0°
0.010 .- 1: Dry fee?
E 0 Wet at 60C 6‘9 ' ‘
0'008 :— 0 Wet at 75C 8 g - x
. . D x
' W 1 1 90C
a 0.006 _- X e a ‘9 x
.8 _ x
b I x
m 0.004 _— x
.. Xx
; x
._ X
0.002 _ x
: x
0.000 r
P
t-
-o'002 I- l I l I l l I
0 50 100 150 200

Temperature (C)

Figure 4-3: Strain versus Temperature (°C) of the MY510/D230 epoxy resin.
Samples have been saturated with distilled water at 45°C, 60°C, 75°C, and 90°C
for 2254 hours.

51

 

 
  
  
 
    

Strain

 

 

 

MY721/D230
0.014 _
0.012 :-
E El Dry
0.010;- 0 Wet at 60C
2 o Wetat75C
0.008 :-
: X Wet at 90C
0.006 :-
0.004 E-
0002 3-
0.000 C-
-0.002 p l l l l l l l l l
0 40 30 120 I60 200

Temperature (C)

Figure 4-4: Strain versus Temperature (°C) of the MY721/D230 epoxy resin.
Samples have been saturated with distilled water at 45°C, 60°C, 75°C, and 90°C
for 2254 hours.

52

For the desorption of the samples, 2.54 cm x 0.32 cm x 1.27 cm wet
samples were dried in an oven at 100°C for 24 hours under vacuum. Then the

samples have undergone T 8 evaluation by the dilatometer, heating rate at 2°C/min.

4.3 Results and Discussion

4.3.1 Glass Transition Temperature (T,)

4. 3. 1. 1 Moisture Absorbed Samples

From Table 4-1, for DGEBA/D230 and MY510/D230 samples, the glass
transition temperatures decreased as the soaking temperatures increased. Due to
the moisture presence in the network, substantial molecular rearrangement of the
epoxy network occurred during the sorption process. Over the many years of
research on the moisture absorption, it is believed that the decrease in the glass

transition temperatures is due to the water acting as a plasticizer.

In a highly crosslinked sample, MY721/D230, the glass transition
temperature of the sample saturated at 60°C decreased substantially, compared to
that of the dry sample. As the soaking temperature increased, however, the glass
transition temperature increased gradually from 83°C to 99°C. This phenomenon
is in disagreement with the results obtained from other epoxy systems of which
the glass transition temperatures decreased consistently as the soaking

temperatures increased. Because the water molecules is polar, it is capable of

53

Table 4-1: Change of the glass transition temperatures due to moisture saturation.

 

 

 

 

 

 

 

Conditioned T8 of T, of T8 of

Temperatures DGEBA/D23 0 MY510/D230 MY721/D23 0
Dry 80°C 98°C 136°C
45°C 62°C ------------
60°C 57°C 68°C 83°C
75°C ‘ 53°C 58°C 91°C
90°C 49°C 54°C 99°C

 

 

 

 

 

54

forming hydrogen bonds with hydroxyl groups of the epoxy network, resulting in
a stiffer network. Higher the soaking temperature, the concentration of the water
would be higher due to the increased rate of the mass transport through the
matrix; therefore, more water molecules become available for the epoxy network
to form hydrogen bonding. Having the highly crosslinked epoxy network, the
crosslink density of the MY721/D230 was intensified by hydrogen bonding formed
between the absorbed water and the epoxy network. Not only translational, but
also rotational movement of the chain was hindered, resulting in the increase of

glass transition temperature.

It is noteworthy that, as shown in Table 4-2 under the constant

 

conditioning temperature, as AT 2 T,—Tg| increases, the diffusion coefficient

decreased (T . is an experiment temperature). In other words, as the conditioning
temperature is further away from the T g, the epoxy resin tends to be more stiff
such that it will be more difficult for the diffusants to penetrate through the epoxy

network.

55

 

Table 4-2: Correlationship between AT = T; — III and the diffusivities of the epoxy

 

 

 

 

glasses.
Materials T , (°C) T . (°C) Dx10'8
(cond. temp.) (cmZ/sec)

DGEBA/D230 80 Dry (as cast) -----

(di-functional) 62 45 0.39

57 60 1.32

53 75 2.63

49 90 8.21

MY510/D230 98 Dry (as cast) -----
(tri-functional) --- 45 0.206
68 60 0.829

58 75 2.165

54 90 6.534

MY721/D230 136 Dry (as cast) -----

(tetra-functional) --- 45 0. 19

83 60 0.62

91 75 1.33

99 90 3.84

 

 

 

 

 

 

D (x 10.11 cmzlsec) = Diffusivity.

56

4. 3. 1.2 Moisture Desorbed Samples

Dilatometry results on the samples that have been conditioned through the
desorption process show that the behaviors of recovering the glass transition
temperatures were different among DGEBA/D230, MY510/D230, and
MY721/D230 epoxy networks. For instance, the recovery of the glass transition
temperatures was not observed in any of MY721/0230 epoxy resin regardless of
their soaking temperatures, even though MY510/D230 and some of DGEBA/D230

samples did recover their T g.

The glass transition temperatures of the DGEBA/D230 samples saturated
at 45°C, 60°C, and 75°C returned to, or returned close to the inherent glass
transition temperature which is 80°C. As shown in Figure 4-5 through Figure 4-7,
one exception was found in the sample saturated at 90°C for which the glass
transition temperature of the desorbed sample reached only to 70°C. The
saturation temperatures of 45°C, 60°C, and 75°C were lower than the glass
transition temperature of the DGEBA/D230 epoxy. The saturation temperature of
90°C, however, is higher than the glass transition temperature of the
DGEBA/D230 epoxy network, so that during the sorption process, the material
was in a rubbery state. As the sorption process proceeded, the molecular
rearrangement of the epoxy network was readily achieved. During the desorption
process, the polymer network would not have enough time for long range

molecular motion to restore its original molecular arrangement. After the

57

desorption, the epoxy would exhibit a network structure in which the newly
formed molecular conformation is “frozen in”, providing the network more room
to relax. This would result in the suppression of the glass transition temperature,

as partially discussed elsewhere [45].

From the sorption curve in Figure 3-2, tri-functional epoxy (MY510/D230)
absorbed the moisture the most. Higher percentage of the moisture uptake can be
explained such that in addition to its higher hydrophilic sites, the tri-functional
epoxy system has a greater mobility in its segmental chain movement. The
conformational change of the network after the moisture introduced to the
network through the sorption process can be readily compensated by the
segmental chain rearrangements. After the desorption process, the chain segments
were able to return to its original conformation such that the glass transition
temperatures of the samples, regardless of their soaking temperatures returned to

the initial 7, (98°C).

MY721/D230 epoxy is a highly crosslinked epoxy resin. When the system
was in the sorption process, it experienced the molecular rearrangement as other
epoxy systems (DGEBA/D230 and MY510/DZ30) encountered. The mobility of
the chain segment, however, was likely to be hindered due to its higher crosslink
density. Low chain mobility due to the higher crosslink density required more
time, or made even impossible for the molecular chains to recover its original

chemical conformation. Therefore, even though the desorbed samples showed

58

their glass transition temperatures in agreement at approximately 112°C, the

initial glass transition temperature (136°C) was not recovered.

59

 

  
 
 
 
 
 
 
  

Strain

 

 

 

DGEBA/D230
0.014 .
:
0.012 .— U D”
: A Wet45C/dried
0.010 .-
: O Wet60C/dried
000“ :— <> Wet75C/dried
0.006 :- X “0‘90““
0.004 :—
0.002 :-
0.000 :-
-0-002 _ l I I I I I l I I I l l l
20 40 60 so 100 120 140 160

Temperature (C)

Figure 4-5: Change of the strain between the dry and moisture desorbed samples
of the DGEBA/D230 epoxy resin were plotted against temperature (°C).

60

 

   
  
  
  

 

 

 

MY 510/D230
0.012 t
I.
0.010 C- D D” o
I o Wet60C/dried
0-003 .— <> Wet75C/dried
E X Wet90C/dried
0.006 :-
.s _
CU
*3 I
m 0.004 :-
P
0.002 :-
0.000 _—
I
41002 1 I l I 1 I 1
0 50 100 150 200

Temperature (C)

Figure 4-6: Change of the strain between the dry and moisture desorbed samples
of the MY510/D230 epoxy resin were plotted against temperature (°C).

61

 

 

 

 

MY721/D230
0.016 _
'_' El Dry 0.9
0.012 :- O Wet60C/dried col
: o Wet75C/dried it?
1- . L‘s ..,' o
a 0.008 _- >< W°t9°C’d"°d 5:" 008
b I. "1:55;“.“719'1110 02:? .
m . .
0.004 '- ....................
1- ‘c~
- 0 « ‘U
0.000 - . ’-
r
-0 004 h . ' . I . I . l .
0 50 100 150 200 250

Temperature (C)

Figure 4-7: Change of the strain between the dry and moisture desorbed samples
of the MY721/D230 epoxy resin were plotted against temperature (°C).

62

4.3.2 Coefficients of Thermal Expansion (CTE)

4. 3. 2. 1 Moisture Saturated Samples

The thermal expansion coefficients of glass region, a8, and of liquid region,
at, were obtained by taking the slopes of the glass and the rubbery regions of
strain versus temperature curves, respectively. The coefficient of thermal
expansion is one way of manifesting the mobility of the molecular chain segment.
Its implication with the molecular chain segment mobility draws an attention that

how the molecular mobility would be affected by the sorption process.

In the glassy state, the coefficients of thermal expansion, a3, for the
various wet samples are not very different; however, the comparison of the
coefficients of thermal expansion among each epoxy network shows that the
magnitude of the CTE follows the same order as the moisture absorption
behaviors such that MY510/D230 having the highest ag, followed by
MY721/D230, and DGEBA/D230 having the lowest (13, refer to Table 4-3.
Higher CTE means higher molecular mobility. As shown from the previous
chapter of this study, MY510/D230 absorbed about 7.5 wt% of moisture, the
highest among three, because it has more hydroxyl groups available in the
network. It was also said that the absorbed water is acting as a plasticizer in the
polymer network. The epoxy resin that has higher number of hydroxyl groups per

mole in the network absorbed more moisture, which plasticizes the material;

63

therefore, MY510/D230 possesses the highest ag followed by MY721/D230 and

DGEBA/D230.

In the rubbery state, it appears that in comparing a; between the dry and
wet samples, wet samples of all three epoxy systems showed a decrease in their
on. When the water molecules were introduced to the crosslink network, the
“voids” between the polymer molecules were filled with water molecules such that
the mobility of the polymer chains were hindered. This restriction of the chain
mobility resulted in the decrease of 01,. Figure 4-2 through Figure 4-4 well
illustrated these effects where or; portions of the curves were leaning towards the
x-axes. Further reasoning of this phenomena was attempted microscopically by
using positron annihilation lifetime spectroscopy (PALS). Detailed discussions on

this subject were presented in section 4.4.

64

Table 4-3: Thermal Expansion Coefficients of DGEBA/D230, MY510/D230, and

 

 

 

 

 

 

 

 

MY721/D230.
a, (10“°c") a. (10"°c")
Temperatur DGEBA! MY510/ MY721/ DGEBA/ MYSIO/ MY721/
e 0230 D230 D230 , D230 D230 D230
(° C)
Dry (23° 0.6824 0.736 0.704 1.8854 1.7430 1.5218
C)
45° c 0.6338 ---------------- 1.4473 ................
60° c 0.6157 0.8585 0.7364 1.4562 1.9360 1.3937
75° c 0.6399 0.7511 0.7398 1.3463 1.8875 1.7778
900 c 0.5871 0.8820 0.8738 1.3407 1.5056 0.7102

 

 

 

 

 

 

 

 

65

4. 3. 2.2 Moisture Desorbed Samples

The coefficients of thermal expansion of moisture desorbed samples were
presented in Table 4-4. Even though there were some variations in T g, the results
showed that the difference between the coefficients of thermal expansion of dry

samples and that of moisture saturated samples was negligible.

Water molecules occupying the “voids” between the polymer molecules
were eventually dried out afier the desorption cycle such that the polymer
network recovered (maybe not 100% but somewhat close) its conformation.
More detailed explanations were presented in section 4.4 with supporting

evidence comprised of PALS results.

66

Table 4-4: Thermal Expansion Coefficient of DGEBA/D230, MYSlO/D230, and
MY721/D230 after the moisture has been desorbed at 100°C for 3

 

 

 

 

 

 

 

 

days.
a, (10" °c") a. (10" °C")
Temperatur DGEBA/ MY510/ MY721/ DGEBA/ MY510/ MY721/
e (° C) D230 D230 D230 13230 D230 D230
Dry (23° 0.6824 0.736 0.704 1.8854 1.7430 1.5218
C)
45° C/dried 0.662 ---------------- 1.965 ................
60° C/dried 0.640 0.717 0.667 1.697 1.630 1.575
75° C/dried 0.674 0.762 0.674 1.967 1.594 1.523
90° C/dried 0.624 0.731 0.686 1.954 1.620 1.532

 

 

 

 

 

 

 

 

67

4.4 Positron Annihilation Lifetime Spectroscopy (PALS)

4.4.1 Background

The free-volume theory has long been assumed to explain the molecular
motion and physical behavior of glassy and rubbery states [6, 16, 22~25, 27]. The
theory has been widely adopted among the polymer scientists because it is
conceptually simple and intuitively plausible for understanding many polymer

properties at a molecular level.

The free volume, by definition, is the space in a solid or liquid sample
which is not occupied by polymer molecules, i.e. the ‘empty space’ between
molecules [50]. In the liquid state, it is supposed that the free volume will be
sensitive to the change in temperature and that most of the thermal expansion of
the polymer rubber or melt can be explained by a change in the free volume. As
the temperature of the melt is decreased, the free volume will be reduced until
eventually there will not be enough free volume to allow molecular rotation or
translation to take place. The temperature at which this happens corresponds to
Tg as below this temperature, the polymer glass is effectively ‘frozen’. The
situation is illustrated schematically in Figure 4-8. The free volume is represented
as a shaded area. It is assumed to be constant at Vf. below T g and to increase as
the temperature is raised above T g. The total sample volume V,, therefore,

consists of volume occupied by molecules V0 and free volume Vf such that

 

 

 

 

69

V, =V0+Vf

Eq. 4-3.

One of the challenging issues in polymer research is in quantifying the
relationship between the molecular motion and free-volume properties. In the
early stage of development, the free volume was merely considered to be a
theoretical concept, something that could not be measured directly. In the last
two decades, however, the PALS method has become a more and more powerful

tool for studying the molecular structure of polymeric materials in general [52].

To take advantage of this notable technique, PALS method was utilized in

this study to examine the nature of the moisture in the polymer network.

4.4.2 PALS

The method makes use of the interactions between the positrons (emitted,
usually, from a radioactive material such as 22Na) and the electrons from the
material under study. One employs the anti-electron, the positron, as a probe and
monitors the lifetime of the positron and positronium, Ps (a bound atom which
consists of an electron and the positron) in the host material. P8 is formed when a
positron is injected into a solid, it captures an electron from the surrounding
medium forming the neutral atom, positronium (Ps). At the present time,
however, the positronium formation process in polymeric materials is not well

understood.

70

Due to the positive-charged nature of the positron, the positron and Ps are
repelled by the core electrons of polymers and trapped in open spaces, for
instance holes, the free volume, and voids. The positron and P8 are localized in
these preexisting local holes and the free volume in polymers [32]. The

annihilation photons come mainly from these open spaces.

Typically, the lifetime of positron is in the range of 0.3 ~ 0.5 ns, while the
lifetime of the P5 depends upon the spin state [51]. Depending on the spin
configuration, two ground states of Ps exist: the singlet state called para-P5 (p-
Ps) and the triplet state called ortho-Ps (o-Ps). In vacuum, o-Ps and p-Ps have
lifetimes of 140 ns and 125 ps, respectively [33]. It has been found that the
lifetimes and relative intensities of the positron itself and p-Ps are not directly
related to the materials characteristics. On the other hand, the lifetime (13) and
relative intensity ([3) of the o-Ps are influenced by the structure of the material
under study [52]. The lifetimes of the p-Ps (n) and e" (12) are much shorter
(0.125 ns and 0.5~0.6 ns, respectively) than the vacuum lifetime of o-Ps (~l40

ns).

In molecular substances, the repulsion between the electron in o-Ps and the
electrons in the surrounding medium localizes the o-Ps in regions with lower
electron densities or ‘voids’. o-Ps annihilates through a process called ‘pick off’
annihilation where the positron of o-Ps annihilates with one of the bound electrons

with opposite spin. As a result of the ‘pick-ofl‘ annihilation, the lifetimes

71

decrease considerably, and for most amorphous polymers are in the range 1.5~3
ns. The lifetime of o-Ps (13) is correlated to the size of the void where o-Ps is
situated at the instant of annihilation, while the intensity of the o-Ps lifetime ([3) is

related to the concentration of voids in the polymer.

4.4.3 Experimental Procedures

Three different conditions of samples - dry, conditioned at 75°C, and
moisture desorbed at 100°C in vacuum for 24 hours - were prepared from each
epoxy systems. The PALS measurements were carried out by using a
conventional fast-timing coincidence method in a temperature-controlled chamber
at a constant temperature of 25 i 1°C. The positron source [~1 x 10" Bq (~30
pCi) of 22Na] was deposited on the surface of an annealed single-crystal copper
disc. As a result, [3 measured was approximately half of that determined with the
source sandwiched between two samples. The spectra were acquired and stored
in a MicroVAX (Digital model 3100) based multichannel analyzer. Each spectrum
was acquired for about 3 hrs. The lifetime spectra were resolved into three or

four components by using the PFPOSFIT program.

4.4.4 Results and Discussions

The results shown in Table 4-5 indicated that between dry and wet samples
of all three epoxy systems, there were no significant changes in the lifetime of 0-

P8 (:3) (r; of wet samples decreased in the range of 0.53~1.91%). The intensity

72

of the o-Ps lifetime ([3), however, showed noticeable changes between dry and
wet samples (10.94~17.3% decrease). From these findings, one can suspect that
when the water molecules were introduced into the epoxy network, they occupied
the ‘voids’ between the polymer molecules such that even though there were no
significant changes in 13 (related to the size of the ‘voids’), 13 (related to the
concentration of ‘voids’) decreased substantially for water molecules occupying
the voids. This reasoning was also supported, in previous sections in this
Chapter, by the dilatometry results shown in Figure 4-2 through Figure 4-4 where

the a; curves of moisture saturated samples were leaning towards the x-axis.

After the samples went through the desorption cycle, again, PALS
measurements were taken on all three epoxy systems. Overall, no distinct changes
in 13 and 13 were observed between dry and moisture desorbed samples. It is
noteworthy, however, that unlike DGEBA and MY510 epoxy systems, in
MY721/DZ30, there were slight increase in 13 and I3. MY721/D230 epoxy
network, being a highly crosslinked system, sustained elevated internal stress by
the water molecules occupying the free volume. After prolonged immersion in the
moisture bath, the crosslinked molecular chains may have been ruptured,

increasing the size and the concentration of the voids in the polymer network.

4. 4. 4. 1 C audition ed-temperatu re effects

In order to investigate how the free volume physically respond to the

moisture, absorbed at different temperatures, another set of an experiment was

73

conducted. Assuming that the physical response would be the same for all three
epoxy networks, the experiment was performed on only one epoxy system, i.e.

DGEBA/D230.

Samples conditioned at 45°C (845), 75°C (875-1, 875-2), and 90°C (890)
were used. After the PALS measurements on wet samples, S45, 875-1, and 875-2
were then dried following 5 different drying cycles; $45, dried at 75°C for 8 days;
875-1, dried at 75°C for 8 days; 875-2, dried at 75°C for 23 hrs, 90°C for 24 hrs,

and 100°C for 52 hrs. Results were tabulated in Table 4-6.

Results showed that 13 and 13 of dried samples were not much different
from those of dry samples. From these findings, we could argue that regardless of
temperatures at which the samples were conditioned, the free volume of the
DGEBA/D230 were not affected by different sample-condition temperatures.
Therefore, in terms of the size and the concentration of the free volume in a
polymer network, moisture temperatures of the service environment does not

affect its sorption behavior.

74

Table 4-5: 13 and 13 results on epoxy systems.

 

 

 

 

 

 

 

 

 

 

 

Epoxy systems Sample conditions 13 (ns) 13 (%)
Dry 1.776 15.415

DGEBA/D230 Wet (75°C) 1.742 12.941
Dried 1.785 14.664

Dry 1.683 15.469

MY510/D230 Wet (75°C) 1.701 12.793
Dried 1.630 14.100

Dry 1.702 14.613

MY721/D230 Wet (75°C) 1.693 13.015
Dried 1.716 14.715

 

 

 

 

 

75

Table 4-6: Conditioned-temperature effects on the free volume. Presented here
are data obtained on DGEBA/D230 epoxy network.

 

 

 

 

Sample condition 13 (ns) 13 (%)
dry 1.774 15.479

S45 1.742 12.996

S75-1 1.745 13.048

875-2 1.742 12.948

S90 1.731 13.014

S45 (75°C / 8 days) 1.763 15.154
875-1 (75°C / 8 days) 1.781 14.421
875-2 (75°C / 23hrs) 1.776 14.481
S75-2 (90°C / 24hrs) 1.793 14.299
S7s-2 (100°C / 52hrs) 1.751 14.543

 

 

 

 

76

4.5 Summary

Depression of T g was observed in moisture saturated samples. There was
an evidence that since the water molecule is polar, it formed hydrogen bonding
with hydroxyl groups of the epoxy network. In a highly crosslinked epoxy
network (MY721/D230), hydrogen bonding formation resulted in increase of T g.
Tgs obtained from the moisture saturated samples were far lower than dry T ;
however, as the condition temperature increased, T 8 also increased. Nevertheless,

its dry T g was never recovered even after the moisture was desorbed.

It is noteworthy that under the constant conditioning temperature, as

AT 2 T—7;| increases, the diffusion coefficient decreased (T, is an experiment

 

temperature). In other words, as the conditioning temperature is further away
from the T , the epoxy resin tends to be more glassy such that it will be more

difficult for the diffusants to penetrate through the epoxy network.

When crosslinked epoxy systems absorb moisture, the water molecules tend
to reside in ‘empty space’ between molecules, so called free volume. Dilatometry
results on the moisture saturated samples showed that the slope of rubbery region
was depressed such that the curve was leaning towards the x-axis. The total
sample volume V consists of the sum of volume occupied by molecules V0 and free
volume Vf. Therefore, when the water molecules occupy the free volume, Vf
decreases resulting in decrease of V, where V0 is constant. After the moisture had

been desorbed, the slope of rubbery region retained its dry slope.

 

77

PALS result also proved the water molecules occupying the free volume in
the epoxy network. It was found that when the samples were saturated, there
were no significant changes in r; (the lifetime of o-Ps, related to the size of the
free volume), but noticeable drop in 13 (the intensity of o-Ps, related to the
concentration of the free volume). When the moisture was desorbed, the
concentration of the free volume was closely returned to its initial concentration.
It is necessary to note that the sample condition temperature affected neither on

the size nor on the concentration of the free volume.

5. MOISTURE EFFECTS ON MECHANICAL PROPERTIES

5.1 Experimental Procedures

When epoxies are utilized as composite matrices and adhesives, they are
often exposed to extreme environments where high content of moisture are
commonly observed. Predicting the durability of epoxies under such environments
has been an important issue especially when epoxies are used in aerospace
structural components. In order to examine the physical effects on the epoxies

due to the absorbed moisture, mechanical testing was necessary.

Tensile testing was performed by using an Instron type 4200 series screw-
driven tensile machine. Sampling rate was 10 pts/sec and the crosshead speed was

1.0 mm/min.

5.2 Samples

Dry and moisture saturated samples of DGEBA, MY510, and MY721 were
tested. The width of the samples were 4.02 mm, the thicknesses were ranged

from 1.20 to 1.46 mm, and the gauge lengths were 25.4 mm (1 inch).

78

 

79

5.3 Results and Discussions

5.3.1 Oxidation and Degradation

Samples have been in the moisture chambers at elevated temperatures over
2000 hours and it is suffice to say that there would be some oxidation on the
surface of the samples due to the oxygen contained in the moisture and in the air.
This assumption was proved to be true when the difference of the samples’ color

were compared between samples soaked at different temperatures.

As the soaking temperatures were increased, sample discoloration became
severe such that the samples’ color became darker as the temperatures of the
soaking environment temperatures increased. Discoloration of the samples can be
said to be an indication that there was a degree of oxidation. Change of sample’s

color is shown in Figure 5-1 through Figure 5-3.

Absorbed moisture caused not only the oxidation but altered the
mechanical properties of the epoxies. Table 5-1 through Table 5-3 show the
results of the mechanical tensile testing. Tensile testing results of DGEBA/D230
in Table 5-1 show that overall mechanical properties decreased except the %
strain-at-break. This is due to the moisture acting as a plasticizer such that the
crosslink network became more susceptible to the deformation. However, once

the soaking temperature is past 60°C, the % strain-at-break started to decrease.

80

DRY 45°C 60°C 75°C 90°C

 

Figure 5-1: Comparison of the color change between dry and moisture saturated

samples. As the soaking temperature increases. the color of the sample becomes
darker. Shown here are DGEBA/D230 epoxy resins.

81

 

 

Figure 5-2: Comparison of the color change between dry and moisture saturated
samples on MY510/D230 epoxy resins.

82

 

 

'Figure 5-3: Comparison of the color change between dry and moisture saturated
samples on MY721/D230 epoxy resins.

83

This is due to the chemical degradation of the crosslink network induced by the
moisture such that while the moisture is behaving as plasticizer, increasing the %
strain-at-break, it also serves as a cause for the chemical degradation. Young’s
modulus dropped from dry state to 45°C by 11%, to 60°C by 17%, to 75°C by
23%, and to 90°C by 25%. Load-at-Break dropped from dry to 90°C by 55%,
stress-at-break dropped by 48%, load-at-0.2% yield by 45%, and stress-at-0.2%
yield by 37%, while % strain-at-break rather increased from dry to 90°C by 42%.
Therefore, the moisture in the DGEBA/D230 epoxy system, drop of Young’s

modulus is also another indication of the chemical degradation.

The behavior of the reduction of mechanical properties in MY510/D230
and MY721/D230 was similar to each other in a sense that sudden drops of
mechanical properties at higher end of the soaking temperature (75" C and 90° C)
were clearly visible in both systems. The DGEBA/D230 showed some degree of
reduction in their mechanical properties; however, the magnitude of the drop from
one temperature to next was not as considerable as those appeared in

MY510/D230 and MY721/D230 systems.

As can be been in Table 5-2, Young’s modulus of MY510/D230 epoxy
system from dry to 90°C dropped by 41% which is 16% more drop than that in
DGEBA/D230 system. Load-at-break dropped from dry to 90°C by 83%, stress-

at-break by 79%, load-at-0.2% yield by 90%, and stress-at-0.2% yield by 87%.

84

Table 5-1: Tensile testing results on DGEBA/D230 epoxy systems.

 

 

 

 

 

 

 

Temperature % Strain at Load at Stress at Load at Stress at Young’s
(°C) Failure Failure Failure 0.2% Yield 0.2% Yield Modulus
(%) (kN) (MPa) (kN) (MPa) (MPa)
as cast (dry) 8.230 0.3295 61.58 0.2763 51.63 1678
45°C 9.469 0.3236 61.73 0.2188 42.40 1497
60°C 14.21 0.2341 43.01 0.2044 37.35 1395
75°C 13.27 0.1908 38.47 0.1617 32.60 1287
90°C 11.65 0.1497 32.16 0.1507 32.35 1257

 

 

 

 

 

 

 

 

85

Table 5-2: Tensile testing results on MY510/D230 epoxy systems.

 

 

 

 

 

 

 

Temperature % Strain at Load at Stress at Load at Stress at Young’s
(°C) Failure Failure Failure 0.2% Yield 0.2% Yield Modulus
(%) (kN) (MPa) (kN) (MPa) (MPa)
as cast (dry) 11.280 0.3720 77.15 0.2493 51.72 1845
45°C 12.825 0.2829 54.37 0.1912 36.74 1428
60°C 17.030 0.2560 48.57 0.1650 33.03 1261
75°C 29.990 0.1961 39.02 0.1232 24.52 1053
90°C 1.498 0.0617 16.05 0.0250 6.669 1085

 

 

 

 

 

 

 

 

86

Table 5-3: Tensile testing results on MY721/D230 epoxy systems.

 

 

 

 

 

 

 

Temperature % Strain at Load at Stress at Load at Stress at Young’s
(°C) Failure Failure Failure 0.2% Yield 0.2% Yield Modulus
(%) (kN) (MPa) (kN) (MPa) (MPa)
as cast (dry) 6.590 0.3884 79.78 0.2682 55.05 1911
45°C 8.952 0.4105 75.23 0.2527 46.31 1652
60°C 8.005 0.4096 74.66 0.2524 46.03 1505
75°C 3.155 0.2443 44.54 0.2338 42.62 1582
90°C 0.778 0.0530 10.48 0.0239 4.96 1376

 

 

 

 

 

 

 

 

87

strain-at-break at dry state was 11.28% and increased by 166% to 29.99% at 75°C
and then dropped drastically to 1.498% by 87%. All these results show

substantial drop in mechanical properties as soaking temperature increases.

MY721/D230, a tetra-functional epoxy, showed similar plunge in
mechanical properties as seen in MY510/D230. Results showed that when
comparing the tensile testing data of dry with 45°C, and 45°C with 60°C, there
was a steady decrease in mechanical properties. When comparing data between
60°C and 75°C, the decrease was not steady anymore, and at 90°C, the decrease
in mechanical properties ranged from 86% to 91% and the Young’s modulus

dropped by 28% from 191 1 MPa to 1376 MPa.

Such decreases in mechanical properties in all three epoxy systems clearly
exhibit that the moisture absorbed by the samples contributed severe damage to
the crosslink networks of the epoxy systems such that chemical degradation of the
network has become an obvious conclusion. In addition, sudden drops of the
mechanical properties at 75°C and 90°C can be explained such that the moisture
which contributed to the chemical degradation caused the chain scission at high

temperature, lowering the mechanical properties.

In DGEBA/D230, di-functional epoxy system, the drop of the mechanical
properties between the samples saturated at 75°C and those saturated at 90°C was
not as drastic as those seen in MY510/D230 and MY721/D230 epoxy systems.

Comparing between the MY510/D230 and MY721/D230, the percentage drop of

the mechanical properties of the MY721/D230 epoxy systems was greater than
that of the MY510/D230. Therefore, it can be concluded that as the functionality
of the epoxy increases, the network formed by the epoxy with higher functional
groups become more susceptible to the chemical degradation, and results in the

catastrophic failure due to the chain scissions induced by the moisture.

To aid visualizing the decrease in the mechanical properties of the moisture
saturated samples, few bar graphs were constructed. % strain at break of each
epoxy system at different soaking temperatures were shown in Figure 5-4. From
this bar-graph, it can be easily noticed that % strain at break gradually increases
as the soaking temperature increases. Then suddenly at a certain temperature, %
strain at break plunges. The increase of the % strain at break in the beginning can
be explained as due to the moisture acting as a plasticizer such that the mobility of

the chemical chain is increased resulting in the increase in the % strain at break.

As the soaking time increases, however, the polymer network starts to
degrade and eventually at a critical point, the network is encountered with chain
scission. This reduces the structural integrity of the polymer network and
resulting in a sudden drop of the % strain at break. This argument can also be
supported by Figure 5-5 and Figure 5-6 where the change of load at break and

Young’s modulus are provided, respectively.

89

 

 

 

 

 

      

 

  
   

 

 

 

 

 

 

 

35
29.99
30 <
DDGEBA/DZ3O
IIIIMY510/0230
A 25 ' EMY721/0230
a; l
g
= 20 -
111
u. 17.03
111'
c 15 .4 14.2
'g 12.83 ‘3-
11.65
m 11.28
23 .
10 i 8. 8.01
8.50 E E
5 " E E 3.16
g E «.7.
o ._ E E
DRY 45C 60C 75C 900

Figure 5-4: Change of % strain at failure (%) due to the moisture absorption.

 

90

0.45

 

0.4105

 

0-41- 01365811170230

Ill MY510/0230
E MY721/0230

 

 

0.35 1 ~

 

.°
(.8

0.2443

.°
N
01

Load at Break (KN)
.0
N

0.15 ‘1

0.1 4»

0.05 1 ~

 

 

|l||||llllllllllllllllllllllllllllllllllllllllllllllllllllllllllIllllllllllllll|||||llllllIlllllIll|||lllllllllllllllllllllllllllllll
lllllllll||Illlllllllllllllllllllllllllll||llllllllllllllllllllllllllllllIlllllllllllllll‘llllllllllllllllllllllllllllllllllllllllllll

 

ll||lllllllllllllllllllllllllllllllll|||llllllllllllllllllllll|||llllllllllllllllllllllllllllllllllllllllllllllllllllllll|||l|

 

 

llllllllllllllllllllll|||l|IIlllllllllllllllllllllllllllll|||||||||||lllIlllIll

 

 

 

 

DRY 450 600 756 900

Figure 5-5: Variation of load at failure (kN) due to the moisture absorption.

 

91

 

2000

El DGEBA/0230

llIlMY510/0230 l

l E MY721/D230
1582

  
 

1800 -~

 

1600 0

1400 <-

1200 ~-

1000 -

800 ‘-

600 <-

Young's Modulus (MPa)

400 <-

200 -»

 

 

 

 

 

 

 

Figure 5-6: Variation of Young’s modulus (MPa) due to the moisture absorption.

92

5.4 Summary

As the sample conditioning temperature increased, the degree of
discoloration of the sample became severe such that a light colored, transparent
sample when it is dry turned into a dark, almost black, and opaque sample.
Prolonged submersion in the water caused discoloration, and it was suggested that
the color change was a direct result of oxidation. Discoloration was not limited

to just one epoxy resin but it appeared in all three.

Besides the findings of color change during the immersion in water, it was
also found that samples immersed in water at high temperatures, especially at
90°C encountered with substantial amount of chemical degradation. The
degradation was so severe that the sample immersed at 90°C snapped even before
the load was applied, during the sample mounting process. Water at higher
temperatures gave much more damage to the samples with higher functionality
such that the overall mechanical properties of MY721/D230 decreased in a great
deal. Epoxy resins of high functionality are more susceptible to the chemical

degradation than those of low functionality.

6. NUCLEAR MAGNETIC RESONANCE SPECTROSCOPY

6.1 Introduction

Nuclear magnetic resonance (NMR) spectroscopy is the experimental study
of the energy levels of certain atomic nuclei of molecules in a magnetic field [42].
The magnetic properties of the nuclei dominate in a magnetic field and all atomic
nuclei possess a characteristic known as nuclear spin [43]. Certain atomic nuclei,
having an odd number of either protons or neutrons possess a nonzero spin that
can be observed by the NMR spectrometer. Such magnetic nuclei behave as
though they were spinning on an axis [44]. A wide variety of nuclei, for instance,
the principal isotopes of hydrogen (1H), carbon (”C ), nitrogen (”N ), fluorine

(”F ), and phosphorus (3 1P) have suitable magnetic pr0perties [43, 44].

A proton is the simplest nucleus, and its odd atomic number (1) implies
that it has a spin [44]. We can visualize a spinning proton as a rotating sphere of
positive charge [44]. The movement of charge is similar to an electric current in a
loop of wire. As the positively charged nucleus spins on its axis, it generates a
magnetic field (H.) that resembles to the field of a small bar magnet Figure 6-1.
This magnetic field is called the magnetic moment of the nucleus. A magnetic
moment is created by moving charge, and when a small bar magnet is placed in the

field of a larger magnet, as shown in Figure 6-2, the magnetic moment tends to

93

94

\D U, \390/ j 0,

I a m 3 A MK, I a . a

 

 

 

 

 

 

 

N 28:85 N
E E

mumaamam 390: .02. om 8:63 awn Samba»

Emcna 97 > $33.5 368: manna—Sm a Bmmaomo moE. 8:3 5 Samson.“
3632:. .25 Samson.“ moi AIL 33328 :5. cm 8 man: can
Bumso». Amoh‘mmfids 5C5

95

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\ > ., > N > >
I o m o
W K w
z _ z _
.622 5.5.3~ 3554 2.9.5.
:55 @825 5mm 5.55

Emcqo ob. >= 9.525. 335.5 :55 $5 52:55 m 325 8 m «:5: 3.. Emma“...
23:5 :5 3: 3.5.5. 8 Emma 3 £5. :5 9.53.: :53. .25
wanwsmoaaa cm :5 65.. 3.55. 5:85: 3:. :5 :55 mm .35.. m:
323. :5: :5 8.3.5535... 9.5.5: amour... :5 :55. AkthkEdS

~35

96

align itself with the field of the larger magnet - a lower-energy arrangement than

an orientation against the field [43, 44].

The same effect can be observed when an external magnetic field (H,) is
applied to a proton, as shown in Figure 6-3. Quantum mechanics requires the
magnetic moment of the proton to be aligned either with the external field or
against the field. The state with the proton aligned with the field corresponds to
a spin quantum number (I) of +1/2, called the alpha-spin (ct-spin) state; the state
with the proton aligned against the external magnetic field corresponds to a spin
quantum number I = -1/2, called the beta-spin (B-spin) state [44]. Since the [3-
spin state is aligned against the external magnetic field, its energy state is higher

than the a-spin state.

Quantum mechanics indicates that a nucleus with spin 1. is characterized by
an angular'momentum with a spin quantum number I and is related to the

magnetic moment, p, by Eq. 6-1;

#=7'71
Eq. 6-1

where 7 is the gyromagnetic ratio (rad/G-s), a constant that depends on the

magnetic moment of the nucleus, and I; is Planck’s constant divided by 27:.

Nuclei with a spin number of zero (I = 0) are not observable in NMR

experiments. There are 118 nuclei that have been studied by NMR [43], but the

97

H0

(externally applied field)

 

I = +1/2 I = -1/2
a-spin state B-spin state
(lower energy) (higher energy)

Figure 6-3: In the presence of an external magnetic field (Ho), a proton has a spin
quantum number (I) of either +l/2 or -1/2. The state with spin 1 =
+1/2 is called the a-spin state, aligned with the external magnetic
field. The state with spin I = -1/2 is called the B-spin state, aligned
against the external field. The B-spin state is higher in energy than
the a-spin state. Adapted from [43].

98

nuclei of primary interest to polymer chemists are the proton (H, I = 1/2),
deuteron (2H, 1 = 1), 13C (I = 1/2) (for the more common 12C isotope, I = 0), and

”F (I = 1/2) [43].

The two factors that determine the sensitivity and utility of a nucleus for
NMR spectroscopy are the natural abundance and the gyromagnetic ratio [44].
For example, the 1H isotope is 100% naturally abundant and has a high
gyromagnetic ratio, and this makes 1H the most sensitive nucleus for NMR study.
On the other hand, the 13C nucleus is present in natural abundance at a level of
only 1.1% and has a gyromagnetic ratio that is 1/4 that of hydrogen to make it 1.6

x 10'2 less sensitive than hydrogen for NMR study [43].

In the absence of a magnetic field, the proton magnetic moments have
random orientations, Fig. 6-4 (a) [43]. However, when a sample is placed in a
homogenous magnetic field, the dipoles will align with the lines of induction or
the force of the applied magnetic field and each proton assumes the on state or the
B state (b). Because of the energy difference between the spin states, there are
slightly more a spins than B spins. Molecules are constantly in thermal motion,
and because the molecules interact with each other, the thermal motion will cause
the magnetic moments of most of the protons to point randomly. However, the
average or net magnetization will be preferentially aligned along the magnetic
field (c). The average of all these magnetic moments is called the thermal

equilibrium magnetization, M0. The intensity of the NMR signal is proportional

99

to M. and is directly related to the magnitude of the applied field. The
macroscopic nuclear magnetization, M, is the vector sum along H, of the each
spin moments. Because the nuclei have angular mo/menta and thermal motions,
they will not align completely parallel to the field, and the torque from the applied
field will cause the magnetic moments to precess about the field direction with a
characteristic angular frequency. The precession frequency is known as the
Larmor frequency (d) and is proportional to the'applied magnetic field, H... The

Larmor frequency, (00 is given by

Eq. 6-2
in radians per second.
The energy difference between the two spin states, a-spin state and B-spin

state, is proportional to the strength of the magnetic field and can be expressed as

in Eq. 6-3

Eq. 6-3

where AE is the energy difference between 01 and [3 states, It is Planck’s constant,
H. is strength of the external magnetic field (in gauss), and y is gyromagnetic

ratio, 26,753 sec’1 gauss'l for a proton. This equation shows that the energy

100

 

 

 

 

2
HO
Mo
/ S y
x
(a) (b)
2 Z
A
H0 H0 (00
N M v, -------------- '
t... _____ _ 21*
M
A, y > Y
X X
(9) (d)

Figure 6-4: The nuclear spin generates a small magnetic field, and in the absence
of an applied magnetic field, the orientation of these dipoles is random
(a). When a sample is placed in a homogenous magnetic field, the
dipoles will align themselves with the lines of induction or force of the
applied magnetic field (b). However, thermal motion does not allow
the alignment to remain, and the average alignment is at a small angle
to the magnetic field (c). The magnetic moment precesses about the
magnetic field at the Larmor frequency (d). Adapted from [43].

101

difference (AE) between the a-spin and B-spin states is proportional to the
strength of the magnetic field (H.) and to the gyromagnetic ratio (y). The energy
levels that occur in the presence of a static magnetic field appear as shown in
Figure 6-5. When a proton is struck by a photon with just the right amount of
electromagnetic energy, the proton’s spin can flip flop from 01 to B or from B to a.
A nucleus aligned with the field can absorb the energy needed to “flip” and align

against the field.

The energy difference falls in the radiofrequency range (107 - 108 Hz). The
nuclei can be induced into a higher energy state by absorption of a radiofrequency
pulse of the appropriate frequency and strength [43]. This radiofrequency, rf, is
produced by using an alternating current of variable frequency that is passed
through a coil whose axis lies in the xy plane perpendicular to the applied
magnetic field. This current excites an oscillating magnetic field that is also
perpendicular to the applied field. As the frequency of the oscillating field varies,
there is a point at which it exactly matches the precessional frequency of the
nuclei, and energy from the radiofrequency is absorbed by the nuclei. When this
absorption occurs, the system is said to be in resonance, as shown in Figure 6-6,

and its absorption of energy is detected by the NMR spectrometer [43].

This absorption of energy is the resonance phenomenon. In resonance,
energy is transferred from the rf radiation to the nuclei, which causes a change in

the spin orientation of the nuclei or, in other words, a change in the spin

102

 

 

 

 

 

I=0
Energy
I=1/2 7N H0 (+1/2)
Energy
yN H0 (-1/2)
7,, H0 (+1)
Energy yN HO (0)
7,, HO (-1)

 

 

Magnetic field Ho

Figure 6-5: Quantized energies of nuclei in a magnetic field. Adapted from [43].

103

B state

hv=AE

(I. state

 

Figure 6-6: A nucleus in “in resonance” when it is irradiated with radio-frequency
photons having energy equal to the energy difference between the spin
states. Under these conditions, a proton in the a-spin state can
absorb a photon and change to the B-spin state. Adapted from [44].

104

populations in the various energy levels. The rf field must oscillate with a
frequency, v, given by the equation of E =hv, showing that the energy E is in
proportion to v, the electromagnetic wave frequency. This equation is put
together with the equation for the energy difference between the spin states to
obtain a relationship between magnetic field and frequency. The Eq. 6-4 shows
that the resonance frequency, v, of a nucleus is in proportion to the applied

magnetic field, 11., and the gyromagnetic ratio;
v - —1— 71‘!
Zn °

Eq. 6-4.

6.2 The Spin-Lattice Relaxation Time (T1)

After the resonance rf pulse has been applied, a higher energy excited state
is present [43]. This higher energy level corresponds to the nuclei that are
elevated to the antiparallel position. To return to the equilibrium or ground state,
this excess energy is passed to the surroundings by stimulated emission, and some
of the antiparallel nuclei revert to the parallel or low-energy state. This decay of
the magnetization proceeds to equilibrium in an exponential manner. The spin-
lattice relaxation time, T1, measuring the longitudinal magnetic relaxation time is

one of the methods to measure the decay of the magnetization [42].

The rate of return to equilibrium along the static field (z axis) is depended

upon the rate of exchange of energy between the nuclei and the environment or,

105

lattice in NMR terminology (Figure 6-4). The longitudinal magnetic relaxation
time, T1, is often called the spin-lattice relaxation time and depended upon the
effectiveness of energy transfer from the excited nuclei to the lattice. Returning
to the ground state is enhanced by the presence of rf energy in the lattice at the
resonant frequency (stimulated emission). This rf energy comes from fluctuating
magnetic nuclei and electrons in the lattice. Magnetic fluctuations that are at or
near the resonant Larmor frequency generate rf magnetic fields which stimulate
the transition from the high-energy to the low-energy states. This coupling
mechanism allows the energy initially added by the rf pulse to be dissipated to the

lattice [42].

The T 1 is unique for each molecule and its environment, and provides
information about the molecular dynamics of the magnetic moments and the
molecules around them. When the nuclear magnetization rotates at the Larmor
frequency, the local fluctuating magnetic fields (perpendicular to the rf axis) are
generated primarily by molecular motion [43]. Thus when the lattice exhibits
considerable molecular motion at the appropriate frequency, the energy coupling
is effective, and the T 1 is short. On this basis, the T 1 is short for liquids where the
molecular motion is extensive over a broad range of frequencies but is quite long
for solids where the motion is restricted and occurs over a narrow range of

frequencies [43].

106

6.3 Cross-Polarization (CP)

Because of its inherent low sensitivity, the 13C isotope addresses a
challenge for NMR spectroscopy [43]. The observable l3C signal is limited due to
the low gyromagnetic ratio, the low natural abundance, and the long spin-lattice
relaxation time of the 13C nucleus. The sensitivity must be increased for the
detection of the resonance of the 13C nuclei. In response to this need,
polarization-transfer methods are employed to transfer the large spin-state
polarization of the protons to the weakly polarized l3C nuclear species. If it is
provided with perfect transfer of polarization from protons to 13C, the 13 C signal

will be improved by a factor of yH/yc = 4.

The CT 1 spin-lattice relaxation times for solids are significantly longer than
the ”T, times, so the maximum number of pulses that can be used is less.
However, with polarization transfer, the delay time between cross-polarization
pulse sequences is depended upon the shorter ”T; times, and further gains in
signal enhancement becomes the extent of (CT/1T1)"2 [43]. In addition, it would
be possible to perform signal averaging experiments much more rapidly. This can

be achieved as follows.

Let us designate the carbon coordinates as x” and y” since they are rotating
at a different speed to x’ and y’, the proton coordinates. Immediately following

the 90° lH pulse and during the 1H spin locking condition radiofrequency along

107

the coordinate x” describing the 13 C magnetization is applied. However, the ‘3 C

power level is altered so that
73H”! = 7cHic

Eq. 6-5.
This is called the Hartmann-Hahn condition or match [43].

Cross-polarization is a double-resonance experiment in which the energy
levels of the 1H and '3 C spins are equalized to the Hartman-Hahn condition in the
rotating frame [43]. Under this condition, energy may be exchanged between the
two coupled spin systems. The result is a growth of the 13 C magnetization at the
expense of the 1H magnetization. The basic pulse sequence used for the cross-
polarization experiment is shown in Figure 6-7. The method is comprised of four
steps: (a) polarization of the 1H spin system, (b) spin-locking of the protons in the
rotating frame, (c) establishment of contact between the carbons and the protons,

and ((1) measurement of the magnetization of the carbon nucleus [43].

The application of a 90° lH pulse rotates the 1H longitudinal magnetization
into transverse magnetization (the xy plane is in the rotating reference frame as
shown in Figure 6-7 (a)). The second step is the nearly immediate (within
microseconds) application of a 90° phase-shifted proton pulse, which produces the
effective field colinear with the proton magnetization and results in spin locking,

Figure 6-7 (b). Spin locking is utilized because the magnetization of the protons

108

in the spin-locked state decays by a rotating frame spin-lattice relaxation time,
H T 1,. Therefore, the magnetization loss of the protons resulting from relaxation

effects during the contact time is small [43].

Simultaneously with the spin locking of the protons, in the third step of the
experiment Figure 6-7 (c), a pulse is applied in the 13C channel, and this pulse is
carefully adjusted so that the energy gap for spin flips agrees exactly to that of the
protons. This pulse is maintained for a time, tcp, and is called the contact time.
This contact time allows for the exchange of energy between the abundant proton-
spin reservoir and the rare carbon-spin system. This exchange, called cross-

polarization, occurs when the Hartmann-Hahn match, which is defined as

wn=wc

Eq. ,6-6

is contended with the rotating frame [43].
The fourth part of the CP experiment Figure 6-7 (d) is to terminate the 13C
pulse and to observe the FID (free induction domain, the observed signal in the
time domain) while the 1H field is used for decoupling. The resulting FID is

Fourier transformed to give the frequency-domain spectrum [43].

In the CP experiment, the 13C spin system is polarized by the 1H spin
system under the Hartmann-Hahn condition, Eq. 6-5. Thus, the longer CT, (1-

1000 s) is replaced by the shorter H T 1 spin-lattice time (1-100 ms) [43].

109

90° Pulse

Irradiation at H frequency

 

 

 

 

 

 

(a)
/ Decay of 1H in Rotating Frame
------------------ l
H Spin-Lock H111
(b)

 

Buildup of 13 C Magnetization
/

 

 

 

 

 

Irradiation at 13 C Frequency

 

Observe

.1

(d)

 

 

 

 

l
= msec = msec

7 _ x
= use tlme —>

C

Figure 6-7: The timing sequence for the cross-polarization experiment. Adapted

from [43].

110

The optimum recycle delay has been measured, and it is 1.25 H T 1 for the situation
in which spin diffusion dominated the spin system [43]. For the case in which spin
diffusion is not efficient, the optimum recycle delay must be longer (4-5 H T 1) to

avoid saturation.

A typical increase in carbon magnetization with contact time is shown in
Figure 6-10. In the CP experiment, polarization is transferred from the
magnetically rich protons to the magnetically poor carbons via their static dipolar
interactions [42]. When both spins are locked in the rotating rf fields and their
amplitudes are equalized by utilizing the Hartmann-Hahn condition, Eq. 6-5, the
time constant for the transfer of polarization under conditions of spin locking is
called the T CH relaxation time. This transfer is spin-spin process, and generally

has a magnitude in the range of 100115.

The result of CP is an initial growth of the 13 C magnetization at a rate that
is inversely proportional to the cross-polarization-transfer rate constant (Tag)
Figure 6-10. After a suitable period (typically 1-10ms), a maximum l3C
magnetization is reached, after which the 13C signal begins to decay at a rate in

proportion to the inverse of the H T 1,, time. The carbon magnetization, S, is given

1 ex -t
S = CXPL H m j[1- ——-p( OJ]
Tlp TCH

by

 

Eq. 6-7

111

where top is the contact time. Thus two opposing relaxation mechanisms operate
during the contact time. For short contact times, the process is dominated by the
Tc” process, and the carbon magnetization increases exponentially. At longer
contact times, the H T 1,, process dominates, and the magnetization decreases

exponentially because of the proton spin-lattice relaxation in the rotating frame.

Provided that the carbons in a polymer repeat unit are subjected to the
same motions, carbons with directly bonded protons are expected to cross-
polarize more rapidly than carbons without direct interactions because the shorter
interaction distance results in a larger local dipolar field. Rapid molecular motion
attenuates the cross-polarization mechanism. For protonated carbons in rigid
solids, most of the signal buildup occurs over about the first 100 us; for

nonprotonated carbons, in the first few milliseconds [43].

The rate of signal decay depends on H T 1,, [43]. If H T 1,, is very short, the
cross-polarization process is jeopardized because of the rapid decay process. By
varying the contact time prior to observation of the spectrum, T CH and H T 1,, may

be determined for the protons coupled with each resolved carbon resonance [43].

6.4 Proton Spin-Lattice Relaxation Times in the Rotating Frame

6.4.1 The rotating frame

The nucleus in the sample possesses a magnetic moment, charge and

angular momentum, and as the magnetic moment interacts with magnetic field H.

112

it will do so in such a way that the field will force the moment to line up with it
[42]. If the nucleus was identical with a simple bar magnet, the moment would
align along the direction of the magnetic field. The latter, by convention, is
termed the z direction. However, unlike a simple bar magnet the nuclei have
angular momentum and the effect is to force the nuclei; therefore, the magnetic
moment to precess about the field H... Figure 6-8 [42]. This motion is analogous
to the spinning top which precesses about the earth’s gravitational field. The
precession velocity 600 of the moment (Larmor speed) is determined by the
gyromagnetic ratio of the nucleus 7 and the strength of the magnetic field as
shown in Eq. 6-2. To express a) in units of hertz rather than radians, a) is divided

by 271', this equation is also shown in Eq. 6-4 [42].

Interaction of the electromagnetic radiation with the moment of the nuclei
when inducing transitions between energy states can be understood by vector
diagrams, since the magnetic moment is a vector quantity, if the coordinates used

for the discussion are changed to rotating frame coordinates [42].

After the sample is put in the magnetic field, the magnetic moment of the
nucleus presecces at the Larmor frequency with the direction of the magnetic field
induced by the magnet, i.e. around 2 in Figure 6-9 (a). The coordinates x, y are
the other coordinates of three-dimensional space in the laboratory of the
laboratory ‘frame’. From nucleus’ point of view, they are rotating at the Larmor

frequency. If we change the x, y coordinates to one rotating at the rotating frame

113

MAGNETIC FIELD
A

Z

 

 

 

 

 

 

 

Figure 6-8: Diagram to show the motion of a spinning nucleus in a magnetic field
of direction 2. Adapted from [48].

114

(x’, y’), i.e. at the Larmor frequency, the magnetic moment appears ‘frozen’ on the
cone Figure 6-9 (a) [42]. In reality, there will be a large number of nuclei; each
will have a magnetic moment frozen at a different position around the cone. The
sum of these moments, the net magnetization, will lie along z-axis (now termed 2’,
Figure 6-9 (b). It will be recalled that the radiofrequency irradiation at resonance
is at the Larmor frequency as well. If this is applied in the laboratory frame at 90°
to the magnetic field H., the magnetic field from the electromagnetic irradiation
(H1) rotates in the x, y plane and at resonance is static ‘frozen’ along x’ in the

rotating frame Figure 6-9 (b) [42].

115

 

 

 

 

 

 

 

 

 

    

(h)
z!
M
y Y'
X X' H1
z' z!
(c) (‘1)
90°
y! 'A y!
M
x! H1 , x1

Figure 6-9: Changing coordinates to the rotating frame: (a) rotation of magnetic
moment 11 about laboratory coordinates; (b) change of coordinates to rotating
frame and application of a radiofrequency induced magnetic field along x’; (c)
after application of field H1 along x’ (note the vector M moves through the z’y’
plane by 90° to become colinear with y’); (d) after changing the phase of the
irradiation field so that H1 is colinear with M and y’. Adapted from [48].

116

6.4.2 Proton spin-lattice relaxation times in the rotating frame

Proton T1,, relaxation in the rotating frame, ” T ,p, is the T 1 process in the
rotating frame that takes place for the proton nuclei [42]. When proton T1,,
relaxations are measured, because 0 spin diffusion, the observed relaxation times
are averaged over all of the coupled protons. Individual proton T ,ps, however,
can be characterized by variations in the normal 13C CP experiments.
Measurements of H T “,5 have the advantage that they probe the molecular motion
in the kilohertz range, whereas the T18 reflect motion in the megahertz range. On
the basis of differences in the spectral density in the kilohertz region, the H T has

may be weakly depended upon the rf field [42].

The H T1,,s are obtained from the final slopes of the plots of observed carbon
magnetization generated by log, equalized spin-lock cross-polarization transfers
from protons to carbons [43]. After the initial increase carbon polarization, the
carbon signal decreases by way of the ”T1,, relaxation process. Under high-
resolution conditions, the individual carbon resonance follow the protons to which
they are most closely coupled. Therefore, it is possible to resolve ” T 1,, differences
between protons on different types of carbons [43]. The measurement of
individual ”TIPS involves the exordium of a variable contact time of cross-
polarization. A semi-log plot of the carbon magnetization vs. the time of spin-

lock cross-polarization transfer generates H T 1 p.

117

As an example, heterogeneities in blends of polymers can be detected
because the H T ,ps are sensitive to spin diffusion in polymer systems on the order
of an internuclear distance of 1 nm. For single-phase systems of multicomponent
systems, the rate of spin diffusion depends on the nature of the spatial mixing of
the polymer chains. When the chains are intimately and homogeneously mixed,
only one H T 1,, may be observed. If the mixing is nonuniform, several different

” T 1 ps may be detected depending on the system [49].

6.5 Sample Preparation for the T c” and H T 1,, Experiment

For each epoxy system, three different types of samples have been
prepared; dry (unconditioned), wet (saturated at 75° C), and dried (wet samples
that had been dried at 100° C under vacuum for 24 hours) samples. Samples were
cut into the dimensions of 1.0 x 0.4 x 0.1 cm’ so that they can fit into a sample
spin tube. Both dry and dried samples were stored individually prior to NMR
experiment at room temperature in polypropylene tubes containing desiccants, and
in particular, each wet samples were stored in distilled water filled polypropylene

tubes to minimize any moisture loss between the experiments.

6.6 Experimental Procedure

The experiments were conducted by Varian VXR-400 NMR spectrometer.
The measurements were obtained by using combined techniques of cross-

polarization and magic-angle spinning at the operating frequency of 100 Mhz.

118

The spinning speed was of the order of 4 kHz. Contact time for the cross-
polarization relaxation time was from 10 us to 2500 us. Total acquisition time

for one sample was 12.6 hours.

6.7 Results and Discussion

6.7.1 lH Spin-Lattice Relaxation Times in the Rotating Frame

”T1,,s on dry, wet, and moisture desorbed samples were obtained for each
epoxy group samples as shown in Table 6-1 through Table 6-3. One of the I vs.

contact time plots is shown in Figure 6-10.

” T 1,, results show that the moisture saturated DGEBA/D230 sample has a
higher ”T,,, values over the range of the peak frequencies, compared to the
unconditioned sample. The increased H T 1,, values, however, returned to its initial
values as the sample has been dried. Unlike the behavior shown by the
DGEBA/D230 epoxy resin, MY510/D230 and MY721/D230 revealed different
characteristics in their proton T1,, relaxation in the rotating frame. Wet samples of
MY510/D230 and MY721/D230 showed insignificant change in H T 1,, comparing
with H T 1,, of dry samples. Even afier the samples have been dried, H T 1,, of dry,

wet, and dried samples could not be differentiated from one another.

 

119

Table 6-1: T C” and "T1,, of DGEBA/D230

 

 

 

 

 

 

 

Dry 75° c 75° c - Dried

Peak freq. Tc” ’7 7,, Peak freq. Tc” ’7 7",, Peak freq. Tc” ”7,,
148.89 6.42 1436.6 146.25 760.21 753.52
62.08 11.44 572.18 64.08 4.10 4971.4 58.26 7.46 478.58
12.65 20.02 142.56 914.28 19.23 88.24 760.59
5.22 125.05 558.58 4.64 137.17 1011.3 4.82 106.36 620.86
-243 116.72 603.56 -2.19 136.94 914.76 -1.28 109.16 617.29

 

 

 

 

 

 

 

 

 

 

 

120

Table 6-2: Tc” and ”T1,, of MY5 10/D230

 

 

 

 

 

 

 

 

Dry 75° C 75° C - Dried
Peak freq. Tc” T1,, Peak freq. Tc” T,,, Peak freq. Tc” T1,,
146.40 195.89 963.56 147.50 37.67 773.59 154.11 50.76 539.08
105.80 60.96 776.43 104.56 8.39 568.01 106.26 8.06 614.81
61.87 12.89 601.11 62.90 12.17 567.02 60.80 12.60 574.13
5.64 92.62 805.99 5.61 124.16 618.35 6.19 103.85 645.68
-l.67 100.49 686.97 -l.83 87.62 690.23 -l.70 107.94 608.20

 

 

 

 

 

 

 

 

 

 

121

Table 6-3: T c” and H T 1,, of MY721/D230

 

 

 

 

 

 

 

Dry - 75° C 75° C - Dried
Peak freq. Tc” ”7,, Peak freq. Tc” #7,, Peak freq. TC” T
122.40 11.86 620.69 120.52 33.88 426.20
63.90 12.77 635.83 61.32 14.78 522.46 65.54 11.56 550.82
6.03 112.37 765.78 6.52 102.01 887.58 6.49 94.062 840.55
-2.68 92.96 674.29 -2.37 94.10 656.99 -2.65 97.18 588.85

 

 

 

 

 

 

 

 

 

 

Intensity

122

 

26- .
- .3

24-

22- £7

20- ,'\l

 

 

 

 

 

 

 

6‘1,

1.

417

2'. r

o L! fi
1

 

-
‘V‘i 'i-YI. 1V.'IY‘YITI‘IITTFI'T'V..OK-YIY‘VYIYW‘IY!.TY‘I'VIIIIIYVI‘I'V—IIWYIY.YVhi‘l’T‘lT‘ll‘ ‘I‘VIYIYIi'l‘VYV'VVI‘l‘UVIIVVIIi'jv‘IFVIVIIIIU'V'

0 200 400 600 000 1000 1200 1400 1600 1000 2000 2200 2400 2500

9

us

Figure 6-10: The change in carbon magnetization with contact time for the cross-
polarization experiment. The initial rise is due to the cross-polarization contact

time, T CH, and the relaxation decrease is governed by the H T ,p.

123

H T 1,, provides information about the molecular dynamics of the magnetic
moments and molecules around them. Also, the measurements of individual H T 1,,
involves the introduction of a variable contact time of cross polarization. When
water molecules were introduced into the epoxy networks through sorption
process, energy exchange occurs between 1H and 13 C spin systems. The result is a
growth of the 13C magnetization at the expense of the 1H magnetization. Thus, if
mobile water molecules are present in the epoxy network, establishment of
contact between the carbons and the protons will take some time. Therefore, as
more mobile water molecules are present in the epoxy network, H T 1,, will become

longer

H T 1,, is a direct representation of the water molecule mobility. As
mentioned above, ”T1,, of moisture saturated DGEBA/D230 samples were
distinctly higher than those of MY721 and MY510 epoxy glasses. In PALS
experiment, it was said that 13 measures the size of the hole, and when comparing
1'3 of the three epoxy systems, I; of DGEBA/D230 was the highest. In addition,
VanderHart et el [41] reported that when water goes into the epoxy, a majority of
water molecularly dispersed in the epoxy rather than aggregated in a waterlike
state within voids (i.e. if voids are present, they can only accommodate one or
two water molecules). Therefore, if the size of the hole is larger, the motion of
water molecule will less likely be hindered. On the other hand, compared to

DGEBA/D230, MY510 and MY721/D230 are tightly crosslinked polymer network

124

such that the size of free volume is smaller; therefore, it is less likely to show any

significant increase in ”T,,,

To further investigate the exact nature of the water in the crosslink

network, quadrupole echo deuterium NMR spectroscopy was suggested.

6.8 Quadrupole Echo Deuterium NMR Experiment

Results from the conventional broad-line NMR spectroscopy were not
satisfactory due to the broad signals it is giving out. For instance, when
comparing spectra between dry and moisture desorbed samples to determine the
nature of the moisture-epoxy chain interaction, it was difficult to affirm the water
molecules were whether still linked to the chains after the desorption or all

completely desorbed.

Quadrupole echo deuterium NMR provides an useful spectroscopic method
for the study of water (i.e. deuterium oxide, D20) absorbed by polymers [50].
The quadrupole echo pulse sequence preserves the line shape of the
inhomogeneously broadened deuterium NMR signal, thus by-passing the problems
normally incurred with broad signals and receiver ringdown. Unlike conventional
broad-line NMR counterpart, quadrupole echo deuterium NMR spectroscopy
lends itself to Fourier transform method. Thus, it is possible to observe weak
signals, using repeated spectral accumulation. In addition, the deuterium line

shapes (acquired from Fourier transformation of the free induction decays) can be

 

125

analyzed in terms of models for molecular motion. Therefore, deuterium NMR
spectroscopy is capable of providing highly specific and detailed information on a

molecular level.

6.8.1 Results and Discussion

Figure 6—11 shows the quadrupole echo deuterium 1H NMR spectrum of
deuterium oxide (D20). It is shown as a sharp peak that the deuterium absorb at a
chemical shift of 516000 (16000 ppm). While this peak operates as a finger print
for the deuterium in the epoxy network, the nature of the moisture can be studied
by examining spectra of pure D20, D20 saturated samples, and D20 desorbed

samples.

Experiments were conducted on samples saturated at 90°C only to examine
the D20 sorption and the desorption processes, disregarding the temperature
effect on D20 sorption behavior. As can be seen in Figure 6-12, mobile
deuteriums, introduced to the epoxy resins by the sorption process are depicted as
sharp peaks. All three systems clearly show the presence of the deuterium in their
networks. After the desorption process (100°C for 4 days), however, there are
no traces of deuterium presence in the epoxy resins. The results are shown in

Figure 6-13.

 

126

1300.. 20?

 

 

*7

14‘

30000 25000 20000 15000 10000 5000

Figure 6-11: Quadrupole echo deuterium 1H NMR spectrum of D20. The sharp
peak at 616000 is indicating the chemical shift of pure deuterium.

127

 

 

 

 

 

 

 

 

 

 

 

(a)
i
4
l
.11
A "hf-‘21 K‘av f,1. .7 _
' ‘ 1 ‘ 1 I . I F 1
50000 25000 20000 15000 10000 5000 pp
I
1‘
-_.J k
1 1 * . 7 1 ‘ 1 P 7 I * fT
20000 25000 20000 25000 20000 5000 P”
(C)
v
- ram-J we: - _
71* 1""TT'Y‘1'17‘If"'T“‘f1
50000 25000 20000 25000 20000 5000 7’

Figure 6-12: Quadrupole echo deuterium 1H NMR spectra of samples saturated

with D20 at 90°C; (a) DGEBA/D230; (b) MY510/D230; and (c)
MY721/D230. ‘

128

(a)

 

 

 

 

1‘11“. 111.1“1‘8‘ M 1‘. 41.111.411.111] 11 #1111 ‘11.. 1.1.1» 11.111 111111.101 11.411.411.111-
1.19111111111111111‘1‘11‘1‘1 11W [MW ill/l W 111W - 111111111le 111.111.111.11
...

M1144 «1141,49 awnLvMM WWW/01411091911111“ 11014191111115 ‘ffl‘flMHWJWfi/NM‘M 11411.1 WM

'''''''
30000 25000 20000 15000 10000 5000 pp

Figure 6-13: Quadrupole echo deuterium 1H NMR spectra of D20 desorbed
samples (100°C for 4 days); (a) DGEBA/D230; (b) MY510/D230; and
(c) MY721/D230.

129

6.9 Summary

Because the water molecules were highly mobile, DGEBA/D230 epoxy
glass needed longer time to establish contact between the carbons and the
protons. In addition, since the majority of water molecularly dispersed in the
epoxy, rather than aggregated in a waterlike state within voids, as water
molecules go into the epoxy network, it will certainly help the motion of water
molecules if the size of free volume is large. From PALS results, it was found
that DGEBA/D230 has larger free volume. Therefore, H T 1,, of wet samples was
significantly higher than that of dry samples. It is noteworthy that after the
desorption process, H T .p returned to its initial value, suggesting that the absorbed

moisture was simply occupying free volume.

Similar results were found in quadrupole echo deuterium NMR experiment.
Figure 6-13 shows no signs of deuterium present in the network after the samples
have undergone the desorption process. Based on this finding, one can conclude
that the moisture absorbed in the epoxy resins is not chemically bonded to the
crosslink network, but rather is simply occupying the voids in the crosslink
network such that when the samples were heated up, D20 molecules were able to

escape out of the epoxy resins.

 

7. CONCLUSIONS

Since the crosslink density increases with the resin functionality, it can be
said that the resin functionality is directly related to the diffusion in an epoxy
glass such that when an epoxy glass is formed by a reaction between hardener and
a monomer with a higher number of functional groups, diffusivity of the epoxy
glass becomes lower. It was also found that the highly crosslinked epoxy glass

was more susceptible to the hygrothermal degradation.

Not only the crosslink density, but also the thickness variation affected the
diffusion behavior in the epoxy network. Swelling effect made a great
contribution in this phenomena such that thicker the sample, lower the diffusion

coefficient and higher the equilibrium moisture content.

It is also noteworthy that under a constant conditioning temperature, as

AT:

 

fi—El increases, the diffusion coefficient decreases, where T. is an

experiment temperature and T g is a glass transition temperature. In other words,
as the conditioning temperature is further away from T 3, it becomes more difficult

for the diffusants to penetrate through the epoxy network.

As discussed earlier, crosslink density increases with a functionality,
resulting in the increase of T g. Moisture uptake, however, was depended upon not
crosslink density, but available hydroxyl sites per mole in the epoxy resin such

that MY510/D230 epoxy glass absorbed the most wt% of water, followed by the

130

 

131

MY721/D230 and the DGEBA/D230. Therefore, it can be concluded that the
hydroxyl site availability in the epoxy network is the dominant factor determining

the % weight gain of the moisture.

Moisture effects on physical properties were extensive such that
plasticization of the polymer was observed. In addition, there was a prevalent
conformational change in the polymer network during sorption and desorption
processes. In terms of the localization of the water molecules in the epoxy
network, positron annihilation lifetime spectroscopy (PALS) provided a great deal
of information. It has shown that even though water molecules primarily tackle
the hydroxyl sites, they occupy free volume between polymer molecules. Upon
desorption, those water molecules, occupying free volume, simply desorbed away.
The solid state NMR of 1H spin-lattice relaxation times in the rotating frame and
quadrupole echo deuterium NMR results also showed that there was no prominent

evidence of water molecules chemically react with the polymer network.

Immersion of the epoxy resins in water at elevated temperatures for an
extended period of time induced oxidation and chemical degradation resulting in
discoloration of the specimens and significant decrease in mechanical properties.
Discoloration and mechanical property drop were most severe with the immersion
temperatures. In addition, epoxy resins of higher functionality were more
susceptible to the chemical degradation than those of lower functionality such that

the MY721/D230 epoxy resin suffered oxidation and degradation in a great deal.

 

132

Finally, the solid state NMR of 1H spin-lattice relaxation times in the
rotating frame and quadrupole echo deuterium 1H NMR results revealed that even
though the available hydroxyl sties are the driving force for the moisture
absorption, the moisture absorbed in the epoxy resins were not chemically bonded

to the network.

 

 

 

REFERENCES

10.

11.

12.

13.

8. REFERENCES

Chi-Hung Shen and George S. Springer, “Moisture Absorption and
Desorption of Composite Materials”, Environmental Effects on Composite
Materials, ed. G. S. Springer, (1981), 15.

Andre Lee and Gregory B. McKenna, “The physical aging response of an

. epoxy glass subjected to large stresses”, Polymer, 31, 423 (1990).

Andre Lee and Gregory B. McKenna, “Effect of crosslink density on
physical aging of epoxy networks”, Polymer, 29, 1812 (1988).

Y. Diamant, G. Marom, and L. J. Broutman, “The effect of network
structure on moisture absorption of epoxy resins”, J. Appl. Polym. Sci., 26,
3015(1981)

V. B. Gupta and L. T. Drzal, “The physical basis of moisture transport in a
cured epoxy resin system”, J. Appl. Polym. Sci., 30, 4467 (1985).

Michael J. Adamson, “Thermal expansion and swelling of cured epoxy resin
used in graphite/epoxy composite materials”, J. Mater. Sci., 15, 1736
(1980)

E. Morel, V. Bellenger, and J. Verdu, “Structure-water absorption
relationships for amine-cured epoxy resins”, Polymer, 26, 1719 (1985).

John B. Enns and John K. Gillham, “Effect of the extent of cure on the
modulus, Glass transition, Water absorption, and Density of an amine-
cured epoxy”, J. Appl. Polym. Sci., 28, 2831 (1983).

R. T. Fuller, R. E. Fornes, and J. D. Memory, “NMR Study of Water
Absorbed by Epoxy Resin”, J. Appl. Polym. Sci.,. 23, 1871 (1979).

Ned D. Danieley and Edward R. Long, Jr., “Effects of Curing on the Glass
Transition Temperature and Moisture Absorption of Neat Epoxy Resin”, J.
Polym. Sci. Chem. Ed, 19, 2443 (1981).

9C. Carfagna, A. Apicella, and L. Nicolais, “The Effect of the Prepolymer

composition of Amino-Hardened Epoxy Resins on the Water Sorption
Behavior and Plasticization”, J. Appl. Polym. Sci., 27, 105 (1982).

P. Moy and F. E. Karasz, “Epoxy-Water Interactions”, Polym. Eng. Sci.,
20,315(1980)

I. Ghorbel and D. Valentin, “Hydrothermal Effects on the Physico-
Chemical Properties of Pure and Glass Fiber Reinforced Polyester and
Vinylester Resins”, Polym. Compos., 14, 324 (1993).

133

 

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

134

Wen-1i Wu and Barry J. Bauer, “Epoxy Network Structure. 3. Neutron-
Scattering Study of Epoxies Containing Monomers of Different Molecular
Weight”, Macromolecules, 19, 1613 (1986).

Wen-1i Wu, Donald L. Hunston, Hsinjin Yang, and Richard S. Stein,
“Epoxy Network Structure. 4. A Neutron Scattering Study of Epoxies
Swollen in a Deuterated Solvent”, Macromolecules, 21, 756 (1988).

J. S. Vrentas and C. M. Vrentas, “Fickian Diffusion in Glassy Polymer-
Solvent Systems”, J. Polym. Sci. Phys. Ed., 30, 1005 (1992).

Donald S. Cohen, “Theoretical Models for Diffusion in Glassy Polymers”,
J. Polym. Sci. Phys. Ed., 21, 2057 (1983).

Donald S. Cohen, “Theoretical Models for Diffusion in Glassy Polymers
II”, J. Polym. Sci. Phys. Ed., 22, 1001 (1984).

B. De’Neve and M. E. R. Shanahan, “Water Absorption by an Epoxy Resin
and its Effect on the Mechanical Properties and Infra-Red Spectra”,
Polymer, 34, 5099 (1993).

J. J. Imaz, J. L. Rodriguez, A. Rubio, and I. Mondragon, “Hydrothermal
Environment Influence on Water Diffusion and Mechanical Behavior of
Carbon Fiber/Epoxy Laminates”, J. Mater. Sci. Lett., 10, 662 (1991).

V. Bellenger, J. Verdu, and E. Morel, “Structure-Properties Relationships
for Densely Cross-linked Epoxide-Amine Systems Based on Epoxide or
Amine Mixtures”, J. Mater. Sci., 24, 63 (1989).

Robert A. Yapel, J. Larry Duda, Xiaohong Lin, and Ernst D. von
Meerwall, “Mutual and Self-diffusion of Water in Gelatin: Experimental

Measurement and Predictive Test of Free-Volume Theory”, Polymer, 35,
2411 (1994).

John M. Zielinski and J. L. Duda, “Predicting Polymer/Solvent Diffusion
Coefficients Using F rec-Volume Theory”, AIChE J., 38, 405 (1992).

Ryszard Steller and Danuta Zuchowska, “Free Volume Effects on
Rheological and Thermal Properties of Polymers and Polymer Blends", J.
Appl. Polym. Sci., 43, 1411 (1991). '

J. S. Vrentas and C. M. Vrentas, “Evaluation of F ree-Volume Theories for
Solvent Self-Diffusion in Polymer-Solvent Systems”, J. Polym. Sci. Phys.
Ed, 31, 69 (1993).

Y. Limoge, “On the concept of Free Volume in Amorphous Materials”,
Scripta Metal]. et. Mat., 26, 809 (1992).

Christopher S. Coughlin, Kenneth A. Mauritz, and Robson F. Storey, “A
General Free Volume Based Theory for the Diffusion of Large Molecules

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

135

in Amorphous Polymers above Tg. 4. Polymer-Penetrant Interactions”,
Macromolecules, 24, 1526 (1991).

R. Kirchheim, “Sorption and Partial Molar Volume of Small Molecules in
Glassy Polymers”, Macromolecules, 25, 6952 (1992).

J. C. Moreland, G. L. Wilkes, and R. B. Turner, “Viscoelastic Behavior of
Flexible Slabstock Polyurethane Foams: Dependence on Temperature and
Relative Humidity. 1. Tensile and Compression Stress (Load) Relaxation”,
J. App]. Polym. Sci., 52, 549 (1994).

Y. Kobayashi, K. Haraya S. Hattori, and T. Sasuga, “Evaluation of
Polymer Free Volume by Positron Annihilation and Gas Diffusivity
Measurements”, Polymer, 35, 925 (1994).

Q. Deng, F. Zandiehnadem, and Y. C. Jean, “Free-Volume Distributions of
an Epoxy Polymer Probed by Positron Annihilation: Temperature
Dependence”, Macromolecules, 25, 1090 (1992).

J. Liu, Q. Deng, and Y. C. Jean, “Free-Volume Distributions of Polystyrene
Probed by Positron Annihilation: Comparison with Free-Volume Theories”,
Macromolecules, 26, 7149 (1993).

Richard C. Macqueen and Richard D. Granata, “Positron Annihilation
Spectroscopy Study of Moisture Absorption in Protective Epoxy
Coatings”, J. Polym. Sci. Phys. Ed., 31, 971 (1993).

Antonio Apicella, Luigi Nicolais, Gianni Astarita, and Enrico Drioli,
“Effects of Thermal History on Water Sorption, Elastic Properties and the
Glass Transition of Epoxy Resins”, Polymer, 20, 1143 (1979).

A. Apicella, L. Nicolais, G. Astarita, and E. Drioli, “Hygrothermal History
Dependence of Moisture Sorption Kinetics in Epoxy Resins”, Polym. Eng.
Sci., 21, 18 (1981).

Antonio Apicella and Luigi Nicolais, “Environmental Aging of Epoxy
Resins: Synergistic Effect of Sorbed Moisture, Temperature, and Applied
Stress”, Ind. Eng. Chem. Prod. Res. Dev., 20, 138 (1981).

David Shiaw-Guang Hu and Mark T. S. Lin, “Water-Polymer Interactions
and Critical Phenomena of Swelling in Inhomogeneous Poly(acrylonitrile-
acrylamide-acrylic acid)gels”, Polymer, 35, 4416 (1994).

Myung Cheon Lee and Nikolaos A. Peppas, “Water Transport in
Graphite/Epoxy Composites”, J. Appl. Polym. Sci., 47, 1349 (1993).

J. Zhou and J. P. Lucas, “Characteristics of Water in Graphite/Epoxy
Composites”, Proceedings of the American Society for Composites, Ninth
Technical Conference, 1213 (1994). ’

 

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.
51.

52.

136

Jiming Zhou and James P. Lucas, “Water-Environment Effects on
Anomalous Absorption Behavior in Graphite-Epoxy Composite”, PhD
Thesis, Michigan State University, 1996.

R. J. Schadt and D. L. VanderHart, “Solid State Proton NMR of a Glassy
Epoxy Exposed to Water”, Macromolecules, 28, 3416 (1995).

J. A. Pople, W. G. Schneider, and H. J. Bernstein, “I_I_igh Resolution
Nuclear Magnetic Resonance”, McGraw-Hill, New York, 1959.

Jack L. Koenig, “Spectroscopy of Polymers”, ACS Professional Reference
Book, American Chemical Society, Washington, DC 1992.

L. G. Wade, Jr., Organic Chemistry. Prentice-Hall, Inc., Englewood Cliffs,
NJ., 1987.

T. C. Wong and L. J. Broutman, “Moisture Diffusion in Epoxy Resins Part
I. Non-Fickian Sorption Processes”, Polym. Eng. Sci., 25, 521 (1985).

T. C. Wong and L. J. Broutman, “Water in Epoxy Resins Part II. Diffusion
Mechanism”, Polym. Eng. Sci., 25, 529 (1985).

J. Crank, “The Mathematics of Diffusion”, Ch. 12, Clarendon Press,
Oxford (1956).

Michael A. Wilson, N.M.R. Techniques and Applications in Geochemistry
and Soil Chemistry, Pergamon Press, 1987.

J. Schaefer, M. D. Sefcik, E. O. Stejskal, and R. A. McKay,
Macromolecules, 14, 188-192 (1981).

R. J. Young, “Introduction to Polymers”, 1st ed., Chapman and Hall, 1981.

Li Xie, David Gidley, Hristo Hristov, and Albert Yee, Polymer, 35, 14-17
(1994)

Liu Yang, Hrist Hristov, Albert Yee, David Gidley, Didier Bauchiere, Jean
Louise Halary, and Monnerie, Polymer, 36, 3997-4003 (1995).

 

llllllllllllllll