Doctoral theses at NTNU, 2019:215

Doctoral thesis
Andrey E. Krauklis

Andrey E. Krauklis
Environmental Aging
of Constituent Materials in
Fiber-Reinforced Polymer
Composites

Andrejs Krauklis, Member of the

Royal Society of Chemistry (MRSC)

ISBN 978-82-326-4026-3 (printed ver.)

ISBN 978-82-326-4027-0 (electronic ver.)
ISSN 1503-8181

Doctoral theses at NTNU, 2019:215

NTNU
Norwegian University of Science and Technology
Thesis for the Degree of
Philosophiae Doctor
Faculty of Engineering
Department of Mechanical and Industrial
Engineering
Andrey E. Krauklis

Environmental Aging
of Constituent Materials in
Fiber-Reinforced Polymer
Composites

Andrejs Krauklis, Member of the

Royal Society of Chemistry (MRSC)

Thesis for the Degree of Philosophiae Doctor

Trondheim, August 2019

Norwegian University of Science and Technology

Faculty of Engineering
Department of Mechanical and Industrial Engineering
NTNU
Norwegian University of Science and Technology

Thesis for the Degree of Philosophiae Doctor

Faculty of Engineering
Department of Mechanical and Industrial Engineering

© Andrey E. Krauklis

ISBN 978-82-326-4026-3 (printed ver.)

ISBN 978-82-326-4027-0 (electronic ver.)
ISSN 1503-8181

Doctoral theses at NTNU, 2019:215

Printed by NTNU Grafisk senter

 “What is the goal of the life? It's to create yourself a soul. For me, … an art... more than an industry.
And it’s the search of the human soul... Have mind. Have power. Have ambition. I wanted to do
something like that. Why not?” Alejandro Jodorowsky, in Jodorowsky’s Dune.

“I have learned to be content with whatever I have. I know what it is to have little, and I know
what it is to have plenty. In any and all circumstances I have learned the secret of being well-fed and
of going hungry, of having plenty and of being in need.” Saint Paul, in Philippians 4:11-12.

“Consider the lilies of the field, how they grow; they neither toil nor spin, yet I tell you, even
Solomon in all his glory was not clothed like one of these.” Jesus, in Matthew 6:28-30.

“The really fine things of life are not things at all.” Anonymous, Springfield, Illinois, USA, 1948.

Nevertheless, …

“Life has three rules: Paradox, Humor, and Change.” Dan Millman, in Way of the Peaceful Warrior.

“We may like to think of ourselves as civilized, but that civilization is in large part bestowed by
material wealth. Without this stuff, we would quickly be confronted by the same basic struggle for
survival that animals are faced with… Materials are a reflection of who we are, a multi-scale expression
of our human need and desires.” Mark Andrew Miodownik, MBE, in Stuff Matters: Exploring the
Marvelous Materials That Shape Our Man-Made World.

But…

“I eat too much. I drink too much. A greedy, selfish such-n-such. But when I wrap my turban on.
My mind is clear. I'm 'Baba Lon'.” Lon Milo DuQuette, in Ask Baba Lon: Answers to Questions of Life and
Magick.

… and finally, there is some seriously complicated matter ahead…

“Do you have the patience to wait until your mud settles and the water is clear?” Lao Tzu, in Tao
Te Ching.

Ora et labora…

i
ii
PREFACE

This thesis has been submitted to the Norwegian University of Science and Technology (NTNU)
for partial fulfilment of the requirements for the degree of Philosophiæ Doctor (Ph.D.). The research
has been carried out at the Department of Mechanical and Industrial Engineering (MTP), known as the
Department of Engineering Design and Materials (IPM) at the time of commencing the Ph.D. research
work, under supervision of Prof. Andreas Thorsten Echtermeyer and co-supervisor Assoc. Prof. Nils-
Petter Vedvik over a period from August 2016 to April 2019. Some work has been conducted by the
author at Rudolfs Cimdins Riga Biomaterials Innovation and Development Centre (RBIAC) and the
University of Latvia in Riga.
This work is part of the Joint Industrial Project (JIP) “Affordable Composites”. The JIP involves
DNV-GL, NTNU and nineteen companies in the industry. The work was financed by the Research
Council of Norway (Project 245606/E30 in the Petromaks 2 programme), with the research partners
Norwegian University of Science and Technology (NTNU) and DNV-GL.

iii
iv
ABSTRACT

Fiber-reinforced polymer (FRP) composites have seen a rapid rise in use in the past 50 years due
to their high strength, stiffness, relatively light weight and relatively high corrosion resistance,
especially when compared with more traditional structural materials such as steel and aluminum.
Composite materials are widely used in structural applications in marine, offshore and oil & gas
industries. A typical design lifetime of offshore FRP structures is 25 or more years in direct contact with
water leading to some deterioration of the material properties. Knowing and forecasting the extent of
the material property deterioration in water is of great interest for designers and users of the offshore
FRP structures. The environmental durability becomes a limiting factor in the use of composites for
structural applications, since the superior material properties are compromised by the uncertainty of
the material interaction with the environment.
Each of the composite constituents (matrix, fibers and sizing-rich interphase) is affected differently
by interaction with water molecules. It is therefore of high importance to understand the degradation
severity and mechanisms of each of the constituent material due to aging in water. The materials
studied are amine-cured epoxy matrix, R-glass fibers and the epoxysilane-based sizing-rich composite
interphase, which, combined, constitute the fiber-reinforced composite laminate. Degradation
mechanisms for each constituent material were identified, and novel tools, i.e. models and methods,
were developed for prediction of long-term properties of composite materials and its constituents.
These practical tools are to be of assistance in partially substituting the rigorous physical testing
procedures in the state-of-the-art situation. Prediction of long-term properties of composites should
significantly reduce costs associated with extensive testing and should already allow a partial transition
towards the multiscale modeling approach.

Brief description of the results

A spectroscopic method for true water content determination and monitoring in polymers and
composites was developed.
Aging mechanisms and changes in the chemical structure of the amine-based epoxy were
investigated. No chain scission (hydrolysis or oxidation-induced) was present, whilst thermo-oxidation
and leaching occurred. Four unique reactive sites responsible for thermo-oxidation were found. The
only mechanism that was important for the strength reduction of the studied polymer was swelling.
Mechanical properties of the epoxy could be regained upon redrying the material to the initial water
content. An analytical method for predicting anisotropic swelling in composites from the swelling of
the epoxy was developed and explained.
Aging of glass fibers occurred in two distinct phases: a short-term non-steady-state (Phase I) and
long-term steady-state (Phase II). Phase I was very complex and involved many competing processes
in parallel, such as ion exchange, gel formation and dissolution. Phase II was dissolution-dominated.
A chemical kinetic model termed the Dissolving Cylinder Zero-Order Kinetic (DCZOK) model was
developed for predicting mass loss and a decreasing radius of glass fibers at various environmental
conditions.
The sizing-rich composite interphase was degrading due to hydrolysis, resulting in the formation
of the interphase flaws. These flaws could further develop into fiber/matrix debondings, matrix cracks
and splitting along the fibers. The internal volume created by the flaws and cracks could then be filled
with water leading to the observed mass increase of a typical composite.
All known environmental aging mechanisms of the studied composite constituents were
systematized and a phenomenologically complete mass balance was presented. Using the mass
balance, it was possible to deduce the dissolution kinetics of the sizing rich-composite interphase.

v
vi
ACKNOWLEDGEMENTS

Prof. Andreas Thorsten Echtermeyer and Assoc. Prof. Nils-Petter Vedvik, for giving me the
opportunity to enter into the Ph.D. programme, gain this enormous knowledge and greatly expand
expertise in the field of materials science over the last 3 years. I am thankful to prof. Echtermeyer for
his guidance, discussions and advices.

Research Council of Norway and DNV-GL, for project financial support. Dr. Ramin Moslemian,
for his feedback and interest in my work.

Prof. Māris Kļaviņš, Prof. Filippo Berto and Doc. Ilo Dreyer, for their support. Prof. Alexei
Vinogradov for an opportunity to try myself as a teaching assistant in Fatigue Design, as well as for his
continuous support and wise advices. I am grateful to Dr. Kristīne Veģere for a (successful!)
opportunity to co-supervise a thesis for the first time. Thanks to Prof. Niklas Hedin, who gave me an
advice that helped me to stay focused and in time.

Kari Elise Dahle, Qazi Sohail Ahmad, Børge Holen, Carl-Magnus Midtbø, Agnes Digranes, Andrea
Steinert, Kristin Wikstrøm and Syverin Lierhagen, for making my work life much easier. A special
mention goes to Natalia Trotsenko who has helped me out so many times. You have saved my work
from deadline problems and subsequently saved more than a few of my nerves! Thank you very much,
Natasha!

I am grateful to the doctoral & post-doctoral fellows at NTNU, especially to my friends Dr. Abedin
Ilirjan Gagani, Dr. Anton Akulichev, Dr. Søren Heinze, Dr. Aslan Ahadi, and (soon Dr.) André Boehme,
with whom I had great fortune to spend exciting time both in and out of the university. Willingly (or
maybe unwillingly!) you have taught me lessons and reminded me what matters the most – the Family.
Without you, my friends, the outcome of this project and my worldview would surely not be the same.

I would like to thank my dear friend and mathematical mastermind Dr. Kamalakshya Mahatab,
whom I met during the international coffee (and cake!) for researchers. You have been one of the closest
friends to me during these years and we will surely stay such, wherever our paths take us.

I am grateful to my old friends Roman Andreev (for 24 years now!), Rihards Kolmanis (for about
20 years now!) and Artur Yanichev (getting there!). Thank you for being there for me. You never forget.
I am happy that we stay in touch after all these years. Now… Pierre Luigi Munier, or Piotr Mel’nik as
some may know him, I am happy that we have built and kept this friendship bond alive since we first
met in Stockholm under the influence of Rock! You are my dear friend and I value our friendship
greatly, wherever we are, mon ami. I am happy I managed to visit you and your girl Èva Hatzigeorgiou
during my Ph.D. By the way, Pierre, thanks for Stuff Matters! See, I even quoted it!

I am grateful to Yury Kovalevsky, who has taught me a lot of the “real-life-down-to-earth”

mechanical and installation work. You have also strengthened my belief that science and spirituality,
or faith, do not have to be in a mutual conflict, and in fact, perfectly add to each other.

I am thankful to Richard “Doc” Nagy, the Datamancer, for creating the keyboard that made my
work process many-many times more pleasant and rewarding.

Dr. Edward F. Edinger, Don Miguel Ruiz, Don Jose Ruiz, Lon Milo DuQuette, Manly Palmer Hall,
Lao Tzu, Saint John of the Cross, Seneca, Alejandro Jodorowsky, Mark Manson, Prof. Morton Smith,
Rev. Adam Hamilton, Prof. Richard G. Swinburne, Prof. Alan Watts, Aleister Crowley, Dr. Bradley

vii
Jersak, Abstractor of the Quintessence Dr. François Rabelais, John D. Hunt and the Apostles for
motivating me and strengthening my worldview of things both seen and unseen.

I would like to thank musicians and producers in Trondheim, with whom I was happy to play and
make music together. POTATUS and the Imp Slappers! Prof. Brian Arthur Grimes, Val, Adriaen
Verheyleweghen, Espen Samseth Hansen, Adrian Denstad Skavlan, Andreas Robertstad and Bjørn Inge
Hindenes, thank you guys! The music & songwriting, the bass & guitar, and the poetry, helped me a
lot, took my mind off work and kept me creative and motivated.

I would like to acknowledge musicians that inspire me. Dir En Grey, Alice in Chains, Black Sabbath,
Otep, The Wailers, Faith No More, Tool, Korn, King’s X, the Gazette, White Zombie, Opeth, Death, Ария,
Children of Bodom, Mastodon, Devin Townsend, Arch Enemy, Melvins, [AMATORY], Parkway Drive, A Day
To Remember, Killing Joke, Егор Летов, Luna Sea, Dream Theater, X Japan, Less Than Jake, Primus, Vistlip,
Japan, Bring Me The Horizon, Black Rebel Motorcycle Club, NOFX, System of a Down, Anti-Flag and Bad
Brains. Led Zeppelin, years ago you started this with “the Rover”.

Jon Alvarez Justo, Abdulla Bin Afif, Yan Li, Eivind Hugaas, Abaynesh Belay Fanta, Ilya Gubins,
Olga Pitkevica, Shaoquan Wang, Sondre Østli Rokvam, Juris Burlakovs, Rūta Ozola, Airida Bekeryte,
Klas Solberg, Ida Moi, Emeric Mialon, Xin Li, Hermann Kaminsky, Inna Samoylenko, Avinash Tiwari,
Xu Lu, Ivan Bunaziv, Svetlana Davydova, Yevgeny Ramensky, Merete Hovde, Cristian Rodriguez,
Bahador Najafiazar, Carlos Valente, Alice Conte, Erik Sæter, Szymon Bernat, Ekaterina Gagarina, Nima
Razavi, Beatriz Galindo-Prieto, Madina Akan, Raj Dhara, Annika Jensen, Kaspar Lasn, Roar Munkebye,
Filippo Abbatinali, Taiji Center’s Frode Strand Karlsen, Tina Lambert and Johnny Brattland, thank you!
I am grateful for meeting you all on my journeys, and for the lessons I have learned from you.

God, thanks!

Oh, and… Thanks goes to motivational words of Halvard Støwer as I was drilling through the
wall and layers of insulation and getting the cable through: “You must be a good Ph.D.!” Thank you
for these words, Halvard! How little we need to be inspired sometimes…

***

Family. Finally, this deserves much more than a separate paragraph and way more than a separate
tome. You have been the closest and strongest support, through all periods of my life. These years were
hard for us – my grandpa Rūdolfs passed away and joined my grandma Tatiana in the eternity. Both
of you always cared and gave me the strength, and you will always stay in my heart. I pray for you.
You have taught me a lot. You keep inspiring me and reminding me, that along my Russian roots, there
is always a place for my Latvian Courlandic heritage. Sacred memories for my great grandma Anna,
great grandpa Ivan and grandpa Vladimir. You have always inspired me with your loving hearts.

My aunt Lyudmila, uncle Andrey and godfather Mikhail Litevsky, I am grateful for having you in
my life. My dearest and only brother Ilya Korneckis, we have been through so much together, that I
cannot put everything what I want into words... yet, I will say, you are one of the most important people
in my life and you will always be. Love you, bro. Thank you for being the broest bro you are! Looking
forward to jam with you in the near future!

My mom Inna, dad Eduard and grandma Valentina. You have always been there for me in the
toughest and the darkest of hours. Your support and love are invaluable, words cannot describe how
grateful I am to have you in my life. I love you endlessly.

viii
 I am infinitely grateful to my beloved bride, my soulmate, a talented chemist and an amazing
graphic artist and painter Oksana Vladimirovna Golubova for her endless love, support, faith and
angelic patience during these years. You mean the world to me, I love you and I will always love you
endlessly. My thesis is also your thesis, since we are, and will always be, Two as One.

Люблю тебя бескрайне и навсегда. Навечно твой Андрей.

By Oksana Vladimirovna Golubova

Andrey Eduardovich Krauklis

Trondheim, Norway
July 2019

ix
x
CONTENTS

Preface ..................................................................................................................................................................iii
Abstract ................................................................................................................................................................. v
Acknowledgements .......................................................................................................................................... vii
Contents ............................................................................................................................................................... xi
Introduction .......................................................................................................................................................... 1
1.1. BACKGROUND ...................................................................................................................................................................................... 1
1.2. MOTIVATION ........................................................................................................................................................................................ 7
1.3. OBJECTIVES AND SCOPE ....................................................................................................................................................................... 7
1.4. THESIS STRUCTURE............................................................................................................................................................................... 8

Methodology ...................................................................................................................................................... 13
2.1. MATERIALS ......................................................................................................................................................................................... 13
2.2. EXPERIMENTAL METHODS ................................................................................................................................................................. 18
2.2.1. Material Characterization and Testing .................................................................................................................................. 18
2.3. MODELING ......................................................................................................................................................................................... 26
2.3.1. Numerical Modeling (FEA) .................................................................................................................................................... 26
2.3.2. Analytical Modeling ................................................................................................................................................................ 26

Key Findings....................................................................................................................................................... 27
3.1. TRUE WATER CONTENT DETERMINATION AND MONITORING IN POLYMERS AND COMPOSITES (PAPER I) .................................. 27
3.2. CHANGES IN THE CHEMICAL STRUCTURE OF THE MATRIX POLYMER DUE TO HYGROTHERMAL AGING (PAPER II)..................... 28
3.3. REVERSIBILITY OF THE MECHANICAL PROPERTIES OF THE MATRIX POLYMER UPON RE-DRYING (PAPER III) ............................... 31
3.4. PREDICTION OF HYGROSCOPIC SWELLING OF THE COMPOSITE FROM SWELLING OF THE MATRIX POLYMER (PAPER IV) ............ 31
3.5. PREDICTION OF MASS LOSS AND RADIUS REDUCTION OF GLASS FIBERS DUE TO DISSOLUTION IN WATER (PAPER V) ................ 34
3.6. PREDICTION OF MASS LOSS OF GLASS FIBERS INSIDE THE COMPOSITE DUE TO HYGROTHERMAL AGING (PAPER VI).................. 37
3.7. PREDICTION OF GLASS FIBER DISSOLUTION AT VARIOUS ENVIRONMENTAL CONDITIONS (PAPER VII) ....................................... 39
3.8. LONG-TERM AGING OF THE SIZING-RICH COMPOSITE INTERPHASE (PAPER VIII) ......................................................................... 41

Conclusions ........................................................................................................................................................ 47
Future Work ....................................................................................................................................................... 49
Abbreviations & Symbols ................................................................................................................................. 53
Bibliography ....................................................................................................................................................... 57
Appendices ......................................................................................................................................................... 65
PAPER I ...................................................................................................................................................................................................... 67
PAPER II..................................................................................................................................................................................................... 83
PAPER III ................................................................................................................................................................................................. 101
PAPER IV ................................................................................................................................................................................................. 111
PAPER V .................................................................................................................................................................................................. 127
PAPER VI ................................................................................................................................................................................................. 141
PAPER VII ................................................................................................................................................................................................ 151
PAPER VIII .............................................................................................................................................................................................. 171

xi
xii
CHAPTER 1

INTRODUCTION

Parts of the Ph.D. thesis are copied from the appended Papers I-VIII (Appendices A-H).

1.1. BACKGROUND

Composites & environmental durability

Fiber-reinforced polymer (FRP) composites have seen a rapid rise in use in the past 50 years due
to their high strength, stiffness, relatively light weight and high corrosion resistance, especially when
compared with more traditional structural materials such as steel and aluminum [1]. The reason for
such superior performance is the synergistic interaction between the constituent materials inside the
composite [1]. There are three constituents in an FRP, namely a polymer matrix, a fibrous reinforcement
and a multi-component coating on the surface of the fibers – the sizing. During the manufacture of
FRPs, the sizing results in the formation of a sizing-rich composite interphase between the reinforcing
fibers and the matrix polymer, chemically and physically bonding fibers and the polymer together [2].
Epoxy, glass fibers and epoxysilane/epoxy sizing are the constituents studied in this work.
FRP laminates are used for structural applications in marine, offshore and oil & gas industries due
to their light weight and corrosion resistance [3–5]. Composites offshore have been implemented in
such applications as risers, tethers, repair patches and ship hulls [6–11]. In these applications, FRPs get
continuously exposed to water and humid environments for decades, a typical design lifetime being
around 25 years or more [7].
Exposure of composite to aqueous and humid environments leads to aging, negatively impacting
the mechanical properties of FRPs [12–18]. Understanding and quantifying the water-induced aging is
especially important for glass fiber-reinforced composites since the glass fibers are hygroscopic [19].
The environmental durability is one of the limiting factors in the structural applications [20], since the
superior strength and stiffness of composite materials are often compromised by the uncertainty of the
material interaction with the environment [21]. Durability is a primary issue because environmental
factors such as moisture, temperature and stress to which the material is exposed can degrade the
properties of the constituent phases. Environmental aging is especially important at higher
temperatures, since the aging is then accelerated. Furthermore, constituents are affected by
environmental aging differently from each other [22]. Thus, understanding the mechanisms and
kinetics of environmental aging of individual constituents is of high importance for the composite
environmental durability. Since water uptake in composites is governed by diffusion [23] and water
concentration has a crucial role in environmental aging [24], of importance is also the development of
a method that allows determination of the true water content inside the FRPs.

Matrix polymer & environmental aging

Epoxy polymers are a common thermoset matrix material known for their relatively high strength,
stiffness, low volatility, chemical resistance, and low shrinkage on curing [25–27]. Epoxies are two-
component systems consisting of an epoxy compound and a hardener (also known as a curing agent),
which react to form the cured polymer matrix network [28]. The epoxy family covers a large diversity
of polymer networks defined by the type of the epoxy compound and hardener employed. All these
variations lead to a different polymer matrix formed, and affect its physical and chemical properties

1
[28,29]. DGEBA-based epoxies, such as in this work, are the most widely used epoxy polymers in
structural applications and constitute more than 75% of epoxy resins sold worldwide [30]. In this work,
the epoxy material system contains two epoxy compounds (DGEBA and HDDGE) and two hardeners
(IPDA and POPA). After the curing reaction, a three-dimensional amine epoxy matrix is obtained [31].
When the epoxy matrix is exposed to aqueous or humid environments, water molecules can
migrate into the polymer and may affect its properties [5,25,32,33]. Highly crosslinked amine-cured
epoxy polymers are hydrophilic and their mechanical properties can significantly deteriorate upon
water uptake [26,34,35]. It is well established that water uptake is an important factor in performance
and durability of epoxy-based composites which undergo plasticization and swelling stresses [36].
Hygrothermal process may induce both reversible and irreversible changes in the epoxy [20,37].
Irreversible changes persist even after redrying the material [32]. Irreversible damage in epoxies may
occur due to susceptibility of the polymer to hydrolysis/chain-scission, residual curing, leaching and
oxidation. For some epoxy systems, water at elevated temperatures can attack the crosslinked network,
causing chain scission and leaching [4]. For the studied epoxy, residual curing and hydrolysis/chain
scission did not occur [3,38], while leaching and oxidation did, but did not exhibit any significant effect
on the mechanical properties of the studied epoxy [38].
The main effect of water on the mechanical property deterioration of the epoxy polymer was
hygroscopic swelling [38]. Swelling is a specific response accompanying moisture diffusion in polymers
and polymer-based composites [4]. Susceptibility of polymers to swelling results in a two-fold effect on
FRPs: on the one hand, it causes a decrease in mechanical strength of the polymeric matrix [38], while
on the other, it results in swelling stresses when the hygroscopic swelling is restrained [39]. The
degradation of the tensile strength can be attributed to the plasticization/swelling and deterioration of
the polymer (ultimate tensile strength of some epoxies may decrease even by 40% due to hygroscopic
swelling) [20,33,40]. While some authors report significant fatigue life reduction of epoxies due to
swelling, others do not observe noticeable changes after water absorption [32]. For the studied epoxy,
both static and fatigue strength dropped by about 20% due to hygroscopic swelling [38].
More details and key findings on environmental aging of the epoxy, including mechanisms and
deterioration of the mechanical properties can be found in Chapter 3 and in Papers I-IV (Appendices
A-D).

Fiber reinforcement & environmental aging

Glass fibers (GFs) are often used as reinforcement in structural composite materials. Furthermore,
the most common fibrous reinforcement material is glass [41]. GFs are produced from raw materials,
which are virtually unlimited in supply [41] and possess such desirable reinforcement material
properties as high hardness, strength and stiffness [41,42]. Various types of GFs exist such as E, ECR, R
and S-glass, listed in the order of increasing mechanical strength [43]. Another common reinforcement
material, carbon fibers, is known to be inert in water, while the glass fibers degrade strongly in aqueous
environments [44,45]. Thus, concerning environmental aging, it is more relevant to expand knowledge
base on the water-induced degradation of glass fibers.
The fact that even water may corrode glass fibers has been known for many years [45]. Glass fibers
are hydrophilic and are susceptible to degradation when exposed to water environments [46].
Hydrolytic degradation of glass fibers can significantly reduce mechanical strength and leads to
corrosion-induced defects [44]. The degradation of glass fibers due to environmental attack can severely
affect the performance of GFRPs [47]. Humid and water environments act primarily to reduce the fiber
strength caused by the growth of flaws, such as surface cracks [45,47]. Unprotected glass fibers exposed
to water lose their strength relatively quickly, especially so if they are mechanically loaded [48,49].
The long-term environmental degradation of glass materials has been previously studied mainly
with respect to nuclear waste applications [50,51]. Most of the existing works on glass dissolution have
been performed with bulk silicate glass, and fibers are not studied often [52]. Recently, there has been
an increasing interest in environmental aging of FRPs, where R-glass and E-glass are often used as

2
reinforcement [3,24,52-55]. Even so, very few studies exist on the kinetics of GF dissolution (mostly on
E-glass) [56,57]. Most studies on environmental degradation of composites are concerned directly with
deterioration of the mechanical properties, and the mechanistic origin and the kinetics of chemical
degradation tend to be overlooked [54,55]. However, a few studies exist that explain general
mechanisms of environmental degradation of glass materials using various approaches that are based
on surface reactions, chemical affinity and diffusion [46,58-62]. Yet again, dissolution experiments in
existing studies are mainly performed with bulk silicate glasses, and glass fibers are not studied often
[52].
However, a few studies on the kinetics of glass fiber dissolution exist. For instance, Mišíková et al
have studied the E-glass fiber leaching kinetics in distilled water at different temperatures [56]. Bashir
et al studied the kinetics of the dissolution of E-glass fibers in alkaline solutions by immersing single
fibers and measuring the diameter change [57]. Recently, Krauklis and Echtermeyer presented an
analytical model termed Dissolving Cylinder Zero-Order Kinetic (DCZOK) that predicts glass
dissolution kinetics during long-term hygrothermal aging of glass fiber bundles and fiber-reinforced
composites at various environmental conditions [24,52,43]. The model is able to predict the mass loss
and fiber radius reduction kinetics during dissolution [24,52]. The dissolution of glass fibers inside
composites is slowed down compared to glass fiber bundles and is addressed in the analytical model
[24].
In the short-term (Phase I, up to about a week in contact with water [43,52]), hydrolytic
degradation involves such competing processes as ion exchange, gel formation and dissolution
[50,52,63]. In the long-term (Phase II), hydrolytic degradation is governed by the glass dissolution
mechanism and follows zero-order reaction kinetics [50,52]. Such kinetics depend on the glass surface
area in contact with water, which is proportional to the fiber radius. As the dissolution continues, the
radius decreases resulting in the mass loss deceleration [52]. Elements that are released during
degradation of R-glass are Na, K, Ca, Mg, Fe, Al, Si and Cl [52]. The total mass of all ions released is the
cumulative mass loss [52]. The total mass loss is what manifests in the radius reduction [52].
During glass-water interaction, several chemical reactions may occur, shown in Chemical
Reactions (a)-(k), after [43,50,52,63-66]:

ሺ‫ ݅ܵ ؠ‬െ ܱܰܽሻ ൅ ‫ ܱ ʹܪ‬՜ ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ܱ‫ܪ‬െ ൅ ܰܽ൅ ሺܽሻ

ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܭ‬ሻ ൅ ‫ ܱ ʹܪ‬՜ ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ܱ‫ܪ‬െ ൅ ‫ܭ‬൅ ሺܾሻ

ሺ‫ ݅ܵ ؠ‬െ ܱሻʹ ‫ ܽܥ‬൅ ‫ ܱ ʹܪ‬՜ ʹሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ʹܱ‫ܪ‬െ ൅ ‫ʹܽܥ‬൅ ሺܿሻ

ሺ‫ ݅ܵ ؠ‬െ ܱሻʹ ‫ ݃ܯ‬൅ ‫ ܱ ʹܪ‬՜ ʹሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ʹܱ‫ܪ‬െ ൅ ‫ʹ݃ܯ‬൅ ሺ݀ሻ

ሺ‫ ݅ܵ ؠ‬െ ܱ െ ‫ ݈ܣ‬ൌሻ ൅ ‫ ܱ ʹܪ‬՞ ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ሺൌ ‫ ݈ܣ‬െ ܱ‫ܪ‬ሻሺ݁ሻ

ሺ‫ ݅ܵ ؠ‬െ ܱሻʹ ‫ ݁ܨ‬൅ ‫ ܱ ʹܪ‬՜ ʹሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ʹܱ‫ܪ‬െ ൅ ‫ʹ݁ܨ‬൅ ሺ݂ሻ

ሺ‫ ݅ܵ ؠ‬െ ܱሻ͵ ‫ ݁ܨ‬൅ ‫ ܱ ʹܪ‬՜ ͵ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ͵ܱ‫ܪ‬െ ൅ ‫͵݁ܨ‬൅ ሺ݃ሻ

ሺ‫ ݅ܵ ؠ‬െ ܱ െ ܵ݅ ‫ؠ‬ሻ ൅ ܱ‫ܪ‬െ ՞ ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ሺ‫ ݅ܵ ؠ‬െ ܱെ ሻሺ݄ሻ

ሺ‫ ݅ܵ ؠ‬െ ܱെ ሻ ൅ ‫ ܱ ʹܪ‬՞ ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ܱ‫ܪ‬െ ሺ݅ሻ

ܱܵ݅ଶ ൅ ʹ‫ܪ‬ଶ ܱ ՞ ‫ܪ‬ସ ܱܵ݅ସ ሺ݆ሻ

ுమ ை
‫݈ܥ݁ܯ‬௫ ሱۛሮ ሺ‫ ݁ܯ‬௫ା ሻ ൅ ‫ ି ݈ܥݔ‬ሺ݇ሻ

3
 Chemical reaction (j) can also be written as a combination of subsequent reactions (l) and (m),
meaning that initially ‫ܪ‬ଶ ܱܵ݅ଷ is formed, which dissociates weakly and further reacts with water to form
silicic acid:

ܱܵ݅ଶ ൅ ‫ܪ‬ଶ ܱ ՞ ‫ܪ‬ଶ ܱܵ݅ଷ ሺ݈ሻ

‫ ͵ܱ݅ܵ ʹܪ‬൅ ‫ ܱ ʹܪ‬՞ ‫ܪ‬Ͷ ܱܵ݅Ͷ ሺ݉ሻ

As shown in the Chemical reactions (a)-(k), various competing reactions happen simultaneously.
Initially these reactions happen at independent rates (short-term non-steady-state), later one process
becomes limiting and dominates the behaviour (long-term steady-state). Therefore, the degradation
process should be divided into two respective stages as was described earlier.
More details and key findings on environmental aging of glass fibers, including analytical models
of chemical kinetics, can be found in Chapter 3 and in Papers V-VII (Appendices E-G).

Sizing-rich composite interphase & environmental aging

The composite interphase is of vital importance since the mechanical properties of composite
materials are often determined by whether the mechanical stresses can be efficiently transferred from
the matrix to the reinforcing fibers [19,67,68]. The quality of the interfacial interaction is strongly
dependent on the adhesional contact and the presence of flaws in the interphase [69]. It is generally
agreed that the composite interphase is often the mechanical weak link and a potential source for the
initiation of defects in fiber-reinforced composite structures [19].
Water and humid environments negatively impact the mechanical properties of FRPs partially
because of a loss of the interfacial bonding [11-14,19]. Flaws in the interphase can be introduced due to
the interaction with water uptaken from the environment [69]. The removal of the sizing material can
also lead to a microcrack initiation at the surface of glass fibers, and that various sizing components can
be extracted by water, resulting in the loss of the material [70-74]. It is therefore of high importance to
understand the environmental aging mechanism and kinetics of a sizing-rich composite interphase.
What complicates the situation is that the sizing, which forms the interphase, has a proprietary
composition. Commercial glass fibers are often supplied with only one or two sizing-related details.
The first being an indication of the chemical compatibility of the sizing with the matrix polymer, e.g.
epoxy, as in this case. The second is a value for the loss on ignition (LOI), which indicates the amount
of sizing [75]. The key functions of the sizing are (1) to protect the glass fibers during handling and
production, (2) to ensure a high level of stress transfer capability across the fiber-matrix interphase and
(3) to protect the composite matrix interphase against environmental degradation [11].
It is known that the W2020 sizing consists of about five various chemicals [76,77]. Furthermore, it
is known, that the sizing contains an organofunctional silane commonly referred to as a coupling agent
[78-80]. This class of chemicals can be considered the most important in the glass fiber sizing, as it is the
main component that promotes adhesion and stress-transfer between the polymer matrix and the fiber
[11].
The structure of the sizing-rich composite interphase is very complex [11], as the sizing itself is
heterogeneous and not uniform [11,81]. Furthermore, it has been observed by various researchers, that
sizing is coated on fibers in “islands”, “islets” or in patches, meaning that fiber surface is only partially
covered by the sizing, also giving some roughness to the surface [11,82-87]. Mai et al investigated silane-
based sizings using AFM and concluded that sized fibers are rougher than the desized fibers [87]. Also,
similar conclusions were drawn by a few other researchers [69,81]. Similar results have been observed
in this work, using Brunauer-Emmett-Teller (BET) theory for specific surface area determination.
The composite interphase can be visualized as a matrix polymer/poly(siloxane)/glass fiber model
(shown in Figure 1.1) [19].

4
 Figure 1.1. The concept of a polymer-siloxane-glass interphase, after [19]. The dotted line indicates that
the sizing is rough [69,81,87].

The siloxanes and poly(siloxanes) form covalent bonds with the glass fiber surface, resulting in a
two-dimensional interface, the thickness of which is governed by the length of the chemical bonds, and
is of an ångstrøm-scale (one tenth of a nanometer) [19]. A composite interphase is a gradient-type blend
of the sizing compounds and the bulk matrix polymer, usually being about a micrometer in thickness
[11,19,32,88,89]. It was observed, that an interfacial failure occurs at 0.5 – 4 nm from the glass surface in
glass/epoxysilane/epoxy interphase, indicating that the interphase region, rather than the two-
dimensional interface is the weak link [19].
There are no direct measurement methods to study the environmental aging of the composite
interphase. It is not known yet how to quantify the interphase loss in the composite due to aging.
Furthermore, it has been noted by Riaño et al that modeling techniques to study the composite
interphase are becoming of high interest to the scientific community and industry [90]. Recently,
Krauklis, Gagani and Echtermeyer proposed a phenomenological mass balance approach for the
hygrothermal aging of fiber-reinforced composites and systematized and quantified known aging
mechanisms of composite microconstituents (Paper VIII) [22]. Mass balance allowed deducing the
kinetics of the hydrolytic degradation of the sizing-rich composite interphase [22].
More details and key findings on environmental aging of the composite interphase and chemical
kinetics of its dissolution can be found in Chapter 3 and in Paper VIII (Appendix H).

True water content of the matrix and the composite

Composites uptake water from their surroundings. Water uptake is governed by water diffusion
and depends on the thickness of the composite structure and the temperature of the environment
[23,91-95]. More details on prediction of water uptake and diffusion for the same GFRP material as in
this work can be found in papers by Gagani et al, for various composite structures both with and
without voids [23].

5
 For polymers, experimental techniques such as differential scanning calorimetry (DSC), ultraviolet
(UV) reflection spectroscopy, attenuated total reflection Fourier transform infrared spectroscopy (ATR-
FTIR) and Fourier transform near-infrared spectroscopy (FT-NIR) have been reported [35,96-101]. It has
also been reported that NIR spectroscopy, e.g. FT-NIR, is a promising technique for the water content
monitoring for various materials, in part due to recent improvements of NIR spectrometers [102-104].
NIR spectroscopy has long been used in such fields as medicine, food and polymers [102,105,106].
The only widely known method for determination of water content in composites is a gravimetric
method, which provides information of weight gain upon water uptake, or the weight loss upon drying
[53]. However, gravimetric curves for composites are not straightforward to interpret, since alongside
typical Fickian water uptake there are also aging processes occurring inside the composite [22]. Another
significant drawback of the gravimetric method is the necessity to know the mass of the absolutely dry
material, which in some cases requires extremely long drying times. Furthermore, it is not possible to
obtain such data precisely due to humidity of the ambient air, unless perfectly dry conditions can be
ensured during drying and weight gain measurements. In order to understand water-induced aging
on the mechanistic microlevel, the absolute or true water content has to be known. Otherwise, an error
is introduced when linking property deterioration due to water influence and the concentration of
water inside the material. Therefore, an alternative NIR spectroscopic method for measuring and
monitoring true water content in epoxy polymers and fiber-reinforced composites was developed [53].
More details on the method can be found in Chapter 3 and in Paper I (Appendix A).

Hygroscopic swelling of the matrix and the composite

Since for the studied epoxy matrix, only mechanism of swelling is important for the strength
reduction [38], it is important to quantitively link hygroscopic swelling with the true water content.
Unlike the polymer itself, in FRPs, the matrix is constricted by fibers, and as a result, this affects
the swelling behavior. What complicates the phenomenon even more is the orthotropic nature of
swelling of composites—fibers, such as glass or carbon, do not swell, while the polymer matrix does
[5,39]. Such incompatible swelling behavior in FRPs leads to swelling stresses at the interfaces, which
may lead to microcrack formation, especially under transient conditions (non-uniform moisture content
distribution) [5,107]. Hygroscopic swelling may affect the mechanical properties of FRPs significantly
[4,38,53,108]. Thus, it is important to know not only the true water content, but also the swelling
behavior, in order to properly characterize the FRP material property change, i.e., strength or modulus,
resulting from moisture absorption [109]. Linear strain increases linearly with increasing water
concentration for both composites and polymers [108,110]. For orthotropic laminates, three Coefficients
of Hygroscopic Expansion (CHEs; ߚ௫ ǡ ߚ௬ ǡ ߚ௭ ) are needed in order to predict swelling [109].
Quantification of the orthotropic CHEs can be performed experimentally using samples with
different fiber orientations. However, it is a time-consuming and tedious process that also tends to
involve quite high experimental scatters. The industrial interest lies in the reduction of testing time and
testing-related expenses. Thus, a modeling approach to swelling of FRPs due to the effects of water
(and also other liquids, such as oil) is of interest [111].
Various studies have been performed on swelling of FRPs [21,39,108,112-119] and, more recently,
on hygroscopic swelling in textile composites [120,121]. The works available in the literature have
addressed several aspects of hygroscopic swelling in composites, from the nature of swelling in
polymeric matrix [5,13,39,122], to the influence of swelling on the fluid diffusion in polymers
[108,111,117,123,124], to the development of micromechanical models to predict transverse swelling
[112,120,121].
An interesting opportunity is the possibility to predict the orthotropic swelling constants (CHEs)
of the composite from the CHE of the matrix polymer, which is isotropic. The matrix properties are
easy to measure. Furthermore, they also may be found in literature for various polymers [122,125].
However, in some cases, the interfacial effects may not be negligible [126], many of the moisture-related
properties of composites are known to be traceable to those of the matrix material [53,108]. Swelling

6
strains of a composite and a matrix polymer should also be related to each other through a proper
analysis [108,127]. Since swelling in polymers does not follow the ideal mixing law [4], i.e., the volume
increase of the polymer is not equal to the volume of the absorbed water, and it is necessary to perform
swelling experiments for the matrix polymer itself, or to find polymer CHE in the literature [122,125].
Recently, Krauklis et al have shown that orthotropic hygroscopic swelling of fiber-reinforced
composites can be analytically predicted from the swelling of the matrix polymer using a model based
on linear elasticity [109]. More details on the model can be found in Chapter 3 and in Paper IV
(Appendix D).

1.2. MOTIVATION

The current situation shows that there is a lack of efficient tools for reliably predicting changes in
properties of composite materials and their constituents due to environmental aging, as well as the
processes involved are not fully understood, i.e. the degradation processes at the microconstituent level
taking place and their respective mechanisms [128]. Thus, an interaction of composite constituents with
the environment was investigated, and novel tools were developed in order to predict environmental
aging.

1.3. OBJECTIVES AND SCOPE

The aim of this work is to identify the environmental aging mechanisms of composite
microconstituent materials, to study the severity of such aging on the mechanical properties, and to
develop novel tools for the prediction of changes in composite properties due to environmental aging.
The microconstituents of the composites were studied in order to evaluate the water-induced
degradation. This resulted in the development of the tools, i.e. analytical models and methods, for
prediction of microconstituent and composite properties in offshore applications. The developed tools
allow an efficient alternative to the state-of-the-art procedures based on physical testing, which is much
more time-consuming than the use of analytical models. Prediction of long-term properties of
composites should significantly reduce costs associated with extensive testing and should already
allow a partial transition towards the multiscale modeling approach.

The reached goals and contribution to the state-of-the-art are the following:
x Providing a true water content determination method for polymers and composites.
x Providing a better understanding of environmental aging mechanisms in fiber-reinforced
composites and microconstituents and its effect on mechanical properties of such materials.
x Providing practical tools, i.e. models and methods, for quantitative prediction of both short-
term and long-term water-induced changes in the microconstituent materials and composites.
The tools are to be of assistance in substituting the rigorous physical testing procedures in the
state-of-the-art situation.

In order to achieve these goals, the following objectives were fulfilled:

x To obtain the water saturation levels and to develop a method for determination and
monitoring of the true water content in composites and microconstituents.
x To study the interaction of composite microconstituents as well as composite material itself
with the affecting environment, i.e. water. Collected data involves both the mechanical side
and the chemical/physico-chemical side of the processes involved.
x To study the material degradation when exposed to environmental aging and evaluate how
the properties of the material are affected.

7
 x Based on environmental aging experiments, to propose and develop analytical models in
order to predict changes in properties of the composite microconstituents due to exposure to
such environments.

1.4. THESIS STRUCTURE

The thesis is made as a compendium of eight scientific articles together with five additional
chapters. Chapter 1 introduces the reader to the scientific problem that is to be addressed and provides
motivation for the research described in this work, the objectives and a description of the scientific
papers. Chapter 2 gives a detailed description of the materials and methods used. Chapter 3 provides
a summary of the results. Chapter 4 outlines the main conclusions. Chapter 5 describes the
recommendations for the future work. The eight scientific articles are incorporated in the Appendices
A-H of the thesis.

Publications included in the Ph.D. thesis (Appendices A-H):

(I) Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Near-Infrared Spectroscopic Method for
Monitoring Water Content in Epoxy Resins and Fiber-Reinforced Composites. Materials (Switzerland)
2018, 11(4), 586-599. DOI: 10.3390/MA11040586.
(II) Krauklis, A.E.; Echtermeyer, A.T. Mechanism of Yellowing: Carbonyl Formation During
Hygrothermal Aging in a Common Amine Epoxy. Polymers (Switzerland) 2018, 10(9), 1017-1031. DOI:
10.3390/POLYM10091017.
(III) Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Hygrothermal Aging of Amine Epoxy:
Reversible Static and Fatigue Properties. Open Engineering (Poland) 2018, 8(1), 447-454. DOI:
10.1515/ENG-2018-0050.
(IV) Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Prediction of Orthotropic Hygroscopic
Swelling of Fiber-Reinforced Composites from Isotropic Swelling of Matrix Polymer. Journal of
Composites Science (Switzerland) 2019, 3(1), 10-23. DOI: 10.3390/JCS3010010.
(V) Krauklis, A.E.; Echtermeyer, A.T. Long-Term Dissolution of Glass Fibers in Water Described
by Dissolving Cylinder Zero-Order Kinetic Model: Mass Loss and Radius Reduction. Open Chemistry
(Poland) 2018, 16(1), 1189-1199. DOI: 10.1515/CHEM-2018-0133.
(VI) Krauklis, A.E.; Echtermeyer, A.T. Dissolving Cylinder Zero-Order Kinetic Model for
Predicting Hygrothermal Aging of Glass Fiber Bundles and Fiber-Reinforced Composites. In 4th
International Glass Fiber Symposium; Gries, Th.; Pico, D.; Lüking, A.; Becker, Th., Eds.; Mainz, G: Aachen,
Germany, 2018; pp. 66–72. ISBN: 978-3-95886-249-4.
(VII) Krauklis, A.E.; Gagani, A.I.; Vegere, K.; Kalnina, I.; Klavins, M.; Echtermeyer, A.T.
Dissolution Kinetics of R-Glass Fibres: Influence of Water Acidity, Temperature and Stress Corrosion.
Fibers (Switzerland) 2019, 7(3), 22-40, in a Special Issue: Advances in Glass Fibers. DOI: 10.3390/fib7030022.
(VIII) Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Long-Term Hydrolytic Degradation of the
Sizing-Rich Composite Interphase. Coatings (Switzerland) 2019, 9(4), 263-286. DOI:
10.3390/coatings9040263.

8
 Figure 1.2. Schematic relationship of the papers and composite constituent materials in the thesis.

The relationship between the articles and the studied materials is represented by a simple diagram
in Figure 1.2. Black lines represent the relationship between the constituents and the composite material
itself. Black dotted lines signify that the interphase exists only when the polymer and matrix are
combined, thus only in the composite. Colored lines show which publications correspond to which
constituent material and the composite itself. Yellow, red, blue and green lines correspond to works on
polymer, interphase, fibers and composite, respectively.
Papers I-IV cover the polymeric matrix material (amine-cured DGEBA-based epoxy). Paper I
presents a method of true water content determination in a polymer matrix. Paper II identifies
hygrothermal aging mechanisms and changes in the chemical structure of the polymer. Paper III
identifies the influence of hygrothermal aging on the mechanical properties of the polymer and shows
that only mechanism of swelling is important for the strength reduction. It shows how severely swelling
affects static tensile strength and fatigue properties of the epoxy, and how these properties can be
regained upon redrying to the initial water content, indicating reversibility. Paper IV shows how water
uptake quantitively manifests in hygroscopic swelling of the polymer.
Papers V-VII cover the reinforcement material (R-glass fibers). Paper V identifies hygrothermal
aging mechanisms in glass. The paper describes long-term aging of R-glass fibers in water and presents
a chemical kinetic model called a Dissolving Cylinder Zero-Order Kinetic (DCZOK) model. The model
can be used for quantitative predictions of the mass loss and fiber radius reduction due to glass
dissolution in water. Paper VI shows the protective effect of sizing against the dissolution of R-glass
fibers. Paper VII describes such environmental effects as pH, temperature and stress corrosion on the
dissolution of glass fibers. The paper reports activation energies and kinetic constants for various
environmental conditions, as well as the extended DCZOK model that takes into account the
environmental effects.
Paper VIII covers the composite interphase. The paper shows how, by identifying all the aging
mechanisms and using a mass balance approach, it is possible to deduce the environmental aging of
the interphase material. The kinetics and the mechanism of hydrolytic aging are described.
Papers I, IV, VI and VIII also cover the composite material directly. Paper I presents a method of
true water content determination and monitoring in a composite material. Paper IV shows how
hygroscopic swelling can be quantitively predicted in composite from swelling of a polymer, since
fibers do not swell. Paper VI extends the use of the DCZOK model to the composites. Paper VIII

9
summarizes environmental aging mechanisms of GFRPs using a phenomenological mass balance
approach and describes long-term gravimetric (mass gain and loss) behavior of the composite.
Furthermore, it provides a link between the water uptake of a composite and the hydrolytic damage of
the interphase.

The contribution of the author and the co-authors is as follows:

A.E. Krauklis formulated the objectives and defined the test program, designed and assembled
the customized rigs in Paper VII. Produced the test specimens, developed test methods in all the Papers
(I-VIII), carried out most experiments and analytical modeling tasks, analyzed and interpreted the data
and wrote all the Papers (I-VIII).

A.I. Gagani produced some of the composite test specimens, carried out FEA simulations and
helped with analytical modeling in Paper IV, has helped with experimental work in Paper III and with
writing literature review in Paper VII.

E. Sæter produced some of the composite test specimens and helped with experimental work in
Paper VIII.

I. Kalnina helped with experimental work in Paper VII.

K. Vegere helped with organizing collaboration with Riga Technical University that allowed some
parts of the experimental testing in Papers II and VII.

M. Klavins contributed with suggestions and comments on Paper VII and helped with organizing
collaboration with Latvian University that allowed some parts of the experimental testing in Paper VII.

A.T. Echtermeyer contributed with intellectual discussions, guidance, suggestions and comments
to all papers.

Relevant publications not included in the Ph.D. thesis:

(IX) Echtermeyer, A.T.; Gagani, A.I.; Krauklis, A.E.; Mazan, T. Multiscale Modelling of
Environmental Degradation—First Steps. In Durability of Composites in a Marine Environment 2. Solid
Mechanics and Its Applications; Davies, P.; Rajapakse, Y.D.S., Eds.; Springer: Cham, Switzerland, 2018;
Volume 245, pp. 135-149. ISBN: 978-3-319-65145-3.
(X) Gagani, A.I.; Krauklis, A.E.; Echtermeyer, A.T. Anisotropic fluid diffusion in carbon fiber
reinforced composite rods: Experimental, analytical and numerical study. Marine Structures
(Netherlands) 2018, 59, 47-59. DOI: 10.1016/j.marstruc.2018.01.003.
(XI) Gagani, A.I.; Krauklis, A.E.; Echtermeyer, A.T. Orthotropic fluid diffusion in composite
marine structures. Experimental procedure, analytical and numerical modelling of plates, rods and
pipes. Composites: Part A (United Kingdom) 2018, 115, 196-205. DOI: 10.1016/j.compositesa.2018.09.026.
(XII) Gagani, A.I.; Monsås, A.B.; Krauklis, A.E.; Echtermeyer, A.T. The effect of temperature and
water immersion on the interlaminar shear fatigue of glass fiber epoxy composites using the I-beam
method. Composites Science and Technology (Netherlands) 2019, 181, 107703-107712. DOI:
10.1016/j.compscitech.2019.107703.
(XIII) Gagani, A.I.; Krauklis, A.E.; Sæter, E.; Vedvik, N.P.; Echtermeyer, A.T. A Novel Method for
Testing and Determining ILSS for Marine Composites. Composite Structures (Netherlands) 2019, 220,
431-440. DOI: 10.1016/j.compstruct.2019.04.040.

10
Scientific conferences:

(I) Echtermeyer, A.T.; Gagani, A.I.; Krauklis, A.E. Long-term degradation of composite laminates
in offshore applications described by a multi-scale approach. 36th International Conference on Ocean,
Offshore and Arctic Engineering Conference OMAE 17. Trondheim, Norway, 2017.
(II) Gagani, A.I.; Krauklis, A.E.; Echtermeyer, A.T. 3D microscale finite element anisotropic model
for predicting diffusion in GF/epoxy composites. 20th International Conference on Composite Structures
ICCS20. Paris, France, 2017.
(III) Echtermeyer, A.T.; Gagani, A.I.; Krauklis, A.E. Effect of anisotropic diffusion of water on the
long-term degradation of composite laminates inside metal end fittings. Oil & Gas Non-Metallics 2017.
London, United Kingdom, 2017.
(IV) Gagani, A.I.; Krauklis, A.E.; Echtermeyer, A.T. Fluid diffusion in fiber reinforced composites
in the presence of cracks and delamination. International Conference on Composite Structures ICCS21.
Bologna, Italy, 2018.
(V) Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Hygrothermal Aging of Fiber-Reinforced
Composites: Introduction to Phenomenological Perspective and Mass Balance Approach. International
Conference on Composite Structures ICCS21. Bologna, Italy, 2018.
(VI) Krauklis, A.E.; Echtermeyer, A.T. Dissolving Cylinder Zero-Order Kinetic Model for
Predicting Hygrothermal Aging of Glass Fiber Bundles and Fiber-Reinforced Composites. Fourth
International Glass Fiber Symposium IGS 2018. Aachen, Germany, 2018.
(VII) Echtermeyer, A.T.; Gagani, A.I.; Krauklis, A.E.; Moslemian, R. Long Term Fatigue
Degradation – Superposition of Dry and Wet Properties. Twenty-second International Conference on
Composite Materials ICCM22. Melbourne, Australia, 2019.

11
12
CHAPTER 2

METHODOLOGY

2.1. MATERIALS

Matrix polymer: amine-cured epoxy

The epoxy resin and amine hardener were supplied by Hexion. Amine-cured epoxy was prepared
by mixing reagents Epikote Resin RIMR135TM and Epikure Curing Agent RIMH137TM
stoichiometrically, in a ratio of 100:30 by weight. The mixture was degassed in a vacuum chamber for
30 minutes to remove bubbles. The density of the polymer (ߩ௠ ) was 1.1 g/cm3.
Resin and hardener system consisted of the following compounds by composition: 63 wt%
Bisphenol A diglycidyl ether (DGEBA; CAS 1675-54-3; number average molecular weight ≤ 700); 14
wt% 1,6-hexanediol diglycidyl ether (HDDGE; CAS 16096-31-4); 14 wt% poly(oxypropylene)diamine
(POPA; CAS 9046-10-0; molecular weight 230); and 9 wt% isophorondiamine (IPDA; CAS 2855-13-2).
Chemical structures of these compounds are shown in Figure 2.1.

Figure 2.1. Molecular structures of epoxy and hardener components: (A) DGEBA monomer; (B) DGEBA
oligomer (n = 1–2); (C) HDDGE; (D) POPA; (E) IPDA.

The dogbone and rectangular sample steel molds, shown in Figure 2.2, were prepared using
computer numerical control (CNC) machining for casting the epoxy into the required geometry.

13
 Figure 2.2. Steel molds for casting epoxy (left) dogbones and (right) rectangular samples.

Degassed resin was cast into the respective mold, followed by curing at room temperature for 24
hours and post-curing in an air oven at 80 °C for 16 hours. Full cure was achieved [3]. After samples
were post-cured, the polymer samples were removed from the mold’s grooves and cut into the desired
length with a vertical bandsaw. Sample preparation was followed by grinding with sandpaper (FEPA
P60, grain size 269 μm). The resin molds allowed making rectangular DMTA (40 mm x 7 mm x 2 mm)
and dogbone-shaped (200 mm x 30 mm x 2 mm with 20 mm width in the narrowest part) specimens
according to ISO 6721 and ISO 527 [129,130]. In both cases, the prepared molds allowed sufficient width
control within a tolerance of 5%. In order to get samples to the right thickness and enable sufficient
thickness control, a metal holder for grinding was prepared and used, as shown in Figure 2.3.
The desired thickness was obtained using grinding and polishing machine Jean Wirtz PHOENIX
2000 and SiC grinding discs (FEPA P500, grain size 30 μm). Exicator grease was used to enable sufficient
adhesion of the sample with the holder. The sufficient thickness control, correct length and width were
ensured within a 5% tolerance. Dogbone-shaped epoxy samples used in static tension and fatigue tests
were prepared in a similar way.
The geometry of dogbone specimens equipped with Tokyo Sokki Kenkyujo strain gauges (gauge
length of 6 mm) are shown in Figure 2.4. The specified dimensions from ISO 6721 and ISO 527 [129,130]
were achieved within 5% tolerance. The placement of strain gauges as shown in Figure 2.4 allowed to
measure strains in both the direction of the applied load and the direction normal to it, thus enabling
the calculation of Poisson’s ratio.

14
 Figure 2.3. Steel sample holder for grinding rectangular epoxy samples.

Figure 2.4. Geometry of dogbone specimens used for static tensile and fatigue tests. The placement of
strain gauges is indicated.

Reinforcement material: R-glass fibers

A typical glass fiber used for marine and oil & gas applications was selected. Boron-free and
fluorine-free high strength, high modulus 3B HiPer-TexTM W2020 R-glass fiber bundles and stitch-
bonded mats were used. These are classified as high-strength, high modulus R-glass (defined by an
international standard ISO 2078 [131]). The material was the same in both cases (bundles and mats) and
possessed the same properties. An average fiber diameter was 17 ± 2 μm [132]. The density of glass (ߩ௙ )
was 2.54 g/cm3 [132]. A single bundle had about 4098 fibers [52]. The specific surface area of glass fibers
was determined to be 0.09 m2/g from geometrical considerations as a product of number, circumference
and length of the fibers [52]. Specific surface area determined with Brunauer-Emmett-Teller (BET) of
sized and bare glass fibers was 0.180 and 0.084 m2/g, respectively [22]. Bare glass fibers were obtained
by desizing glass fibers via heat cleaning. It should also be noted, that the heat cleaning might have had
an effect on the density and the chemical nature of the surface layer of glass fibers, which could affect
the initial dissolution of the desized glass fibers (this would affect ONLY the bare fibers). R-glass fiber
bundles were used for dissolution experiments and fiber bundle tensile tests, while R-glass fiber mats

15
were used for making the laminates. A typical glass fiber Young’s modulus value was taken from
literature (72.4 GPa) [47]. This value was used throughout this work. However, later it was found that
the modulus of the studied R-glass was slightly higher (86 – 89 GPa), according to the most recent
datasheet by 3B [133]. This, however, did not affect the results significantly and did not change the
conclusions at all. All fibers used throughout this work were sized, unless otherwise stated, and are
shown in Figures 2.5 and 2.6.

Figure 2.5. R-glass fiber bundles.

Figure 2.6. Micrograph of R-glass fiber bundles taken with a digital microscope Keyence VHX6000.

16
Sizing & the sizing-rich composite interphase

The sizing is a multi-component coating that results in the formation of the composite interphase
during the manufacture of GFRPs [11]. This microconstituent has a proprietary composition. However,
it is known that typical sizings consist of about five various chemicals [76,77]. Furthermore, it is known,
that the sizing contains an organofunctional silane commonly referred to as a coupling agent [78-80].
This class of chemicals can be considered the most important in the glass fiber sizing, as it is the main
component that promotes adhesion and stress-transfer between the polymer matrix and the fiber [11].
It also provides improvements in interphase strength and hygrothermal resistance of the composite
interphase [80,133]. The silane coupling agents have the general structure [X-Si(-O-R)3] where R is a
methyl or ethyl group and X is a reactive group in respect to the polymer. When applied to fibers, it is
first hydrolysed to a silanol in presence of water. It is unstable and further condenses onto the fibers by
producing a siloxane network, which then partially becomes covalently bonded to the glass fiber
surface. During the composite manufacture, the X reactive groups of the silane may still be available to
react with the thermosetting polymer, leading to a strong network bridging between the fiber and the
matrix polymer [11]. The most common coupling agents are silane compounds [70]. According to a size
formulation patent review by Thomason and specifically a patent EP2540683A1 by Piret, Masson and
Luc of 3B, the coupling agent in the studied W2020 sizing was an epoxysilane [76,77]. Usually sizings
contain about 10 wt% of the coupling agent [88].
The composition of the sizing also consists of a number of multi-purpose components, such as a
film former which, holds the filaments together in a strand and protects the filaments from damage
through fiber-fiber contact. Film formers are as closely compatible to the polymer matrix as possible.
Epoxies, such as in this case [70], are very common film formers [78]. Usually sizings contain about 70-
80 wt% of the film former [88].
The sizing may also contain cationic or non-ionic lubricants, that reduce fiber-fiber abrasion, or
other additives, such as antistatic agents, emulsifiers, chopping aids, wetting agents or surfactants, and
antioxidants [11].
The exact composition of the sizing used in this study was not known, but based on technical
details on the given R-glass fibers elsewhere [70], it is assumed that the sizing is based on the general
characteristics described above. The results obtained are compatible with this assumption.

Glass fiber-reinforced composite laminates

Composite laminates were prepared via vacuum-assisted resin transfer molding (VARTM) using
the same curing and post-curing procedure as for the polymer. The constituent glass fibers and matrix
polymer materials were the same, as described before. The composite laminates were cut into
rectangular bars and subsequently into composite plates with dimensions 20 mm x 20 mm x 1 mm
(Papers I, IV and VI) and 50 mm x 50 mm x 1.5 mm (Paper VIII) with fibers oriented parallel (C1 plates)
and normal (C3 plates) to the large face of the plate, respectively, as shown in Figure 2.7. The thickness
was adjusted within 5% tolerance using grinding and polishing machine Jean Wirtz PHOENIX 2000
and SiC discs (FEPA P500, grain size 30 μm). The specified dimensions were achieved within 5%
tolerance.

Figure 2.7. Composite plate configurations: (left) C1 and (right) C3.

17
Reagents and other chemicals

The distilled water (resistivity 0.5-1.0 MΩ·cm) was used for conditioning of the epoxy, glass fibers
and composite samples. It was produced using the water purification system Aquatron A4000. The pH
of the distilled water was 5.650 ± 0.010, being lower than neutral due to dissolved CO 2 from atmosphere
in equilibrium.
IUPAC standard buffer solutions made by Radiometer analytical were used for studying the effect
of pH on kinetics of GF and GFRP dissolution. The solutions of pH 1.679 ± 0.010, 4.005 ± 0.010, 5.650 ±
0.010, 7.000 ± 0.010 and 10.012 ± 0.010 were used. All of the samples were put dry into the water
solutions, meaning that they were all saturated at respective pH and temperature.

2.2. EXPERIMENTAL METHODS

This section describes experimental methods used for investigating the mechanisms of
environmental aging of composite and constituent materials, and their severity. In additional modeling
techniques that were employed are described briefly.

2.2.1. Material Characterization and Testing

Determination of fiber fractions in composites

Fiber volume fractions were obtained by density measurements (Papers I, IV and VI) and burnoff
tests (Paper VIII).
The polymer and composite plates were exposed to water. The density of matrix polymer (ߩ௠ ) and
glass fiber (ߩ௙ ) was 1.1 g/cm3 and 2.54 g/cm3, respectively. The density of the composite (ߩ௖௢௠௣௢௦௜௧௘ ) was
determined to be 1.97 g/cm3 by measuring mass and dimensions of a large composite block (Papers I,
IV and VI). The volume and mass fractions of matrix polymer were calculated using the following
equations, respectively:

ߩ௖௢௠௣௢௦௜௧௘ െ ߩ௠
ܸ௙ ൌ
ߩ௙ െ ߩ௠

ߩ௙ ή ܸ௙
݉௙ ൌ
ߩ௠ ή ሺͳ െ ܸ௙ ሻ ൅ ߩ௙ ή ܸ௙

For the composite laminate in Papers I, IV and VI, the volume and mass fraction of the fibers were
ܸ௙ = 0.606 and ݉௙ = 0.780, respectively. The void content was very low (less than 0.02 %) and could be
neglected.
In Paper VIII another laminate was used made out of the same constituents. The fiber volume
fraction of the composite was ܸ௙ = 0.595 and was determined using the burn-off test, after the ASTM
Standard D3171 [135]. The void volume fraction of the composite was 0.44 % and was measured by
image analysis of optical microscope images, as was described elsewhere [23].

Determination of number of fibers & glass fiber surface area

The amount of glass fibers ݊ in samples was calculated from density and geometrical
considerations, using the following equation:

18
 ݉଴
݊ൌ
ߩ௚௟௔௦௦ ߨ݈‫ݎ‬଴ଶ

where ݈ is the length of fibers; ߩ௚௟௔௦௦ is the density of the glass; ݉଴ is the initial mass of a fiber bundle
sample; ‫ݎ‬଴ is the initial fiber radius. The number of fibers per bundle was determined to be about 4098.
External surface area of glass fibers was evaluated from geometrical considerations as a product
of number, circumference and length of fibers. The specific surface area of a studied sample can be
calculated using the following equation:

௦௣௘௖௜௙௜௖ ܵ଴ ʹߨ݈݊‫ݎ‬଴
ܵ଴ ൌ ൌ 
݉଴ ݉଴

௦௣௘௖௜௙௜௖
Specific surface area ܵ଴ was determined to be 0.09 m2/g glass. Additionally, Brunauer-
Emmett-Teller (BET) tests were performed in order to determine specific surface area of sized and bare
glass fibers. Specific surface area determined with Brunauer-Emmett-Teller (BET) of sized and bare
glass fibers was 0.180 and 0.084 m2/g, respectively [22].

Loss on ignition & sizing amount determination

The Loss On Ignition (LOI) value of the fiber bundles was determined according to the standard
practice ASTM D4963 [75]. This technique allows measurement of the weight loss of a sized glass
sample. Since the weight loss is due to the burning off of the sizing, the method can be used to determine
the amount of sizing on the fiber [11]. According to the LOI measurements the sizing was 0.64 wt% of
the fibers. The temperature during the LOI measurement was about 565 °C applied for about 5 – 5.5
hours.
The obtained LOI is consistent with literature. LOI of most glass fiber reinforcement products is
below 1.2 wt% [11]. For instance, Zinck and Gerard [136] also studied an silane-based sizing which had
a similar LOI value of 0.77 wt%.

Weight measurements of the wet samples

Samples were weighed using analytical scales Mettler Toledo AG204 (± 0.1 mg). The surface layers
of the samples immersed in water were dried using a soft and dry paper cloth, enabling to account for
only the absorbed water inside the samples. The water bath and the scale were in the same room in
vicinity of each other, which enabled performing weight measurements very quickly. We did not
experience problems related to weight instability. Furthermore, using a set of samples provided
statistical confidence in the weight gain results.

Conditioning and drying of polymer and composite samples

Conditioning in water was performed for polymer and composite plates using a batch system. A
heated distilled water (60 ± 1 °C) bath was used. Samples were weighed using analytical scales Mettler
Toledo AG204 (± 0.1 mg). Polymer and composite samples were conditioned at least until saturation
with water was achieved. Up to and at the saturation point, the samples were taken out of the water
bath, weighed and analyzed with an FTIR spectrophotometer. FTIR spectra at different water contents
allowed to determine the true water content in polymer and composites and lead to the development
of a novel spectroscopic water content monitoring method for composites (Paper I).
The drying of saturated polymer samples was performed in a drying cabinet ESAB PK-410 at 60 ±
1 °C in air atmosphere, with natural convection and relative humidity of 13 RH%. After that, samples
were reconditioned at ambient conditions in the air to regain its their initial water content.

19
 The water uptake and drying curve for polymer is shown in Figure 2.8, indicating the three sets of
samples (dry, saturated and redried) used for mechanical testing (Paper III) and for determination of
aging mechanisms (Paper II).

Figure 2.8. Water uptake and drying curves for epoxy polymer indicating true water content in dry,
saturated and redried samples.

Transmission FT-NIR spectroscopy & true water content determination

Polymer and composite characterization with different water contents was performed using
Fourier transform infrared spectroscopy (FTIR) in the near-infrared range (NIR). FT-NIR was used in
order to develop a true water content determination method for polymer and composites (Paper I).
Near-infrared spectra were obtained using the Fourier transform spectrophotometer Foss
NIRSystems 6500 operated in transmission mode, shown in Figure 2.9. An optical fiber probe and a
spectral analysis software Foss Vision were used.

Figure 2.9. Epoxy sample analyzed with transmission FT-NIR spectroscopy (left) and FT-NIR
spectrometer (right).

20
 Spectra were taken in Vis-NIR wavenumber range of 4000–25000 cm−1, using 32 scans per spectrum
with a resolution of 4 cm−1. FT-NIR spectroscopy was used to determine that the initial and redried
polymer samples had the same water content.

HR-ICP-MS & glass dissolution experiments from fiber bundles and composites

HR-ICP-MS was used to determine dissolution of R-glass fibers from bundles and composites
(Papers V-VII). Unlike the polymer, glass does not absorb water, but rather loses mass due to
dissolution.
Dissolution experiments in water were performed for glass fiber bundles and fiber-reinforced
composite plates using a batch system. Samples for dissolution study were weighed using analytical
scales Mettler Toledo AG204 (± 0.1 mg) before the experiments. The samples were placed in inert closed
vessels filled with 50 mL of distilled water or pH buffer solutions. The tight sealing of samples was
ensured. The water-tight vessels, with samples and water solutions (or buffer solutions with pH levels
of 1.679, 4.005, 5.650, 7.000 and 10.012; with an accuracy of ± 0.010) in them, were placed in the water
bath. The water´s temperature (25, 40, 60, 80 °C) in the bath was controlled via PID-controlled heating,
giving an accuracy of ± 1 °C. Two-stage heating system was used in order to ensure that there is no
contact of the sample water with other potential ion release sources, such as the heating element itself.
The concentration of the dissolved ions in the water from the vessels was analyzed in time via
High Resolution Inductively Coupled Plasma Mass Spectrometry (HR-ICP-MS) providing glass
dissolution kinetics. Experimentally, the total mass loss of glass material was measured as a sum of all
ions’ release quantified with HR-ICP-MS cumulatively over time. Analyses were performed using a
double focusing magnetic sector field HR-ICP-MS Thermo-Scientific Finnigan ELEMENT 2, equipped
with a sample introduction system ESI/Elemental Scientific PrepFAST and a pre-treatment/digestion
Milestone UltraClave. Acidification of samples was performed using ultra-pure grade HNO3 in order
to avoid adsorption of ions to the wall of the sample vials.
Infinite water availability conditions were ensured by using fresh distilled water for each separate
measurement.
The benefit of HR-ICP-MS versus gravimetric analysis is that it allows measuring dissolution
kinetics of each separate ions as well as the total mass loss [52]. The data obtained from the HR-ICP-MS
experiments are in the form of mass concentration at each time point (non-cumulative) c (g/L) and need
to be converted to the ݉ௗ௜௦௦௢௟௩௘ௗ form by using the following equation:

௧
݉ௗ௜௦௦௢௟௩௘ௗ ൌ ܸ௪௔௧௘௥ න ܿ ݀‫ݐ‬
଴

where ܸ௪௔௧௘௥ is the volume of a water sample in the HR-ICP-MS measurement. ܸ௪௔௧௘௥ used for
experiments was 50 mL. This equation is valid for each individual ion release and for the total mass
loss.
For studying the influence of external loads on dissolution kinetics, a stress corrosion rig was
designed, built and used during the experiments. The experimental rig is shown in Figure 2.10, and
design schematics of the stress corrosion setup are shown in Figure 2.11.

21
 Figure 2.10. Glass fiber stress corrosion experimental setup.

The principle is that glass fibers are inside an inert cylinder pushed by an inert rod, which transfers
the stress from the weights to the fibers. The water samples for HR-ICP-MS analyses are taken from the
main cylinder, where distilled water is in contact with stressed glass fibers. The temperature was PID-
controlled using two-stage heating system, ensuring temperature of 60 ± 1 °C. Polymeric spheres, seen
in Figure 2.10, were used to reduce the evaporation rate of the heating water.

22
 Figure 2.11. Design schematics of the stress corrosion rig for glass fiber bundles, indicating also the
placement of fibers and weights.

pH measurements & HR-ICP-MS for studying leaching from the epoxy

HR-ICP-MS and pH measurements were used to determine leaching from the epoxy polymer
(Paper II).
Leaching experiments in water were performed for polymer plates using a batch system. Samples
were weighed using analytical scales Mettler Toledo AG204 (± 0.1 mg) before the experiments. The
samples were placed in inert closed vessels filled with 50 mL of distilled water. The tight sealing of
samples was ensured. The water-tight vessels, with samples and distilled water in them, were placed
in the water bath. The water´s temperature (60 °C) in the bath was controlled via PID-controlled
heating, giving an accuracy of ± 1 °C. Two-stage heating system was used.
The concentration of the dissolved compounds in the water from the vessels was analyzed in time
via high resolution inductively coupled plasma mass spectrometry (HR-ICP-MS) providing leaching
kinetics. Experimentally, the total mass loss of glass material was measured as a sum of all elements’
release quantified with HR-ICP-MS cumulatively over time, similarly as was done in glass dissolution
experiments.
pH measurements were performed using standard pH-meter Radiometer analytical MeterLab
PHM210 (pH ± 0.01). IUPAC standard buffer solutions produced by Radiometer analytical were used
for calibration of the pH meter. The pH was determined for each sample, resulting in the pH kinetics
curves, that indicate the change of H3O+ concentration in water due to leaching. After pH was
determined, the samples were acidified using ultra-pure grade HNO3 and further analyzed with HR-
ICP-MS. HR-ICP-MS analyses were performed using a double focusing magnetic sector field HR-ICP-
MS Thermo-Scientific Finnigan ELEMENT 2, equipped with a sample introduction system
ESI/Elemental Scientific PrepFAST and a pre-treatment/digestion Milestone UltraClave.

23
Optical microscopy, glass fiber radius reduction & polymer and composite swelling measurements

Optical microscopy was performed using a digital microscope Hirox RH-2000 equipped with lens
MXB-2500REZ with a magnification of 140 and resolution of 1.06 μm. Microscopy was used for
measuring changes in glass fiber radius after various immersion times in water (Paper V) and for
determination of hygroscopic swelling of polymer and composite plates (Paper IV).
Changes in the length and width of the polymer and composites plates were measured using the
edge dimensions of the plates. The strains and swelling coefficients were obtained using the following
equations for matrix polymer and composites, respectively:

݈ െ ݈଴
ߝ௠ ൌ ൌ ߚ௠ ܹ
݈଴

݈ െ ݈଴
ߝ௜ ൌ ൌ ߚ௜ ܹ௠
݈଴

where ݅ stands for ‫ݔ‬,‫ ݕ‬and ‫ ݖ‬for respective swelling directions. Composite C1 was used to obtain
swelling coefficients in the direction parallel to fibers (ߚ௫ ) and transverse to fibers (ߚ௬ ), while composite
C3 also provided transverse-to-fibers swelling (ߚ௭ ) which was similar to and consistent with ߚ௬ .

SEM & EDX

Scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDX)

experiments were performed using Tescan Mira/LMU in a backscattered electron regime, with working
voltage of 15 kV. SEM and EDX was used to study the changes in polymer due to environmental aging
(Paper II).

ATR-FT-IR spectroscopy

Fourier transform infrared (FT-IR) spectra were recorded using Varian Scimitar 800 FT-IR in the
Attenuated Total Reflectance (ATR) mode via Pike technologies GladiATRTM mode. Spectra were
obtained at 4 cm−1 resolution, co-adding 50 scans over a range of wavenumbers from 400 to 4000 cm −1.
ATR-FT-IR was used to compare spectra of the polymer before and after environmental aging in order
to deduce aging-induced changes in the chemical structure (Paper II).

DMTA & glass transition temperature determination

Dynamic Mechanical Thermal Analysis (DMTA) tests, for determination of glass transition
temperature ܶ௚ , storage and loss moduli for dry and saturated polymer samples, were conducted using
a Netzsch GABO Eplexor qualimeter, equipped with a 1.5 kN load cell operated in displacement control
with a constant static strain of 0.4%, and a cyclic strain of 0.1% applied with a frequency of 1 Hz (Papers
II and III). The temperature sweep range was from 20 up to 120 °C, with a heating rate of 1 °C/min. The
glass transition temperature ܶ௚ was determined using DMTA as the crossing of tangents to the
inflection points in the storage modulus curves, after standard practice ISO 6721-11 [32,129]. The setup
and the sample placement are shown in Figure 2.12.

24
 Figure 2.12. Placement of the specimen (left) and the DMTA machine (right).

Uniaxial tensile testing

To evaluate the effect of the true water content on the ultimate tensile strength (UTS), Young’s
modulus and Poisson’s ratio of the dry, saturated and redried epoxy polymer, tensile tests were
conducted using the servo hydraulic test machine Instron Model 1342, shown in Figure 2.13. Dogbone-
shaped epoxy samples were used in order to determine tensile strength; the rate was set to 1 mm/min
of controlled displacement (Paper III). The temperature during the tests was about 23 °C (room
temperature). Tensile tests were performed with 4 specimens for each configuration (dry, wet and
dried). Average values and experimental scatter were reported for ultimate tensile strength, Young’s
modulus and Poisson’s ratio for each group.

Uniaxial tension-tension fatigue testing

To evaluate the effect of water content on the fatigue behavior of the dry, saturated and redried
epoxy polymer, tension-tension fatigue tests were conducted using a servo hydraulic test machine
Instron Model 1342, shown in Figure 2.13. Dogbone-shaped epoxy samples were used in order to obtain
S-N curves (Paper III). The testing frequency was chosen in order to keep a constant strain rate of
0.05%/min. The temperature during the tests was about 23 °C (room temperature). Tests were
performed at R ratio of 0.1. In fatigue, between 11 and 13 dogbone specimens were used for obtaining
S-N curves for each case (dry, wet and dried).

25
 Figure 2.13. Placement and failure of the specimen in the test machine.

2.3. MODELING

2.3.1. Numerical Modeling (FEA)

FEA is used in this work to validate the analytical prediction of the hygroscopic swelling of the
fiber-reinforced composite plates in Paper IV. The commercial FEA software package Abaqus was used
in all FEA simulation experiments. The detailed description of the FEA approach is given in Paper IV
(Appendix D).

2.3.2. Analytical Modeling

Analytical models are described shortly in the key findings of the thesis (Chapter 3) and in more
detail in respective papers. Paper I (Appendix A) describes a model for the true water content
determination in polymer and composites. Paper IV (Appendix D) describes a model for prediction of
orthotropic hygroscopic swelling of composites from isotropic swelling of a polymer matrix. Papers V-
VII (Appendices E-G) describe a model of glass dissolution prediction of fiber bundles and composites
at various environmental conditions. Paper VIII (Appendix H) describes a phenomenological mass
balance model for composites in water and predicts the kinetics of the composite interphase hydrolysis.

26
CHAPTER 3

KEY FINDINGS

3.1. TRUE WATER CONTENT DETERMINATION AND MONITORING IN

POLYMERS AND COMPOSITES (PAPER I)
It is common to report water content without accounting for moisture initially present in the
material. In reality, some water is already present in the polymer uptaken from the air, which has a
certain humidity. The true water content shows the amount of water with respect to the absolutely dry
material and is defined as:

݉௪௔௧௘௥ ሺ‫ݐ‬ሻ ݉ሺ‫ݐ‬ሻ െ ݉௔௕௦௢௟௨௧௘௟௬ௗ௥௬

ܹ ‫ כ‬ሺ‫ݐ‬ሻ ‫ؠ‬ ή ͳͲͲΨ ൌ ቆ ቇ ή ͳͲͲΨ
݉௔௕௦௢௟௨௧௘௟௬ௗ௥௬ ݉௔௕௦௢௟௨௧௘௟௬ௗ௥௬

where ܹ ‫ כ‬is the true water content (wt%); ݉௪௔௧௘௥ is the mass of water (g); ݉௔௕௦௢௟௨௧௘௟௬ௗ௥௬ is the mass of
an absolutely dry specimen (with null water content; g); ݉ is the mass of a specimen with some water
taken up (g).
A benefit of using a spectroscopic method over the conventional gravimetric analysis is the
possibility of deducing the mass of an absolutely dry material and the true water content, which is an
important indicator of water content-dependent properties, i.e. strength, stiffness and fatigue property
changes due to water content. Theoretically, it is possible to obtain mass of the absolutely dry material
with conventional drying, but usually this requires extremely long drying times. Furthermore, it is not
possible to obtain such data precisely due to humidity of the ambient air, unless perfectly dry air can
be ensured during drying and weight gain measurements. These technical problems can be easily
avoided by using a spectroscopic method.
The true water content at initial conditions was already 0.63 wt% for the studied epoxy. These
conditions are what would usually be denoted as a ‘dry’ material. Thus, the initial water content is
significant and should not be neglected. The provided method allows determining and quantifying it.
The true water content at saturation with water was 3.44 wt%. For a composite, equilibrium true water
content value (scaled by polymer mass fraction) was only slightly higher (3.56 wt%) than that of the
polymer itself. The difference was attributed to the sizing-rich interphase also absorbing water.
A detailed Fourier Transform Near-Infrared (FT-NIR) spectroscopic method for estimating and
monitoring true water content in epoxy resins and fiber-reinforced composites was developed using
the maxima of the absorbance band at about 5200 cm−1 in the Near-Infrared (NIR) combination mode
region correlated with the true water content. Based on extensive measurements of epoxy polymer and
composite samples of varying water content and thickness, regression was performed, and the
quantitative absorbance dependence on water content in the materials was successfully established.
The model equations for monitoring water content in epoxy resin and composite material samples were
obtained and experimentally validated. The model was related to the Beer–Lambert law and explained
in such terms. The details of the method were reported, allowing the use of the method in practical
applications.

Model in short

Based on the Beer–Lambert law, absorbance (‫ )ܣ‬is dependent on the molar attenuation coefficient
(ߝ), concentration (ܿ) and the path length (݈) [137]:

27
 ‫ܫ‬଴
‫ ܣ‬ൌ ݈‫݃݋‬ଵ଴ ൬ ൰ ൌ ߝ݈ܿ 
‫ܫ‬

In this case, the concentration term is the true water content (ܹ ‫) כ‬, and the path length is the
thickness of the sample (ߜ). Thus, the Beer–Lambert law can be expressed as:

‫ܭ‬௦௟௢௣௘  ‫כ‬
‫ ܣ‬ൌ ߝ ‫ ߜ כ ܹכ‬ൌ ܹ ߜ
ߜ

where ߝ ‫ כ‬is the attenuation coefficient (wt%-1mm-1) and ‫ܭ‬௦௟௢௣௘ is the attenuation coefficient for a specific
sample thickness (wt%-1), for more details see the method described in Paper I (Appendix A).
The method was extended to glass fiber-reinforced composite materials. The true water content of
the polymer matrix in the composite can be calculated as:

݉௖௢௠௣ ሺ‫ݐ‬ሻ െ ቀͳ െ ݉௙ೝ೐ೞ೔೙ ቁ ή ݉௖௢௠௣ ௜௡௜௧௜௔௟ െ ݉௥௘௦௜௡ೌ್ೞ೚೗ೠ೟೐೗೤೏ೝ೤

‫כ‬ ೔೙೔೟೔ೌ೗
ܹ௥௘௦௜௡ ൌ ή ͳͲͲΨ
݉௥௘௦௜௡ೌ್ೞ೚೗ೠ೟೐೗೤೏ೝ೤

‫כ‬
where ܹ௥௘௦௜௡ is the true water content (wt%); ݉௖௢௠௣ is the mass of a composite specimen with some
water taken up (g); ݉௖௢௠௣ ௜௡௜௧௜௔௟ is the initial mass of a composite specimen (g); ݉௙ೝ೐ೞ೔೙ is the initial
೔೙೔೟೔ೌ೗
mass fraction of the epoxy in the composite (g/g); ݉௥௘௦௜௡ೌ್ೞ೚೗ೠ೟೐೗೤೏ೝ೤ is the mass of an absolutely dry
epoxy (with null water content; g).
‫כ‬
Calculated ܹ௥௘௦௜௡ is then to be used for predicting the water content in the composite:

‫כ‬
‫ ܣ‬ൌ ‫ܭ‬௦௟௢௣௘ ܹ௥௘௦௜௡ 

Since composite materials have components of different absorbance, there is a necessity to correct
for the summary absorbance of the components via the baseline shift. In order to do so, an additional
parameter is required: a ratio, termed ܲ݁ܽ݇‫ݎ݋ݐܿܽܨ‬, of the maximum absorbance for the composite
(‫ܣ‬௣௘௔௞೎೚೘೛ ) and the polymer (‫ܣ‬௣௘௔௞ೝ೐ೞ೔೙ ) at water saturation:

‫ܣ‬௣௘௔௞೎೚೘೛
ܲ݁ܽ݇‫ؠ ݎ݋ݐܿܽܨ‬ 
‫ܣ‬௣௘௔௞ೝ೐ೞ೔೙

The final water content monitoring equation for the composite system:

‫כ‬ ‫כ‬
‫ܣ‬௣௘௔௞೎೚೘೛ ൌ ߝ ‫ כ‬ή ܲ݁ܽ݇‫ ݎ݋ݐܿܽܨ‬ή ߜ ή ܹ௥௘௦௜௡ ൌ ‫ܭ‬௦௟௢௣௘ ή ܲ݁ܽ݇‫ ݎ݋ݐܿܽܨ‬ή ܹ௥௘௦௜௡ 

Paper I (Appendix A) describes a model for the true water content determination in polymers and
composites in more detail.

3.2. CHANGES IN THE CHEMICAL STRUCTURE OF THE MATRIX POLYMER DUE

TO HYGROTHERMAL AGING (PAPER II)

The hygrothermal aging may involve both reversible and irreversible processes [20,37].
Irreversible changes are those that persist even after redrying the material to its initial water content
[32]. The epoxy yellows irreversibly, indicating that the mechanistic origin of the color change lies
among irreversible degradation pathways. Morphology was found to be unaffected. Changes in the
chemical structure of the epoxy were studied and the mechanism of yellowing was identified.
Experimental evidence in this work was obtained using a combination of FT-NIR, ATR-FT-IR, EDX,
HR-ICP-MS, pH measurements, optical microscopy, SEM, and DMTA.

28
 Irreversible aging mechanisms, which have been reported in the literature to occur during
hygrothermal influence on general epoxies are [4,32,37,138]:

1. Hydrolysis (involves chain scission)

2. Thermo-oxidation (might involve chain scission, backbone modifications or crosslinking)
3. Photo-oxidation (might involve chain scission, backbone modifications or crosslinking)
4. Residual curing (additional crosslinking)
5. Leaching (initially present additives, impurities or degradation products)
In this work, photo-oxidative effects were avoided by conducting experiments in the absence of
high-intensity light sources [139,140]. For the studied material, no chain scission (hydrolysis or
oxidation-induced) occurred [38]. To avoid residual curing, the material was fully cured. Based on this,
hydrolysis, photo-oxidation, and residual curing were excluded, whilst thermo-oxidation and leaching
occurred and were investigated further. Compounds involved in leaching were identified to be
epichlorohydrin and inorganic impurities but were found to be unrelated to yellowing. It was found,
that yellowing occurred due to the thermo-oxidative carbonyl formation in the epoxy carbon-carbon
backbone via nucleophilic radical attack.
“Weak points” for radical attack in the epoxy network were identified and are shown in Figure
3.1. In the network, 12 unique sites potentially involved in thermo-oxidation were found. Furthermore,
8 of these sites were excluded based on experimental evidence and literature [28,31,141-146]. Sites
marked in green were the identified main reactive sites (δ +DGEBA-II, δ+POPA-I, δ+HDDGE-III, and
δ+HDDGE-IV).

Figure 3.1. Chemical structure of the studied DGEBA/HDDGE/IPDA/POPA amine epoxy network
(mixing ratios are not considered). Marked sites represent “weak points” for radical attack in the
network. Sites marked in red are excluded based on experimental evidence and literature. Sites marked
in green are the main reactive sites.

Four of these unique reactive sites were involved in thermo-oxidation. One reactive site was
involved in minor thermo-oxidative crosslinking of the HDDGE segments, as shown in Scheme 1. The
other three sites were linked to carbonyl formation. Noteworthy that all three sites involved in carbonyl
formation had similar structures, containing highly reactive polyoxypropylene and i-propanol
moieties.

29
 Respective reactions were proposed. A crosslinking mechanism of HDDGE segments was
proposed in Scheme 1, involving sites δ+HDDGE-IV. This reaction is analogous to thermo-oxidative
crosslinking of DGEBU, after [143].

Scheme 1. Crosslinking reaction of the HDDGE segments via reactive sites δ+HDDGE-IV.

δ+POPA-I site as a polyoxypropylene moiety-containing segment is very susceptible to radical

attack under oxidation, due to low stability of the tertiary C-H bond and the destabilizing effect of the
neighboring ether group [31,147]. The proposed carbonyl formation reaction on this site is shown in
Scheme 2.

Scheme 2. Carbonyl formation involving polyoxypropylene moiety on reactive site δ +POPA-I.

The δ+DGEBA-II and δ+HDDGE-III sites as i-propanol moiety-containing segments are also highly
susceptible to oxidation [144,148]. The carbonyl formation on these two sites follows the same reaction,
as shown in Scheme 3 [144].

Scheme 3. Carbonyl formation reaction involving i-propanol moiety on reactive sites δ+DGEBA-II
and δ+HDDGE-III.

ܶ௚ is a useful parameter revealing chemical changes for polymers [149]. The ܶ௚ of a redried epoxy
(84.7 °C) was slightly higher than for the initial material (81.7 °C), indicating that no chain scission
occurred [141,147]. A likely reason of a ܶ௚ increase was a combination of polymer relaxation [32], anti-
plasticizing effect of leaching, and a minor crosslinking [141,143] of the HDDGE segments (Scheme 1).
It is speculated that yellowing could be prevented or delayed by adding phenolic antioxidants,
such as hindered phenols.

30
3.3. REVERSIBILITY OF THE MECHANICAL PROPERTIES OF THE MATRIX
POLYMER UPON RE-DRYING (PAPER III)
Exposure of the epoxy to water caused the material to swell. In order to investigate the influence
of water on the mechanical properties of the epoxy, static tensile and tension fatigue tests were
performed.
Swelling/plasticization is the only hygrothermal process that significantly affects mechanical
properties of the studied polymer. It was found, that mechanical properties of the studied epoxy are
reversible upon re-drying the material to its initial water content. The material regains its initial
strength, Poisson’s ratio, Young’s modulus and fatigue performance after re-drying to its initial water
content. Thus, swelling/plasticization was fully reversible for the studied epoxy.
When epoxy was saturated with water, the ultimate tensile strength (UTS) decreased from the
initial 66.4 ± 3.0 MPa down to 48.5 ± 3.3 MPa, resulting in a relative decrease of about 20% due to
swelling/plasticization by water. The results of fatigue tests indicated that the S-N curve of a wet epoxy
also shifted by 20% without a change in slope. The S-N curves of dry and dried material were identical,
further proving the reversibility in mechanical properties.
Furthermore, it is shown experimentally that the tension fatigue S-N curve of a wet epoxy resin
can be estimated by shifting the S-N curve of a dry material proportionally to a reduction in static
tensile strength due to swelling/plasticization.
Results show that Poisson's ratio increases for saturated epoxy but returns to the initial values after
drying to the initial water content, indicating that this effect is reversible. An increase in Poisson’s ratio
of epoxy due to absorbed water is consistent with literature [150]. Such increase in Poisson’s ratio is
likely due to the absorbed almost incompressible water or the reduction of glass transition temperature
(ܶ௚ ) due to the plasticizing action [150,151]. The ܶ௚ of dry, saturated and re-dried epoxy was 81.7, 59.1
and 84.7 °C, respectively.
The strain to failure increased from initial 3.71 ± 0.10% to wet 4.78 ± 0.51%. It was a brittle fracture,
and shattering often occurred for dry and re-dried specimens, while there was no shattering for the
saturated epoxy. On the micro level failure is believed to occur due to a combination of crosslink bond
breakage and disentanglement of macromolecular chains.

3.4. PREDICTION OF HYGROSCOPIC SWELLING OF THE COMPOSITE FROM

SWELLING OF THE MATRIX POLYMER (PAPER IV)
Swelling in fiber-reinforced composites is anisotropic. For orthotropic laminates, three CHEs
(ߚ௫ ǡ ߚ௬ ǡ ߚ௭ ) are needed in order to predict swelling. Quantification of the orthotropic CHEs can be
performed experimentally using samples with different fiber orientations. However, it is a time-
consuming and tedious process that also tends to involve quite high experimental scatters. Thus, an
analytical tool based on linear elasticity was developed that predicts the orthotropic swelling constants
(CHEs) of the composite from the CHE of the matrix polymer, which is isotropic. The matrix properties
are easy to measure. Furthermore, they also may be found in literature for various polymers [122,125].
The method has an advantage that it is simple-to-use in practice and requires only a swelling coefficient
of the matrix polymer, elastic constants of matrix and fibers and a known fiber volume fraction of the
composite.
Hygroscopic strains were measured after various times of exposure to water for both matrix
polymer and composite samples, and CHEs were obtained in each direction experimentally. Linear
strain behavior has been observed experimentally with increasing moisture concentration for both
composites and polymers, in agreement with [108,110]. Swelling behavior in the fiber direction (ߚ௫ )
was constrained by the non-swelling fibers and was close to null, while swelling in the transverse
directions (ߚ௬ , ߚ௭ ) was found to occur freely – similar to the unconstrained polymer.

31
 CHEs are systematized in Table 3.1. Numerical FEA and analytical prediction as well as
experimental results are shown in Table 3.1 and Figure 3.2.
Good agreement was obtained and was reported between experimental swelling data, analytical
and numerical results for composite laminates. To the best knowledge of the author, this is the first
micromechanical model that predicts the anisotropic swelling of composites from isotropic swelling of
the polymer.
During the review of the dissertation, it was pointed out that the modulus of the studied R-glass
should be slightly higher than was used in the calculations (a typical glass fiber modulus was used).
This, however, did not affect the transverse swelling predictions and did not change the conclusions.
As for the axial swelling, conclusions were also unaffected, but a slightly lower analytical axial CHE
(ߚ௫ ) was obtained (0.029 instead of 0.036).

Table 3.1. Coefficients of hygroscopic expansion. Experimental scatter is reported with one standard
deviation for the coefficient of hygroscopic expansion (CHE). * indicates an input parameter for the
models. ‡ indicates a value for a typical E-glass fiber.

CHE Experimental Analytical Numerical

βm 0.332 ± 0.021 0.332* 0.332*

βX 0.060 ± 0.018 0.029 (0.036 ‡) 0.044

βY 0.569 ± 0.066 0.546 0.552

βZ 0.576 ± 0.059 0.546 0.573

Figure 3.2. Numerical and analytical fit of swelling with the experimental data for composites.
Analytical results are shown in red lines; global averaged out numerical results are shown in blue lines,
while a numerical scatter is shown with blue sectors indicating scatter on the local scale due to various
fiber arrangements.

32
Brief description of the model

Composite transverse swelling strains (directions ‫ ݕ‬and ‫ )ݖ‬can be predicted as a serial connection
of fiber and matrix:

ߝ௬  ൌ ܸ௙ ߝ௙ ൅ ൫ͳ െ ܸ௙ ൯ߝ௠ 

where ߝ௬ is the composite transverse swelling strain, ܸ௙ the fiber volume fraction, ߝ௙ is the fiber swelling
strain and ߝ௠ is the matrix swelling strain. For many engineering reinforcements (carbon fibers, glass
fibers, etc.) the swelling is null, ߝ௙ = 0. In these cases:

ߝ௬ ൌ ൫ͳ െ ܸ௙ ൯ߝ௠ 

The transverse swelling coefficient is defined in the following equation [47]:

ߝ௬ ሺͳ െ ܸ௙ ሻߝ௠ ܹ
ߚ௬  ൌ ൌ ൌ ൫ͳ െ ܸ௙ ൯ ߚ௠ 
ܹ௖ ܹ௖ ܹ௖

where ܹ is the moisture content in the matrix and ܹ௖ the moisture content in the composite.
Composite axial swelling (direction ‫ )ݔ‬can be predicted by employing a parallel connection model
of fiber and matrix. The swelling of the matrix in this case is strongly constrained by the stiffness of the
fibers. The constrained swelling strain in the matrix ߝ௠ generates a stress equal to:

ߪ௠ ൌ ‫ܧ‬௠ ߝ௠

where ߪ௠ is the stress in the matrix and ‫ܧ‬௠ is the stiffness of the matrix.
The load transferred from the matrix to the fibers, ‫ ܮ‬can be estimated as:

‫ ܮ‬ൌ ߪ௠ ‫ܣ‬௠ ൌ ߪ௠ ‫ܣ‬ሺͳ െ ܸ௙ ሻ ൌ ߪ௙ ‫ܣ‬௙ ൌ ߪ௙ ‫ܸܣ‬௙

where ‫ ܣ‬is the cubic cell lateral surface area (in the ‫ ݔ‬െ ‫ ݖ‬plane), ‫ܣ‬௠ ൌ ‫ܣ‬ሺͳ െ ܸ௙ ሻ is the matrix part of
the cubic cell lateral surface area and ‫ܣ‬௙ ൌ ‫ܸܣ‬௙ is the fiber part of the cubic cell lateral surface area.
The stress transferred to the fiber, ߪ௙ :

ͳ െ ܸ௙
ߪ௙  ൌ ߪ௠ 
ܸ௙

Finally, the composite axial strain, ߝ௫ , can be predicted as equal to the fiber axial strain, ߝ௙ , as the
axial swelling of the cubic cell is governed by its behaviour:

ߪ௙ ߪ௠ ͳ െ ܸ௙
ߝ௫ ൌ ߝ௙ ൌ ൌ
‫ܧ‬௙ ‫ܧ‬௙ ܸ௙

where ‫ܧ‬௙ is the fiber stiffness.

The axial swelling coefficient is defined as, after [47]:

ߝ௫ ߪ௠ ͳ െ ܸ௙ ‫ܧ‬௠ ߝ௠ ͳ െ ܸ௙ ‫ܧ‬௠ ߚ௠ ܹ ͳ െ ܸ௙
ߚ௫ ൌ ൌ ൌ ൌ
ܹ௖ ܹ௖ ‫ܧ‬௙ ܸ௙ ܹ௖ ‫ܧ‬௙ ܸ௙ ‫ܧ‬௙ ܹ௖ ܸ௙

Paper IV (Appendix D) describes a model for prediction of orthotropic hygroscopic swelling of

composites from isotropic swelling of a polymer matrix in more detail.

33
3.5. PREDICTION OF MASS LOSS AND RADIUS REDUCTION OF GLASS FIBERS
DUE TO DISSOLUTION IN WATER (PAPER V)

The long-term (3194 hours) dissolution of R-glass fibers in water was studied experimentally. An
analytical model termed Dissolving Cylinder Zero-Order Kinetic (DCZOK) model was developed and
successfully used to describe the kinetics of mass loss and fiber radius reduction during the dissolution
in water. Experimentally mass loss was obtained using HR-ICP-MS for separate released ions.
The model differentiates between the complex short-term and dissolution-dominated long-term
processes. Furthermore, the novelty of the model is the ability to describe both dissolution and radius
reduction kinetics without the necessity for introducing additional terms such as conversion factor. The
model is able to predict both mass loss and radius reduction kinetics using the same four parameters:
initial fiber radius (‫ݎ‬଴ ), rate constants for both short-term degradation (‫ܭ‬଴ூ ) and steady-state degradation
(‫ܭ‬଴ூூ ) and the time when steady-state kinetics are reached (‫ݐ‬௦௧ ). All parameters can be easily determined
from initial radius measurements and mass loss evolution in time.

Figure 3.3. Separation of mass loss or cumulative ion release curves into (Phase I) short-term non-steady-
state and (Phase II) long-term steady-state regions.

The methodology was provided offering the guidelines on how to obtain the required parameters
in order to use the model in practice. The developed model is useful for both predicting the time
evolution of fiber radius reduction and material mass loss.
Elements released during degradation were determined to be Na, K, Ca, Mg, Fe, Al, Si and Cl. The
total material loss and release of each separate ion was modeled using the developed kinetic equations,
and rate constants were obtained and reported. Si contribution to the total mass loss was the largest
(56.1 wt%) and governed the dissolution process. Ca and Mg are released at approximately similar rates
to each other and contributed 14.3 and 15.1 wt% to the total mass loss, respectively, while all other
elements contributed less than 7 wt% individually, as shown in Figure 3.4. It was speculated that in the
steady-state some equilibrium composition of glass (different from the bulk composition) is obtained
in the outer layers that are contact with water, allowing elements to dissolve at some limiting rate and
proportionally to their content in the outer layers of the glass.
The rate constants ‫ܭ‬଴ூ and ‫ܭ‬଴ூூ for the glass dissolution and individual ions’ release are
systematized in Table 3.2. Rate constants are often given in relation to the surface area of a material,
thus describing dissolution behavior from a unit of the material’s surface area [61]. Obtained values are
reasonable (similar order of magnitude) compared with dissolution rates of other glass fibers, after
[152].

34
 The kinetics of radius reduction were reported. The radius reduction was found to be linear with
time. However, the induction period before the linear regime, as was also observed elsewhere [57],
which can be explained with the complexity of the short-term non-steady-state process and, to the best
knowledge of the author, was not explained before. The rate constant and the density of the glass
described the rate (proportionality) of the dissolution.

Figure 3.4. Comparison of ion release rates in the steady-state (values have been calculated from the rate
constants in the steady-state).

Table 3.2. Rate constants of glass dissolution and individual ions’ release (‫ݐ‬௦௧ = 166 hours).

ࡷࡵ૙ (g/m2·s) ࡷࡵࡵ

૙ (g/m ·s)
2

Na 6.80·10-11 1.80·10-11
K 4.85·10-11 9.80·10-12
Ca 8.72·10 -10 9.70·10-11
Mg 4.85·10-10 1.02·10-10
Fe 2.20·10 -12 4.40·10-13
Al 1.45·10-10 4.68·10-11
Si 1.10·10 -9 3.80·10-10
Cl 7.80·10-11 2.28·10-11
Glass (all ions) 3.00·10-9 6.68·10-10

Radius reduction predicted by the model after 3194 hours was only 0.0036 μm (0.04 % radius loss).
The full dissolution of the studied R-glass fibers would take about 1025 years. The radius loss after 25
years, a typical design lifetime, would be 2.45 %. These values are for 60 °C. For lower temperatures,
the radius reduction would be even less [50,51,63].

35
Brief description of the model

The model is similar in concept to an older solid-state model called the contracting cylinder [153]
or shrinking cylinder [57], which relates the evolution of conversion to time. The model presented in
this work differentiates between the complex short-term (Phase I) and dissolution-dominated long-
term (Phase II) stages. Furthermore, DCZOK model describes both mass loss and radius reduction
kinetics due to glass dissolution without the necessity for introducing additional terms such as a
conversion factor.

Figure 3.5. Schematic representation of a fiber bundle and geometrical dimensions.

Experimentally ݉ௗ௜௦௦௢௟௩௘ௗ was measured as a sum of all ions’ release quantified with HR-ICP-MS
cumulatively over time. A schematic representation of a fiber bundle and important dimensions for the
model are shown in Figure 3.5. The number of fibers is ݊ (-); the initial radius of the fibers is ‫ݎ‬଴ (m); the
length of fibers is ݈ (m).
The model involves the following assumptions. As a simplification, this model is deterministic
and all fibers are assumed to have the same initial radius, which is ‫ݎ‬଴ ; and the cross-sectional surface
area at the end of the fibers is assumed to be negligible in calculations of the surface area. The length of
the long fibers ݈ is assumed to be constant during the whole dissolution process. During the whole
degradation process, the density of the glass material stays constant (ߩ௚௟௔௦௦ ).
Dissolution is a surface reaction. The rate of the dissolution is dependent on the constant
describing the rate of the reaction (‫ܭ‬଴ ), the glass surface area exposed to water (ܵ). In infinite water
availability conditions, the surface reaction can be well-described with zero-order kinetics [57,153],
which can then be represented by a following differential equation:

߲݉
ൌ ‫ܭ‬଴ ܵ
߲‫ݐ‬

where ݉(g) is a total cumulative mass dissolved after time ‫( ݐ‬s), ‫ܭ‬଴ (g/m2·s) is a zero-order reaction
kinetic constant and ܵ (m2) is the glass surface area in contact with water.
As the reaction proceeds, the radius of the fibers is reduced and the total surface area (ܵ) is
decreased, thus leading to a decrease in the rate of mass loss. The overall ion release rate decreases
proportionally to the decrease in total surface area or a decrease in fiber radius.
The volume of a single fiber is ߨ‫ ݎ‬ଶ ݈, where ݈ is the cylinder length and ‫ ݎ‬is the cylinder radius. For
݊ fibers, the volume is ݊ߨ‫ ݎ‬ଶ ݈ and mass is ߩ௚௟௔௦௦ ݊ߨ‫ ݎ‬ଶ ݈. The surface area of a single fiber is ʹߨ‫݈ݎ‬. For ݊
fibers it is ʹ݊ߨ‫݈ݎ‬. Substituting mass and surface area expressed in such terms into the following
equation:

36
 ߲‫ ݎ‬ଶ ʹ‫ܭ‬଴
ൌ ‫ݎ‬
߲‫ݐ‬ ߩ௚௟௔௦௦

The final mass loss kinetic model equation in differential form is obtained:

߲݉ ‫ܭ‬଴ଶ
ൌ ʹ݊ߨ݈ ቆ‫ݎ‬଴ ‫ܭ‬଴ െ ‫ݐ‬ቇ
߲‫ݐ‬ ߩ௚௟௔௦௦

The final model equations combining approximated non-steady-state short-term and physical
steady-state long-term dissolution kinetics are proposed. The complete radius reduction DCZOK
model:

‫ܭ‬଴ூ
‫ۓ‬ ‫ ݐ‬൑ ‫ݐ‬௦௧ ǣ‫ ݎ‬ൌ ‫ݎ‬଴ െ ‫ݐ‬
ۖ ߩ௚௟௔௦௦
‫۔‬ ‫ܭ‬଴ூூ
ۖ‫ ݐ‬൐ ‫ݐ‬௦௧ ǣ‫ ݎ‬ൌ ‫ݎ‬௧ೞ೟ െ ሺ‫ ݐ‬െ ‫ݐ‬௦௧ ሻ
‫ە‬ ߩ௚௟௔௦௦

The complete mass loss DCZOK model:

‫ܭ‬଴ூଶ ଶ
‫ۓ‬ ‫ ݐ‬൑ ‫ݐ‬௦௧ ǣ݉ௗ௜௦௦௢௟௩௘ௗ ൌ ݊ߨ݈ ቆʹ‫ݎ‬଴ ‫ܭ‬଴ூ ‫ ݐ‬െ ‫ ݐ‬ቇ
ۖ ߩ௚௟௔௦௦
‫۔‬ ‫ܭ‬଴ூூଶ
ۖ‫ ݐ‬൐ ‫ݐ‬௦௧ ǣ݉ௗ௜௦௦௢௟௩௘ௗ ൌ ݉ௗ௜௦௦௢௟௩௘ௗ೟ೞ೟ ൅ ݊ߨ݈ ቆʹ‫ݎ‬௧ೞ೟ ‫ܭ‬଴ூூ ሺ‫ ݐ‬െ ‫ݐ‬௦௧ ሻ െ ሺ‫ ݐ‬െ ‫ݐ‬௦௧ ሻଶ ቇ
‫ە‬ ߩ௚௟௔௦௦

where ‫ܭ‬଴ூ and ‫ܭ‬଴ூூ are the rate constants (g/m2·s) for the short-term non-steady-state and long-term
steady-state regions, respectively; ‫ݎ‬௧ೞ೟ (m) and ݉ௗ௜௦௦௢௟௩௘ௗ೟ (g) are the fiber radius and lost mass after time
ೞ೟
‫ݐ‬௦௧ (s), when steady-state is reached.
Paper V (Appendix E) describes a model for prediction of glass dissolution of fiber bundles in
more detail. Also note, that in this Section 3.5 and in Paper V (Appendix E), rate constants ‫ܭ‬଴ include
the protective effect of the sizing (ߦ௦௜௭௜௡௚ ), and are further treated as apparent dissolution rate constants
‫ܭ‬଴‫ כ‬௦௜௭௘ௗ௙௜௕௘௥௦ , as is described in Papers VI and VII (Appendices F-G).

3.6. PREDICTION OF MASS LOSS OF GLASS FIBERS INSIDE THE COMPOSITE

DUE TO HYGROTHERMAL AGING (PAPER VI)

An analytical model for prediction of long-term dissolution of glass fibers termed Dissolving
Cylinder Zero-Order Kinetic (DCZOK) model was extended for thin fiber-reinforced composite plates
and was presented and explained. The model describes mass loss kinetics during hygrothermal aging
of desized and sized R-glass fiber bundles and R-glass fiber-reinforced composite plates with various
fiber orientations. Effects of sizing, availability of water and accumulation of degradation products
were discussed. The model predicts the mass loss during hygrothermal aging of bare and sized glass
fiber bundles and fiber-reinforced composites.
The sizing slowed down glass dissolution of fiber bundles dramatically. The protective effect of
sizing ߦ௦௜௭௜௡௚ (0.165) on glass fiber dissolution was found to be about six times.
Compared to dissolution of sized fiber bundles, the long-term dissolution of glass from composites
was slowed down by 36.84% and 65.26% depending on fiber orientation. Slower dissolution from
composites compared to unprotected desized glass was explained with the effect of sizing, limited
water availability and due to silica degradation product accumulation inside the composite.
A method to decouple water availability and accumulation terms was proposed by using
measurements for composite samples with different fiber orientations.

37
 Apparent glass dissolution rate constants were obtained from experimental long-term dissolution
kinetics data for Si loss from sizeless and sized glass fiber bundles as well as for composite plates.
Obtained rate constants are reported in Table 3.3. The goodness of fit with experimental data is
represented by determination coefficients R2.

Table 3.3. Long-term dissolution of R-glass fiber bundles and reinforced composite plates.

K0* (g Si/m2·s) R2
Sizeless Fiber Bundle 2.30·10-9 0.9576
Sized Fiber Bundle 3.80·10-10 0.9781
Composite Plate C1 1.32·10 -10 0.9953
Composite Plate C3 2.40·10-10 0.9923

Brief description of the model

The model presented here is a general case of the DCZOK model presented in Paper V (Appendix
E). The rate of the dissolution is dependent on the zero-order reaction kinetic constant (‫ܭ‬଴ ), the glass
surface area exposed to water (ܵ), presence of sizing (ߦ௦௜௭௜௡௚ ), the availability of water (‫ܥ‬ுమை ) and the
order of the reaction (݊௢௥ௗ௘௥ ). In addition, as the aging proceeds, degradation products are accumulated
in the composites and slow down the rate of the reaction. Since the long-term reaction is governed by
Si dissolution [52], the silica hydrolysis products are what causes the deceleration of glass dissolution
inside the composites. In the model, the accumulation term is accounted for as a driving force term,
௘௤
that shows that rate of the mass loss is proportional to the difference between saturation (‫ܥ‬ௌ௜ைమ ) and
current concentration (‫ܥ‬ௌ௜ைమ ) of degradation products in the composite and the order (݉௢௥ௗ௘௥ ). The
global model (general case) can then be mathematically expressed as the following equation:

߲݉ ௡ ௘௤
௠೚ೝ೏೐ೝ
ൌ ‫ܭ‬଴ ߦ௦௜௭௜௡௚ ܵ‫ܥ‬ுమ೚ೝ೏೐ೝ
ை ቀ‫ܥ‬ௌ௜ைమ െ ‫ܥ‬ௌ௜ைమ ቁ ൌ ‫ܭ‬଴‫ܵ כ‬
߲‫ݐ‬

where ݉is a total cumulative mass dissolved after time ‫ܭ ;ݐ‬଴‫ כ‬is an apparent reaction kinetic constant
that can be obtained from regression of experimental data. While ‫ܭ‬଴ is a material property, ‫ܭ‬଴‫כ‬
incorporates effects of sizing, water availability and degradation product accumulation.
For sized fibers, effect of sizing (ߦ௦௜௭௜௡௚ ) should be accounted for. The model simplifies to:

߲݉
ൌ ‫ܭ‬଴ ߦ௦௜௭௜௡௚ ܵ ൌ ‫ܭ‬଴‫ כ‬௦௜௭௘ௗ௙௜௕௘௥௦ ܵ
߲‫ݐ‬

For bare (desized) fibers the model simplifies to:

߲݉
ൌ ‫ܭ‬଴ ܵ ൌ  ‫ܭ‬଴‫ כ‬௦௜௭௘௟௘௦௦௙௜௕௘௥௦ ܵ
߲‫ݐ‬

Paper VI (Appendix F) describes a model for prediction of glass dissolution prediction of sized,
desized fiber bundles and composites in more detail.

38
3.7. PREDICTION OF GLASS FIBER DISSOLUTION AT VARIOUS
ENVIRONMENTAL CONDITIONS (PAPER VII)
The long-term dissolution of R-glass fibers in water was studied experimentally at various
environmental conditions, such as varying pH levels, temperature and under the effect of stress
corrosion. Experimentally mass loss was obtained using HR-ICP-MS for separate released ions.
The analytical model termed Dissolving Cylinder Zero-Order Kinetic (DCZOK) model was
successfully used to explain the long-term glass dissolution experiments of R-glass fibers at various
environmental conditions. The kinetic constants were obtained for various pH and temperature
conditions as well as for different stress levels. The protective effect of sizing was accounted for in this
model [24]. The obtained rate constants are systematized in Tables 3.4, 3.5 and 3.6 for the effect of pH,
temperature and stress, respectively. Obtained values are comparable with dissolution rates of other
glass fibers studied in the literature [24,52,152].
Temperature showed an Arrhenius-type influence on the kinetics, increasing rate of dissolution
exponentially with increasing temperature:

ாಲ ሺ௣ுǡఙሻ
‫ܭ‬଴ ൌ ‫ି ݁ܣ‬ ோ்

where ‫ ܣ‬is the pre-exponential factor (g/(m2·s)); R is the gas constant being 8.314 J/(mol·K); T is the
absolute temperature (K); ‫ܧ‬஺ is the activation energy (J/mol). Both pH and stress corrosion are thought
to affect the activation energy term in the Arrhenius equation [154,155].
Activation energy of the steady-state glass dissolution was obtained and reported at pH 5.65 and
no stress (53.46 and 34.84 kJ/mol for Si and total glass dissolution, respectively). Obtained values are
consistent with values reported in literature, being slightly lower than for E-glass (58–79 kJ/mol in
alkaline solutions [57]). Activation energy for Si dissolution of various silica-based materials may be
around 60 kJ/mol [156]. The steady-state (Phase II) was achieved after about a week.

Table 3.4. The glass dissolution rate constants obtained via regression of the experimental data of
R-glass fiber bundles using DCZOK model for Si and total mass loss at 60 °C and various pH.

pH ࡷࡵ૙ࡿ࢏ (g/(m2·s)) ࡷࡵࡵ

૙ࡿ࢏ (g/(m ·s))
2 ࡷࡵ૙࢚࢕࢚ࢇ࢒ (g/(m2·s)) ࡷࡵࡵ
૙࢚࢕࢚ࢇ࢒ (g/(m ·s))
2

1.679 ± 0.010 (3.10±0.44)·10-7 (1.25±0.09)·10-7 (1.70±0.19)·10-6 (1.16±0.08)·10-6

4.005 ± 0.010 (2.59±0.33)·10-8 (1.70±0.11)·10-8 (8.48±1.21)·10-8 (6.24±0.36)·10-8

5.650 ± 0.010 (6.67±1.03)·10-9 (2.30±0.16)·10-9 (1.82±0.29)·10-8 (4.05±0.29)·10-9

7.000 ± 0.010 (3.64±0.53)·10-8 (2.55±0.19)·10-8 (5.46±0.82)·10-8 (4.85±0.38)·10-8

10.012 ± 0.010 (8.97±1.27)·10-8 (4.56±0.32)·10-8 (1.39±0.16)·10-7 (1.11±0.07)·10-7

The activation energy of glass dissolution was affected by pH. The influence of pH is complicated
and may be described by a parabolic polynomial function – the activation energy peaks at pH 5.65 and
decreases towards more basic, as well as more acidic conditions. In comparison with neutral conditions,
basic and acidic aqueous environments showed an increase in dissolution rates, affecting the lifetime
of glass fibers negatively. GFRPs should not be used in strongly acidic conditions. This is in agreement
with an observation in another study, stating that many GFRPs fail catastrophically after a critical time
when exposed to acids [47]. The steady-state (Phase II) was achieved after about a week.

39
 Table 3.5. The glass dissolution rate constants obtained via regression of the experimental data
using DCZOK model for Si and total mass loss at pH 5.65 and various temperatures.

T (°C) ࡷࡵ૙ࡿ࢏ (g/(m2·s)) ࡷࡵࡵ

૙ࡿ࢏ (g/(m ·s))
2 ࡷࡵ૙࢚࢕࢚ࢇ࢒ (g/(m2·s)) ࡷࡵࡵ
૙࢚࢕࢚ࢇ࢒ (g/(m ·s))
2

25 ± 1 (1.46±0.23)·10-9 (2.60±0.18)·10-10 (1.04±0.12)·10-8 (1.42±0.11)·10-9

40 ± 1 (2.62±0.37)·10-9 (1.08±0.08)·10-9 (1.37±0.19)·10-8 (2.72±0.19)·10-9

60 ± 1 (6.67±1.03)·10-9 (2.30±0.16)·10-9 (1.82±0.29)·10-8 (4.05±0.29)·10-9

80 ± 1 (2.19±0.31)·10-8 (8.91±0.73)·10-9 (4.24±0.59)·10-8 (1.47±0.11)·10-8

The activation energy of glass dissolution is affected by stress corrosion. The activation energy
decreases linearly as the stress increases. The steady-state (Phase II) was achieved after about 5 days
(120 h), sooner than in all unstressed cases, indicating that stress may slightly accelerate the transition
towards the steady-state dissolution.

Table 3.6. The glass dissolution rate constants obtained via regression of the experimental data
using DCZOK model for Si and total mass loss at 60 °C and pH 5.65 at various stress levels.

࣌ ࡷࡵ૙ࡿ࢏ ࡷࡵࡵ
૙ࡿ࢏ ࡷࡵ૙࢚࢕࢚ࢇ࢒ ࡷࡵࡵ
૙࢚࢕࢚ࢇ࢒
(MPa) (g/(m ·s))
2 (g/(m ·s))
2 (g/(m ·s))
2 (g/(m2·s))

0.0 (6.67±1.03)·10-9 (2.30±0.16)·10-9 (1.82±0.29)·10-8 (4.05±0.29)·10-9

0.2 (7.27±1.15)·10-9 (2.38±0.21)·10-9 (2.08±0.42)·10-8 (4.12±0.33)·10-9

26.4 (7.45±1.03)·10-9 (2.58±0.19)·10-9 (2.04±0.33)·10-8 (4.35±0.35)·10-9

42.2 (8.30±1.33)·10-9 (3.47±0.25)·10-9 (2.01±0.41)·10-8 (5.56±0.44)·10-9

52.7 (9.21±1.33)·10-9 (4.73±0.33)·10-9 (1.65±0.44)·10-8 (8.12±0.67)·10-9

The increase in stress reduces the activation energy of dissolution linearly and accelerates the glass
dissolution rates exponentially.

Brief description of the model

For the sized fiber bundles, the mass loss kinetic DCZOK model equation in differential form is
the following [52]:

ଶ
߲݉ ‫ܭ‬଴ଶ ߦ௦௜௭௜௡௚
ൌ ʹ݊ߨ݈ ቆ‫ݎ‬଴ ‫ܭ‬଴ ߦ௦௜௭௜௡௚ െ ‫ݐ‬ቇ
߲‫ݐ‬ ߩ௚௟௔௦௦

where ݊ is the number of fibers (-); ݈ is the length of fibers (m); ‫ݎ‬଴ is the initial fiber radius (m), and ߩ௚௟௔௦௦
is the density of glass (g/m3).
pH, temperature and stress corrosion affect the material-environment energy-activated
interactions, thus affecting the dissolution rate constants ‫ܭ‬଴ :

‫ܭ‬଴ ൌ ݂ሺ‫ܪ݌‬ǡ ܶǡ ߪሻ

where ‫ ܪ݌‬is the acidity of the environment (-), ܶ is its temperature (K), and ߪ is stress (MPa).

40
 The DCZOK model was expanded to account for the environmental conditions (pH, T, σ):

ாಲ ሺ௣ுǡఙሻ ଶ

߲݉
ଶ
൫‫ܭ‬଴ ߦ௦௜௭௜௡௚ ൯ ‫ۇ‬ ாಲ ሺ௣ுǡఙሻ ൬‫ି ݁ܣ‬ ோ் ߦ௦௜௭௜௡௚ ൰ ‫ۊ‬
ൌ ʹ݊ߨ݈ ൭‫ݎ‬଴ ‫ܭ‬଴ ߦ௦௜௭௜௡௚ െ ‫ݐ‬൱ ൌ ʹ݊ߨ݈ ‫ݎ‬଴ ‫ ି ݁ܣ‬ோ் ߦ௦௜௭௜௡௚
‫ۈ‬ െ ‫ۋ ݐ‬
߲‫ݐ‬ ߩ௚௟௔௦௦ ߩ௚௟௔௦௦
‫ۉ‬ ‫ی‬

where ݉ is a total cumulative mass dissolved after time ‫ܭ ;ݐ‬଴ is a material/environment interaction
property; ߦ௦௜௭௜௡௚ is the protective effect of sizing; ‫ ܪ݌‬is the acidity of the environment (-), ܶ is its
temperature (K); ߪ is stress (MPa); ݊ is the number of fibers (-); ݈ is the length of fibers (m); ‫ݎ‬଴ is the
initial fiber radius (m); ߩ௚௟௔௦௦ is the density of glass (g/m3); ‫ ܣ‬is the pre-exponential factor (g/(m2·s)); R
is the gas constant being 8.314 J/(mol·K); T is the absolute temperature (K); ‫ܧ‬஺ is the activation energy
(J/mol).
In this equation, ‫ Ͳܭ‬andߦ௦௜௭௜௡௚ are the time-independent parameters, whereas ܵ changes with time.
As the dissolution proceeds, the radius of the fibers is reduced and the total surface area ܵ is decreased,
thus leading to a decrease in the dissolution rate. For each ion, individual ‫ܭ‬଴ூ and ‫ܭ‬଴ூூ values have to be
obtained, while ‫ݐ‬௦௧ and ‫ݎ‬଴ for each individual ion release are the same and are equivalent to that for the
total mass loss [52]. ‫ܭ‬଴ூ and ‫ܭ‬଴ூூ are the dissolution rate constants (g/(m2·s)) for the short-term non-
steady-state (Phase I) and long-term steady-state regions (Phase II), respectively.
The model involves the following assumptions. As a simplification, this model is deterministic
and all fibers are assumed to have the same initial radius ‫ ; Ͳݎ‬the cross-sectional surface area at the end
of the fibers is assumed to be negligible in calculations of the surface area; the length of the long fibers
݈ is assumed to be constant during the whole dissolution process. During the whole degradation
process, the density of the glass material is assumed constant ߩ݈݃ܽ‫ ݏݏ‬. The effect of sizing ߦ‫ ݃݊݅ݖ݅ݏ‬is
assumed to be independent of environmental conditions and time [24,52]. For free fiber bundles (not
embedded in the composite) the conditions of infinite availability of water are ensured by using large
volumes of water, thus making the rate of reaction independent of the water concentration [24,52].
The model can also predict fiber radius reduction kinetics during the dissolution using the same
four parameters, that can be determined experimentally [52]. The DCZOK model can be used to
describe dissolution for each element separately as well as for the total mass loss.
Paper VII (Appendix G) describes a model for prediction of glass dissolution prediction of sized
fiber bundles at various environmental conditions in more detail.

3.8. LONG-TERM AGING OF THE SIZING-RICH COMPOSITE INTERPHASE

(PAPER VIII)
Composite plates of two configurations C1 and C3, as was shown in Figure 2.7, were conditioned
for a period of about a year in a heated bath with distilled water (60 ± 1 °C). The C1 was representative
of a typical composite, whereas C3 was unusual with a short fiber-matrix interface length and the
interphases being connected to the large sample´s surface.
In water, the sizing-rich composite interphase was exposed to hydrolytic degradation. Glass fiber
composites absorb water with time and the mass of the composites increase subsequently. When
measuring diffusivity and saturation level of water according to ASTM D5229 testing is stopped when
the mass increase with time stops [157], i.e. it is reaching a plateau, in this case at about 200 hours. If
the water uptake experiments are stopped as suggested by ASTM, then the long-term behavior is not
captured, as shown in Figure 3.6. However, continuing the tests exposing the laminates to water for
longer, the mass of a typical composite C1 increases again, measured up to 9 months. This observation
is also consistent with the results of another study on long-term water uptake by composite plates [158].
This additional water uptake was found to be due to the hydrolytic degradation of the sizing-rich fiber
matrix interphase.

41
 Figure 3.6. Long-term water uptake by composite laminates. Dashed line corresponds to a time when a
test following standard practice ASTM D5229 would be stopped [157].

Due to water-induced dissolution interphase flaws were formed, as shown schematically in Figure
3.7, which developed further into matrix cracks, observed in micrographs in Figure 3.8. The internal
volume created by the flaws and cracks can be filled with water leading to the observed mass increase.
The microscopically measured size of the flaws matches the order of magnitude of the volume required
for obtaining the measured additional mass increase.

Figure 3.7. Interphase flaw is formed and gets filled with water.

42
 Figure 3.8. Micrograph of a composite sample exposed to water for 6673 hours at 60 °C. The micrograph
indicates the (A) Fiber matrix debondings; (B) Matrix transverse cracks, (C) Splitting along the fibers.

Three damage mechanisms were observed in the micrographs:

1. Fiber matrix debondings, shown in Figure 3.8(A).

2. Matrix transverse cracks, shown in Figure 3.8(B). These cracks seem to be inside the bundle.
This location may be also a result of the weakening of the fiber matrix interphase, which was
covered in point 1.
3. Splitting along the fibers, shown in Figure 3.8(C).

Fiber matrix debonding appears to be the first failure mechanism, caused by hydrolysis of the
interphase. When these debondings accumulate creating a weakened local region they can easily
combine into a longer “matrix crack” due to a release of curing, thermal and swelling stresses, resulting
in a crack formation. The reason for the observed splitting along fibers is less clear. It could be related
to the matrix cracks, but it could also be caused by the fibers used for stitching the reinforcing mat. All
these flaws (cracks) create volume that can be filled with water and increases the mass of the composite.
When C3 specimens were conditioned in water their mass increased during the first 200 hours
similar to C1. Continuing the test for longer times lead, however, to a mass loss, seen in Figure 3.6. For
these specimens the flaws created by the fiber matrix interphase hydrolysis were open towards the
surface of the test specimen, since the interphase length (and fiber length) was so short, 1.5 mm. The
reaction products of the hydrolysis could migrate into the surrounding water bath leading to a mass
drop. This mass loss allowed determining the product of the dissolution rate constant and the surface
area of the interphase.
The small specimens tested here would degrade the entire interphase within 22 to 30 years at 60
°C. The calculation is based on a full mechanistic mass balance approach considering all the composite´s
constituents: water uptake and leaching of the matrix, dissolution of the glass fibers and dissolution of
the composite interphase. These processes were modeled using a combination of Fickian diffusion and
zero-order kinetics. The mass loss due to long-term gravimetric behavior of composite C3 was
successfully modeled, because the C3 samples did not have a significant accumulation of the
degradation products.
Based on long-term test data from the literature tested for close to 10 years it seems that typical
composites, such as C1, will initially absorb extra water in the flaws and cracks created by interphase
hydrolysis. Eventually these cracks will create a network that is connected to the surface of the
composite laminate. When this network is formed reaction products can leave the laminate and the
mass will be reduced similarly to C3.
Damage caused by the hydrolytic aging of the sizing-rich composite interphase very likely leads
to a decrease in interfacial strength. For instance, Gagani et al [159] and Rocha et al [32] have reported
the composite interphase-related deterioration of the mechanical properties due to aging in water. It is

43
likely that the formation of the interphase flaws described in this work is the mechanistic origin of the
interfacial strength deterioration of composites.
Paper VIII (Appendix H) covers hydrolysis of the composite interphase, but the same approach
should be applicable for all other environmental agents and solvents (in general, solvolysis).

Brief description of the model

The combination of the phenomenological perspective and mass balance approach provide a
useful tool for analyzing mass uptake/loss processes in composites during hygrothermal aging by
breaking down a complex process into constituent-related processes. The processes that affect weight
gain or loss of composites are summarized in Table 3.7. The proposed model equation should be a
phenomenological full representation of the interaction between the composite material and the water
environment.

Table 3.7. Summary of the processes during hygrothermal aging of composites that affect the mass balance.

Process Sign Reference

Water uptake of the polymer matrix + [23]

Water uptake by the composite interphase + [53]
Water uptake by the voids + [11,23,160]
Thermo-oxidation of the polymer matrix + [3]
Leaching from polymer matrix - [3]
Glass fiber dissolution - [24,43,52]
Sizing-rich interphase dissolution - [22]

Gravimetric measurements determine the sample´s mass over time during conditioning in water.
The mass consists of the following terms:

݉௚௥௔௩௜௠௘௧௥௜௖ ሺ‫ݐ‬ሻ ൌ ݉ௗ௥௬ ൅ ݉௪௔௧௘௥௨௣௧௔௞௘ ሺ‫ݐ‬ሻ ൅ ݉௢௫௜ௗ௔௧௜௢௡ ሺ‫ݐ‬ሻ െ ݉௟௘௔௖௛௜௡௚ ሺ‫ݐ‬ሻ െ ݉௚௟௔௦௦ௗ௜௦௦௢௟௨௧௜௢௡ ሺ‫ݐ‬ሻ െ ݉௜௡௧௘௥௣௛௔௦௘ௗ௜௦௦௢௟௨௧௜௢௡ ሺ‫ݐ‬ሻ

The dissolution of the interphase is then simply given by:

݉௜௡௧௘௥௣௛௔௦௘ௗ௜௦௦௢௟௨௧௜௢௡ ሺ‫ݐ‬ሻ ൌ ݉ௗ௥௬ ൅ ݉௪௔௧௘௥௨௣௧௔௞௘ ሺ‫ݐ‬ሻ ൅ ݉௢௫௜ௗ௔௧௜௢௡ ሺ‫ݐ‬ሻ െ ݉௟௘௔௖௛௜௡௚ ሺ‫ݐ‬ሻ െ ݉௚௟௔௦௦ௗ௜௦௦௢௟௨௧௜௢௡ ሺ‫ݐ‬ሻ െ ݉௚௥௔௩௜௠௘௧௥௜௖ ሺ‫ݐ‬ሻ

The water uptake and leaching can be calculated using the Fickian diffusion [23,157]:

௠ ௩
‫ܯ‬ஶ ሺߥ௠ ൅ ߥ௜ ሻߩ௠ ൅ ‫ܯ‬ஶ ߥ௩ ߩ௪௔௧௘௥
‫ܯ‬ஶ ൌ 
ߥ௙ ߩ௙ ൅ ሺߥ௠ ൅ ߥ௜ ሻߩ௠

஽௧ బǤళఱ
ି଻Ǥଷቀ మ ቁ
‫ܯ‬ሺ‫ݐ‬ሻ ൌ ‫ܯ‬ஶ ቈͳ െ ݁ ௛ ቉

where ߩ௠ is the matrix density, ߩ௙ is the fiber density, ߩ௪௔௧௘௥ is the water density, ߥ௙ is the fiber volume
fraction, ߥ௠ is the matrix volume fraction,ߥ௜ is the interphase volume fraction, ߥ௩ is the void volume
௠ ௩
fraction (ߥ௙ ൅ ߥ௠ ൅ ߥ௜ ൅ ߥ௩ ൌ ͳ), ‫ܯ‬ஶ is the matrix saturation water content (3.44 wt%) and ‫ܯ‬ஶ is the

44
void saturation water content (100 wt%). ‫ܯ‬ሺ‫ݐ‬ሻ is the water content, ‫ܯ‬ஶ is the water saturation content,
‫ ݐ‬is time, ݄ is the thickness and ‫ ܦ‬is the diffusivity in the thickness direction of the plate.

Figure 3.9. Experimental composite C3 plate mass change during the conditioning in water, shown over
a square root of time. Water uptake and mass balance are modeled.

The dissolution kinetics of glass and the interphase can be calculated using zero-order kinetics.
The full Dissolving Cylinder Zero-Order Kinetics (DCZOK) model is the following [52]:

‫ܭ‬଴‫כ‬ூଶ ଶ
‫ۓ‬ ‫ ݐ‬൑ ‫ݐ‬௦௧ ǣ݉ௗ௜௦௦௢௟௩௘ௗ ൌ ݊ߨ݈ ቆʹ‫ݎ‬଴ ‫ܭ‬଴‫כ‬ூ ‫ ݐ‬െ ‫ ݐ‬ቇ
ۖ ߩ௙

‫۔‬ ‫ܭ‬଴‫כ‬ூூଶ
ۖ‫ ݐ‬൐ ‫ݐ‬௦௧ ǣ݉ௗ௜௦௦௢௟௩௘ௗ ൌ ݉ௗ௜௦௦௢௟௩௘ௗ೟ೞ೟ ൅ ݊ߨ݈ ቆʹ‫ݎ‬௧ೞ೟ ‫ܭ‬଴‫כ‬ூூ ሺ‫ ݐ‬െ ‫ݐ‬௦௧ ሻ െ ሺ‫ ݐ‬െ ‫ݐ‬௦௧ ሻଶ ቇ
‫ە‬ ߩ௙

where ݊ is the number of fibers; ݈ is the length of fibers; ‫ݎ‬଴ is the initial fiber radius; ߩ௚௟௔௦௦ is the density
of glass ; ‫ܭ‬଴‫כ‬ூ and ‫ܭ‬଴‫כ‬ூூ are the apparent dissolution rate constants for the short-term non-steady-state
(Phase I) and long-term steady-state (Phase II) regions, respectively; ‫ݎ‬௧ೞ೟ and ݉ௗ௜௦௦௢௟௩௘ௗ೟ are the fiber
ೞ೟
radius and lost mass after time ‫ݐ‬௦௧ (s), when steady-state is reached.
Paper VIII (Appendix H) describes a model for prediction of the hydrolytic degradation kinetics
of the composite interphase in more detail.

45
46
CHAPTER 4

CONCLUSIONS

The microconstituents of the composites were studied in order to evaluate the water-induced
degradation. Environmental aging mechanisms of composite microconstituent materials (epoxy, glass
fibers and sizing) were successfully identified. Models and methods were developed for studying and
predicting the changes in constituent and composite materials’ properties due to the environmental
aging. The severity of hygrothermal aging on the mechanical properties was also obtained and
explained for epoxy, whereas deterioration of interfacial strength for the same material system was
covered together with Dr. Abedin I. Gagani from the same group at NTNU elsewhere [159].
A spectroscopic method of the true water content determination in the matrix and in composites
was developed. The true water content at saturation was 3.44 wt%, whereas at initial conditions it was
already 0.63 wt% for the studied epoxy. These initial conditions are what would usually be denoted as
a ‘dry’ material. Thus, the initial water content is significant and should not be neglected.
Mechanical properties and chemical structure of the epoxy and the effect of water environment
was studied. For the studied amine-based epoxy, mechanical properties were affected (negatively) only
via plasticization/swelling mechanism in the presence of water. This negative effect was fully reversible
upon redrying the material to its initial water content. Changes in chemical structure involved thermo-
oxidation, which took a few distinct pathways: minor thermo-oxidative crosslinking and carbonyl
formation in the carbon-carbon backbone of the polymer. The carbonyl formation caused the material
to yellow irreversibly. In addition, small molecular size compounds, that were initially present in the
polymer, were diffusing out into the water due to leaching. Thermo-oxidation and leaching did not
show any significant effect on the deterioration of mechanical performance of the epoxy.
Since swelling was found to be the only mechanism that affects the mechanical properties of the
studied epoxy polymer significantly, an analytical model for the prediction of hygroscopic swelling in
fiber-reinforced composites from the isotropic swelling data of the matrix polymer was developed.
For the studied R-glass, long-term dissolution experiments were performed, and the dissolution
behavior was obtained. Environmental aging occurred in two distinct phases: short-term non-steady-
state (Phase I) and long-term steady-state (Phase II). The first one involved many subprocesses
occurring in parallel, but relatively quickly transitioned into the second phase, in about a week’s time.
This means that Phase II is of most interest for the long-term degradation. In Phase II, the degradation
was dominated by glass dissolution and was successfully modeled using the Dissolving Cylinder Zero-
Order Kinetics (DCZOK) model at various environmental conditions (pH, temperature and stress
corrosion). In addition, glass fiber dissolution from thin composites was successfully modeled using
the DCZOK model.
The sizing-rich composite interphase degrades due to hydrolysis, resulting in the formation of the
interphase flaws. These flaws may further develop into fiber/matrix debondings, matrix cracks and
splitting along the fibers, as was observed in micrographs. The internal volume created by the flaws
and cracks can be filled with water leading to the observed mass increase. Based on the combination of
experimental evidence of about a year and long-term test data from the literature [158] tested for close
to 10 years it seems that typical composites will initially absorb extra water in the flaws and cracks
created by interphase hydrolysis. Eventually these cracks will create a network that is connected to the
surface of the composite laminate. When this network is formed reaction products can leave the
laminate and the mass will be reduced continuously.
Another important aspect that has to be noted is that when measuring diffusivity and saturation
level of water according to ASTM D5229 testing is stopped when the mass increase with time stops
[157], i.e. it is reaching a plateau. However, the water uptake keeps increasing after that due to the

47
hydrolytic degradation of the sizing-rich fiber matrix interphase. If the water uptake experiments are
stopped as suggested by ASTM, then the long-term behavior is not captured.
A phenomenological mass balance model was developed and successfully used in order to obtain
the dissolution kinetics of the sizing-rich composite interphase.
The thesis provides a better understanding of environmental aging mechanisms of the constituents
in fiber-reinforced composites and its effect on mechanical properties of such materials. Based on
environmental aging experiments, novel analytical models were proposed and developed in order to
predict changes in properties of the composite microconstituents due to exposure to such
environments. These practical tools, i.e. models and methods, were provided for quantitative
prediction of water-induced changes in the microconstituent materials and composites. The tools are
to be of assistance in partially substituting the rigorous physical testing procedures in the state-of-the-
art situation. Prediction of long-term properties of composites should significantly reduce costs
associated with extensive testing and should already allow a partial transition towards the multiscale
modeling approach.

48
CHAPTER 5

FUTURE WORK

This thesis addresses the problem of environmental aging of composites and their constituents.
Nevertheless, there are related aspects and research topics that were only touched upon or not fully
investigated. For industry a couple of aspects would be of high interest:

x The application and validation of the proposed methodology to study environmental aging of
composites with different matrix, fiber and sizing materials.
x The application and validation of the proposed methodology to study aging via exposure to
oil or composite pipe and riser rinsing with methanol.

The questions for further work include:

x Water-induced and temperature effects on the viscoelastic behavior, i.e. creep, of matrix
polymers and their prediction in composites.

x The developed true water determination method described in Paper I (Appendix A) is

potentially useful also for other polymers and composite systems with different fiber or resin
fractions, as well as for diffusant media other than water. However, there is one significant
drawback of the method at the moment that it has its limitations for thick and non-transparent
composite samples. Thus, a combination of methods or the improvement of the current
method is recommended. Other spectroscopic methods such as Raman, while not as sensitive
to water, and reflectance FTIR, while mostly providing information about the surface of the
material, might also be considered in developing water monitoring methods, especially in
cases when composites are non-transparent to the IR light.

x Other common epoxies and matrix polymers should be studied using a similar approach to
Paper II (Appendix B) also shown in Figure 5.1. This should be done in order to see whether
there are other aging mechanisms in other matrix polymers, i.e. hydrolysis, as expected for
example, for the anhydride-based epoxy. The proposed approach which was followed in
Paper II is shown schematically in Figure 5.1.

x Other common matrix polymers should be studied analogously as in Paper III (Appendix C).
Similar research would be of high interest to thermoplastic-based composites, e.g. using highly
crosslinked polypropylene (HXPP) or polyamides (PA6, PA11, PA12). The methodology
described here should be applicable to other matrix materials. It is expected that mechanical
properties of HXPP should be reversible, whereas polyamides are expected to age irreversibly
due to chemical degradation. Hydrolytic degradation kinetics of polyamides and the effect of
such aging on mechanical properties of these common thermoplastic matrix materials was
studied in detail by Mazan et al [161,162]. They have reported that the hydrolysis induced
chain scission and chemicrystalization were the two main mechanisms of property change
[162]. Furthermore, they were able to model mechanical deterioration behavior of PA11. For
the studied epoxy it was shown experimentally that the tension fatigue S-N curve of a wet
epoxy resin can be estimated by shifting the S-N curve of a dry material proportionally to a
reduction in static tensile strength due to hygrothermal effects. This observation and its

49
 universality should be investigated also for other polymers that do not degrade chemically,
e.g. not affected by hydrolysis or chain scission.

x The imperfections of the fiber–matrix interface are neglected in the analytical model for
hygroscopic swelling prediction for composites, described in a Paper IV (Appendix D). A good
agreement between experimental, numerical and analytical results indicated that this effect is
negligible for the material system studied. For a more advanced model, there should be
additional studies concerning this aspect.

Figure 5.1. Schematic representation of the logic during investigation of the changes in the chemical
structure of the polymer due to aging.

x Aging of R-glass fibers was studied in this work. However, there should be no limitations to
apply the Dissolving Cylinder Zero-Order Kinetics (DCZOK) model to other types of glass
fibers. Validation of the model with various types of glass fibers such as E, ECR, S is advised.
The model should be applicable to other types of glass since SiO2 is the major component in
virtually all types of glass [57], but it would be beneficial to validate this model experimentally
with other types of glass fibers.

x The influence of pH, temperature and stress corrosion on the rate constants and dissolution
activation energies was obtained in Paper VII (Appendix G), however influence of each
parameter was studied one at a time. Thus, a suggestion for future work includes a cross-
parametric study for pH, temperature and stress corrosion in order to deduce a general
analytical solution for the environmental influence on the activation energy of dissolution, and
to study whether there is any coupled effect. The DCZOK model should also be extendable to
include the effect of ionic strength.

x Validating the DCZOK model for seawater conditions (about 1.84 – 12.62 mg SiO2/kg water;
pH of seawater 7.8) would be highly beneficial, especially for the marine and offshore
industries, since the real-life structures most often operate in the seawater environment. When
GFs are used in seawater, the dissolution of glass occurs slower due to the presence of silica in
the seawater, which is in seawater from the contact with sand and minerals [163-166]. The

50
 approach in distilled water is conservative with regards to seawater, meaning that structures
implementing glass fibers designed for distilled water conditions should not encounter
penalties to their service time in the seawater.

x Silica degradation products are large and are likely unable to escape thick composite
structures. It is recommended to perform diffusion studies and determine diffusivity of the
degradation products inside the composites. In order to study the transport of SiO 2 through
the matrix, a permeability experiment using the polymer membrane should be performed,
where on one side of the membrane would be SiO2-saturated water, while on the other side,
there would be distilled SiO2-free water. This would allow to determine the permeability and
diffusivity of the silica degradation products in the polymer material used as a matrix in the
composite. Furthermore, tests with fiber bundles not embedded in the composite immersed in
the SiO2-saturated water should be performed to see whether the fiber dissolution stops
completely. Furthermore, the thickness of the composite may have an influence on the
accumulation of the degradation products inside the composites. It should be harder for
degradation products to leave the thicker composite. This would mean that the effect of
degradation product accumulation should be better protecting the thicker GFRP structures
from glass dissolution. Additional research is needed to test this hypothesis.

x It would be interesting to combine the fiber dissolution study with a fiber strength study to
obtain some indication of how fiber mass loss affects fiber (and composite) performance. By
modeling the dissolution kinetics of glass, it should be possible to predict the long-term
deterioration of mechanical properties of the glass fibers due to hydrolytic crack growth.
Micromechanical models to predict the crack growth of the glass fibers have been previously
proposed by Wierderhorn and Bolz [167] for stress-corrosion and Sekine and Beaumont [168]
for glass corrosion in acids. However, to the best knowledge, the link from the dissolution
chemistry-based kinetic models to the crack growth models has not been done yet, even
though it has been proposed by Charles back in 1958 [169]. The DCZOK model is seen as a
potential chemical kinetics component for modeling such hydrolytic crack growth in glass
fibers. This aspect should be further investigated.

51
52
ABBREVIATIONS & SYMBOLS

Abbreviations:

ATR-FT-IR Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy

BET Brunauer-Emmett-Teller theory
C1 Composite plate with fibers oriented parallel to the plate face
C3 Composite plate with fibers oriented normal to the plate face
DCZOK Dissolving Cylinder Zero-Order Kinetic model
DGEBA Bisphenol A diglycidyl ether
DMTA Dynamic Mechanical Testing Analysis
E - glass “Electrical” glass
ECR - glass “Electrical/Chemical Resistance” glass
EDX Energy-Dispersive X-ray spectroscopy
FE Finite Element
FEA Finite Element Analysis
FRP Fiber-Reinforced Polymer, same as fiber-reinforced composite
FT-NIR Fourier Transform Near Infrared spectroscopy
GF Glass Fiber
GFRP Glass Fiber-Reinforced Polymer, same as Glass Fiber-Reinforced Composite
HDDGE 1,6-Hexanediol diglycidyl ether
HR-ICP-MS High Resolution Inductively Coupled Plasma Mass Spectrometry
IPDA Isophorondiamine
LOI Loss On Ignition
PEO Poly(ethylene oxide)
PDMS Polydimethylsiloxane
POPA Poly(oxypropylene)diamine
PPO Poly(propylene oxide)
R - glass “Reinforcement” glass
RVE Representative Volume Element
S - glass “Strength” glass
SEM Scanning Electron Microscopy
VARTM Vacuum-Assisted Resin Transfer Molding

53
Symbols:

ࢼ࢓ Hygroscopic swelling coefficient for matrix polymer (-)

ࢼ࢞ Hygroscopic swelling coefficient for composite in along-the-fiber direction (-)
ࢼ࢟ ǡ ࢼࢠ Hygroscopic swelling coefficient for composite in transverse directions (-)
ࢿ Linear strain (%)
ࢿࢎ Linear hygroscopic strain (%)
ࢿ࢓ Linear strain of the polymer matrix (%)
ࢿࢌ Linear strain of the fibers (%)
ࢿ࢓ࢇ࢞ Linear strain to failure (%)
ࢿ࢞ Linear strain for composite in along-the-fiber direction (%)
ࢿ࢟ ǡ ࢿࢠ Linear strain for composite in transverse-to-fiber directions (%)
ࢾ࢏ Thickness of the composite interphase (m)
ࣇࢌ Poisson’s ratio of glass fibers (-) in: Paper IV
ࣇࢌ Volume fraction of the fibers (m3/m3)
ࣇ࢓ Poisson’s ratio of the matrix polymer (-) in: Papers III & IV
ࣇ࢓ Volume fraction of the matrix polymer (m3/m3)
ࣇ࢏ Volume fraction of the composite interphase (m3/m3)
ࣇ࢜ Volume fraction of the voids (m3/m3)
࣋ࢉ࢕࢓࢖࢕࢙࢏࢚ࢋ Density of the composite (g/m3)
࣋ࢌ Density of the glass fibers (g/m3)
࣋ࢍ࢒ࢇ࢙࢙ Density of the glass (g/m3), same as ࣋ࢌ
࣋࢏ Density of the sizing-rich composite interphase (g/m3)
࣋࢓ Density of the matrix polymer (g/m3)
࣋࢝ࢇ࢚ࢋ࢘ Density of the water (g/m3)
ࣈ࢙࢏ࢠ࢏࢔ࢍ Protective effect of sizing against glass dissolution (-)
࣌ Stress (MPa)
࣌ Stress in glass fibers (MPa)
ࢌ

࣌࢓ Stress in the matrix polymer (MPa)

࡭ Pre-exponential factor (g/(m2·s))
࡭ Cubic cell lateral surface area (-) in: Paper IV
࡭ࢌ Fiber part of the cubic cell lateral surface area (-)
࡭࢓ Matrix part of the cubic cell lateral surface area (-)
ࢉ Non-cumulative ion mass concentration measured with HR-ICP-MS (g/L)
CHE Coefficient of Hygroscopic Expansion (-), same as ࢼ
ࡰ Through-thickness water diffusivity of the material (mm2/h)
ࡰ࢒ࢋࢇࢉࢎ࢏࢔ࢍ Through-thickness leachable compound diffusivity of the material (mm2/h)
ࡱ Young’s modulus (MPa)
ࡱ࡭ Dissolution activation energy (J/mol)
ࡱࢌ Stiffness of glass fibers (MPa)

54
ࡱ࢓ Stiffness of the matrix polymer (MPa)
ࢎ Thickness of a material plate (m)
ࡷ૙ Glass dissolution rate constant (g/(m2·s))
ࡷࡵ૙ Glass dissolution rate constant (non-steady-state; Phase I) (g/(m2·s))
ࡷࡵࡵ
૙ Glass dissolution rate constant (steady-state; Phase II) (g/(m2·s))
ࡷࡵ૙ࡿ࢏ Dissolution rate constant for Si only (non-steady-state) (g/(m2·s))
ࡷࡵࡵ
૙ࡿ࢏ Dissolution rate constant for Si only (steady-state) (g/(m2·s))
ࡷࡵ૙࢚࢕࢚ࢇ࢒ Dissolution rate constant for glass (non-steady state) (g/(m2·s))
ࡷࡵࡵ
૙࢚࢕࢚ࢇ࢒ Dissolution rate constant for glass (steady-state) (g/(m2·s))
ࡷ‫כ‬૙ Apparent glass dissolution rate constant (g/(m2·s))
ࡷ‫ࡵכ‬
૙ Apparent glass dissolution rate constant (non-steady-state; Phase I) (g/(m2·s))
ࡷ‫ࡵࡵכ‬
૙ Apparent glass dissolution rate constant (steady-state; Phase II) (g/(m2·s))
ࡷ૙࢏ Zero-order rate constant of the composite interphase dissolution (g/(m2·s))
࢒ Length of fibers and the interphase (m)
ࡸ࢞ ǡ ࡸ࢟ ǡ ࡸࢠ Dimensions of the RVE (-)
࢓ Mass of a polymer including moisture (g) in: Paper I
࢓; ࢓ࢊ࢏࢙࢙࢕࢒࢜ࢋࢊ Glass mass loss due to dissolution (g)
࢓ࢊ࢏࢙࢙࢕࢒࢜ࢋࢊ࢚࢙࢚ Dissolved glass mass when the steady-state is reached (g)
࢓ࢇ࢈࢙࢕࢒࢛࢚ࢋ࢒࢟ࢊ࢘࢟ Mass of an absolutely dry specimen (0 wt% water content) (g)
࢓ࢌ Mass fraction of fibers in a composite laminate (g/g)
࢓࢏ Mass of the composite interphase (g)
࢓࢏ ૙ Initial mass of the composite interphase (g)
࢓࢕࢘ࢊࢋ࢘ Order of the degradation product accumulation (-)
࢓ࡿ࢏ Si mass loss due to dissolution (g)
࢓࢝ࢇ࢚ࢋ࢘ Mass of moisture uptaken by the specimen (g)
ࡹ Water content of the composite (wt%)
ࡹஶ Saturation water content of the composite (wt%)
࢓
ࡹ Water content of the matrix polymer (wt%)
ࡹ࢓
ஶ Saturation water content of the matrix polymer (wt%)
ࡹ࢜ஶ Saturation water content of the voids (wt%)
ࡹ࢒ࢋࢇࢉࢎ࢏࢔ࢍ Content of leached compounds from the polymer (wt%)
ࡹ૙࢒ࢋࢇࢉࢎ࢏࢔ࢍ Initial leachable compound content in the polymer (wt%)
࢔ Number of fibers (-)
࢔࢕࢘ࢊࢋ࢘ Order of the water availability term (-)
ࡺ Number of cycles before specimen fails in fatigue (-)
࢖ࡴ Acidity of the environment (-)
࢘ Fiber radius (m)
࢘૙ Initial fiber radius (m)
࢚࢙࢚࢘ Fiber radius when the steady-state dissolution is reached (m)
ࡾ Universal gas constant (8.314 J/(mol·K))
ࡿ Stress (MPa)

55
ࡿ Glass fiber surface area (m2)
ࡿ૙ Initial glass fiber surface area (m2)
࢙࢖ࢋࢉ࢏ࢌ࢏ࢉ
ࡿ૙ Specific glass fiber surface area (m2/g)
ࡿࢍ࢒ࢇ࢙࢙ Total glass fiber surface area in a composite plate (m2)
ࡿ࢏ Surface area of the composite interphase (m2)
ࡿ࢏ ૙ Initial surface area of the composite interphase (m2)
࢙࢖ࢋࢉ࢏ࢌ࢏ࢉ
ࡿ࢏ Specific surface area of the composite interphase (m2/g)
࢚ Time (s)
࢚࢙࢚ Time when steady-state dissolution is reached (s)
ࢀ Absolute temperature (K)
ࢀࢍ Glass transition temperature (K)
࢛࢞ ǡ ࢛࢟ ǡ ࢛ࢠ Relative displacements in Finite Element simulations (-)
ࢁࢀࡿ Ultimate Tensile Strength (MPa)
ࢂࢌ Volume fraction of fibers in a composite laminate (m3/m3)
ࢂ࢝ࢇ࢚ࢋ࢘ Volume of a water sample used for the HR-ICP-MS (L)
ࢃ Gravimetric water content (wt%) in: Paper I
‫כ‬
ࢃǢ ࢃ True water content (wt%)
ࢃࢉ Moisture content in the composite (wt%)
ࢃ࢓ Moisture content scaled by polymer matrix fraction of the composite (wt%)
࢞ǡ ࢟ǡ ࢠ Cartesian coordinate system

56
BIBLIOGRAPHY

1. Berg, J.; Jones, F.R. The role of sizing resins, coupling agents and their blends on the formation of the
interphase in glass fiber composites. Composites Part A 1998, 29, 1261–1272, doi:10.1016/S1359-835X(98)00091-
8.
2. Feih, S.; Wei, J.; Kingshott, P.; Sørensen, B.F. The influence of fiber sizing on the strength and fracture
toughness of glass fiber composites. Composites Part A 2005, 36, 245–255,
doi:10.1016/j.compositesa.2004.06.019.
3. Krauklis, A.; Echtermeyer, A. Mechanism of Yellowing: Carbonyl Formation during Hygrothermal Aging in
a Common Amine Epoxy. Polymers 2018, 10, 1017–1031, doi:10.3390/polym10091017.
4. Xiao, G.Z.; Shanahan, M.E.R. Swelling of DGEBA/DDA epoxy resin during hygrothermal ageing. Polymer
1998, 39, 3253–3260, doi:10.1016/S0032-3861(97)10060-X.
5. Toscano, A.; Pitarresi, G.; Scafidi, M.; Di Filippo, M.; Spadaro, G.; Alessi, S. Water diffusion and swelling
stresses in highly crosslinked epoxy matrices. Polym. Degrad. Stab. 2016, 133, 255–263,
doi:10.1016/j.polymdegradstab.2016.09.004.
6. Grabovac, I.; Whittaker, D. Application of bonded composites in the repair of ships structures—A 15-year
service experience. Compos. Part A 2009, 40, 1381–1398, doi:10.1016/j.compositesa.2008.11.006.
7. McGeorge, D.; Echtermeyer, A.T.; Leong, K.H.; Melve, B.; Robinson, M.; Fischer, K.P. Repair of floating
offshore units using bonded fibre composite materials. Compos. Part A 2009, 40, 1364–1380,
doi:10.1016/j.compositesa.2009.01.015.
8. Gustafson, C.G.; Echtermeyer, A. Long-term properties of carbon fibre composite tethers. Int. J. Fatigue. 2006,
28, 1353–1362, doi:10.1016/j.ijfatigue.2006.02.035.
9. Salama, M.M.; Stjern, G.; Storhaug, T.; Spencer, B.; Echtermeyer, A. The First Offshore Field Installation for a
Composite Riser Joint.; OTC-14018-MS; Offshore Technology Conference: Houston, TX, USA, 2002,
doi:10.4043/14018-MS.
10. Echtermeyer, A.T.; Gagani, A.I.; Krauklis, A.E.; Mazan, T. Multiscale Modelling of Environmental
Degradation—First Steps. In Durability of Composites in a Marine Environment 2. Solid Mechanics and its
Applications, Davies, P., Rajapakse, Y.D.S., Eds.; Springer: Cham, Switzerland, 2018; Volume 245, pp. 135–149,
ISBN: 978-3-319-65145-3.
11. Thomason, J.L. Glass Fiber Sizings: A Review of the Scientific Literature. Middletown, DE, USA, 2012, ISBN:978-
0-9573814-1-4.
12. Weitsman, Y. Coupled damage and moisture-transport in fiber-reinforced, polymeric composites. Int. J.
Solids Struct. 1987, 23(7), 1003-1025, doi:10.1016/0020-7683(87)90093-X.
13. Weitsman, Y.J.; Elahi, M. Effects of fluids on the deformation, strength and durability of polymeric
composites—An overview. Mech. Time-Depend. Mater. 2000, 4, 107–126, doi:10.1023/A:1009838128526.
14. Roy, S. Moisture-Induced Degradation. In Long-Term Durability of Polymeric Matrix Composites; Pochiraju,
V.K.; Tandon, P.G.; Schoppner, A.G., Eds.; Springer: Boston, MA, USA, 2012; pp 181-236, ISBN:978-1-
4419-9307-6.
15. Apicella, A.; Nicolais, L. Effect of Water on the Properties of Epoxy Matrix and Composite. Adv. Polym. Sci.
1985, 72, 69-77, doi:10.1007/3-540-15546-5_3.
16. Lefebvre, D.R.; Elliker, P.R.; Takahashi, K.M.; Raju, V.R.; Kaplan, M.L. The Critical Humidity Effect in the
Adhesion of Epoxy to Glass: Role of Hydrogen Bonding. J. Adhes. Sci. Technol. 2000, 14, 925-937,
doi:10.1163/156856100742988.
17. Guermazi, N.; Elleuch, K.; Ayedi, H.F. The Effect of Time and Aging Temperature on Structural and
Mechanical Properties of Pipeline Coating. Mater. Des. 2009, 30, 2006-2010, doi:10.1016/j.matdes.2008.09.003.
18. Wu, C.F.; Xu, W.J. Atomistic Simulation Study of Absorbed Water Influence on Structure and Properties of
Crosslinked Epoxy Resin. Polymer 2007, 48, 5440-5448, doi:10.1016/j.polymer.2007.06.038.
19. DiBenedetto, A.T. Tailoring of interfaces in glass fiber reinforced polymer composites: a review. Mater. Sci.
Eng. A 2001, 302, 74-82, doi:10.1016/S0921-5093(00)01357-5.
20. Wang, M.; Xu, X.; Ji, J.; Yang, Y.; Shen, J.; Ye, M. The hygrothermal aging process and mechanism of the
novolac epoxy resin. Composites Part B 2016, 107, 1-8, doi:10.1016/j.compositesb.2016.09.067.

57
21. Halpin, J.C. Effects of Environmental Factors on Composite Materials; Technical report AFML-TR-67–423, Air
Force Materials Laboratory: Dayton, OH, USA, 1969.
22. Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Long-Term Hydrolytic Degradation of the Sizing-Rich
Composite Interphase. Coatings 2019, 9(4), 263-286, doi:10.3390/coatings9040263.
23. Gagani, A.I.; Fan, Y.; Muliana, A.H.; Echtermeyer, A.T. Micromechanical modeling of anisotropic water
diffusion in glass fiber epoxy reinforced composites. J. Compos. Mater. 2017, 52(17), 2321-2335,
doi:10.1177/0021998317744649.
24. Krauklis, A.E.; Echtermeyer, A.T. Dissolving Cylinder Zero-Order Kinetic Model for Predicting
Hygrothermal Aging of Glass Fiber Bundles and Fiber-Reinforced Composites. In 4th International Glass Fiber
Symposium; Gries, Th.; Pico, D.; Lüking, A.; Becker, Th., Eds.; Mainz, G (Verlag): Aachen, Germany, 2018; pp.
66–72, ISBN:978-3-95886-249-4.
25. Maggana, C.; Pissis, P. Water sorption and diffusion studies in an epoxy resin system. J. Polym. Sci. Part B
1999, 37(11), 1165–1182, doi:10.1002/(SICI)1099-0488(19990601)37:11<1165::AID-POLB11>3.0.CO;2-E.
26. Lee, M.C.; Peppas, N.A. Water transport in epoxy-resins. Prog. Polym. Sci. 1993, 18(5), 947–961,
doi:10.1016/0079-6700(93)90022-5.
27. Popineau, S.; Rondeau-Mouro, C.; Sulpice-Gaillet, C.; Shanahan, M.E.R. Free/bound water absorption in an
epoxy adhesive. Polymer 2005, 46(24), 10733–10740, doi:10.1016/j.polymer.2005.09.008.
28. Tennent, N.H. Clear and Pigmented Epoxy Resins for Stained Glass Conservation: Light Ageing Studies. Stud.
Conserv. 1979, 24, 153–164, doi:10.2307/1505777.
29. Ernault, E.; Richaud, E.; Fayolle, B. Origin of epoxies embrittlement during oxidative ageing. Polym. Test.
2017, 63, 448–454, doi:10.1016/j.polymertesting.2017.09.004.
30. Pham, H.Q.; Marks, M.J. Epoxy resins. In Ullmann’s encyclopedia of industrial chemistry; 2005, Wiley-VCH. doi:
10.1002/14356007.a09_547.pub2, ISBN:9783527303854.
31. Mailhot, B.; Morlat-Thérias, S.; Ouahioune, M.; Gardette, J.-L. Study of the Degradation of an Epoxy/Amine
Resin, 1 Photo- and Thermo-Chemical Mechanisms. Macromol. Chem. Phys. 2005, 206, 575–584,
doi:10.1002/macp.200400395.
32. Rocha, I.B.C.M.; Raijmaekers, S.; Nijssen, R.P.L.; van der Meer, F.P.; Sluys, L.J. Hygrothermal ageing
behaviour of a glass/epoxy composite used in wind turbine blades. J. Compos. Struct. 2017, 174, 110–122,
doi:10.1016/j.compstruct.2017.04.028.
33. Startsev, V.O.; Lebedev, M.P.; Khrulev, K.A.; Molokov, M.V.; Frolov, A.S.; Nizina, T.A. Effect of outdoor
exposure on the moisture diffusion and mechanical properties of epoxy polymers. Polym. Test. 2018, 65, 281–
296, doi:10.1016/j.polymertesting.2017.12.007.
34. Li, L.; Yu, Y.; Wu, Q.; Zhan, G.; Li, S. Effect of Chemical Structure on the Water Sorption of Amine-Cured
Epoxy Resins. Corros. Sci. 2009, 51, 3000-3006, doi:10.1016/j.corsci.2009.08.029.
35. Wang, J.; Gong, J.; Gong, Z.; Yan, X.; Wang, B.; Wu, Q.; Li, S. Effect of Curing Agent Polarity on Water
Absorption and Free Volume in Epoxy Resin Studied by PALS. Nucl. Instrum. Methods Phys. Res. B 2010, 268,
2355-2361, doi:10.1016/j.nimb.2010.04.010.
36. Morel, E.; Bellenger, V.; Verdu, J. Structure-Water Absorption Relationships for Amine-Cured Epoxy Resins.
Polymer 1985, 26, 1719-1724, doi:10.1016/0032-3861(85)90292-7.
37. Clancy, T.C.; Frankland, S.J.V.; Hinkley, J.A.; Gates, T.S. Molecular modeling for calculation of mechanical
properties of epoxies with moisture ingress. Polymer 2009, 50(12), 2736–2742,
doi:10.1016/j.polymer.2009.04.021.
38. Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Hygrothermal Aging of Amine Epoxy: Reversible Static and
Fatigue Properties. Open Eng. 2018, 8, 447–454, doi:10.1515/eng-2018-0050.
39. Ibarra, L.; Chamorro, C. Short fiber–elastomer composites. Effects of matrix and fiber level on swelling and
mechanical and dynamic properties. J. Appl. Polym. Sci. 1991, 43, 1805–1819, doi:10.1002/app.1991.070431004.
40. Chen, Y.; Davalos, J.F.; Ray, I.; Kim, H.-Y. Accelerated aging tests for evaluations of durability performance
of FRP reinforcing bars for concrete structures. Compos. Struct. 2007, 78(1), 101–111,
doi:10.1016/j.compstruct.2005.08.015.
41. Wallenberger, F.T. Commercial and Experimental Glass Fibres. In Fibreglass and Glass Technology;
Wallenberger, F.T.; Bingham, P.A., Eds.; Springer: US, 2010; ISBN:978-1-4419-0735-6.

58
42. Steinmann, W.; Saelhoff, A.-K. Essential Properties of Fibres for Composite Applications. In Fibrous and Textile
Materials for Composite Applications, Textile Science and Clothing Technology; Rana, S.; Fangueiro, R., Eds.;
Springer: Singapore, 2016; ISBN:978-981-10-0232-8.
43. Krauklis, A.E.; Gagani, A.I.; Vegere, K.; Kalnina, I.; Klavins, M.; Echtermeyer, A.T. Dissolution Kinetics of R-
Glass Fibres: Influence of Water Acidity, Temperature and Stress Corrosion. Fibers 2019, 7(3), 22-40, in a Special
Issue: Advances in Glass Fibers, doi:10.3390/fib7030022.
44. Brown, E.N.; Davis, A.K.; Jonnalagadda, K.D.; Sottos, N.R. Effect of surface treatment on the hydrolytic
stability of E-glass fiber bundle tensile strength. Compos. Sci. Technol. 2005, 65, 129–136,
doi:10.1016/j.compscitech.2004.07.001.
45. Bledzki, A.; Spaude, R.; Ehrenstein, G.W. Corrosion Phenomena in Glass Fibres and Glass Fibre Reinforced
Thermosetting Resins. Compos. Sci. Technol. 1985, 23(4), 263-285, doi:10.1016/0266-3538(85)90040-5.
46. Tournié, A.; Ricciardi, P.; Colomban, P. Glass Corrosion Mechanisms: A Multiscale Analysis. Solid State Ionics
2008, 179(38), 2142-2154, doi:10.1016/j.ssi.2008.07.019.
47. Agarwal, B.D.; Broutman, L.J. Analysis and Performance of Fibre Composites, 2nd ed.; John Wiley and Sons, Inc.:
Hoboken, USA, 1990; pp. 339-359, ISBN:978-0-471-51152-6.
48. Schutte, C.L. Environmental durability of glass-fiber composites. Mater. Sci. Eng. R Reports 1994, 13, 265-323,
doi:10.1016/0927-796X(94)90002-7.
49. Renaud, C.M.; Greenwood, M.E. Effect of glass fibers and environments on long-term durability of GFRP
composites, In Proceedings of 9 EFUC Meeting, Wroclaw, Poland, 2005.
50. Grambow, B.; Müller, R. First-order dissolution rate law and the role of surface layers in glass performance
assessment. J. Nucl. Mater. 2001, 298(1-2), 112-124. doi:10.1016/S0022-3115(01)00619-5.
51. Grambow, B. A General Rate Equation for Nuclear Waste Glass Corrosion. Mat. Res. Soc. Symp. Proc. 1985, 44,
15-27, doi:10.1557/PROC-44-15.
52. Krauklis, A.E.; Echtermeyer, A.T. Long-Term Dissolution of Glass Fibers in Water Described by Dissolving
Cylinder Zero-Order Kinetic Model: Mass Loss and Radius Reduction. Open Chem. 2018, 16(1), 1189-1199,
doi:10.1515/chem-2018-0133.
53. Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Near-Infrared Spectroscopic Method for Monitoring Water
Content in Epoxy Resins and Fiber-Reinforced Composites. Materials 2018, 11(4), 586-599,
doi:10.3390/ma11040586.
54. Stamenović, M.R.; Putić, S.S.; Rakin, M.B.; Medjo, B.; Čikara, D. Effect of alkaline and acidic solutions on the
tensile properties of glass–polyester pipes. Mater. Des. 2011, 32(4), 2456-2461,
doi:10.1016/j.matdes.2010.11.023.
55. Amaro, A.M.; Reis, P.N.B.; Neto, M.A.; Louro C. Effects of alkaline and acid solutions on glass/epoxy
composites. Polym. Degrad. Stab. 2013, 98(4), 853-862, doi:10.1016/j.polymdegradstab.2012.12.029.
56. Mišíková, L.; Liška, M.; Galusková, D. CORROSION OF E-GLASS FIBRES IN DISTILLED WATER. Ceram.
Silikaty 2007, 51(3), 131-135.
57. Bashir, S.T.; Yang, L.; Liggat, J.J.; Thomason, J.L. Kinetics of dissolution of glass fibre in hot alkaline solution.
J. Mater. Sci. 2018, 53(3), 1710-1722. doi:10.1007/s10853-017-1627-z.
58. Delage, F.; Ghaleb, D.; Dussossoy, J.L.; Chevallier, O.; Vernaz, E. A mechanistic model for understanding
nuclear waste glass dissolution. J. Nucl. Mater. 1992, 190, 191-197, doi:10.1016/0022-3115(92)90086-Z.
59. Geisler-Wierwille, T.; Nagel, T.J.; Kilburn, M.R.; Janssen, A.; Icenhower, J.; Fonseca, R.O.C.; Grange, M.L.;
Nemchin, A.A. The Mechanism of Borosilicate Glass Corrosion Revisited. Geochim. Cosmochim. Acta 2015, 158,
112-129, doi:10.1016/j.gca.2015.02.039.
60. Icenhower, J.; Steefel, C.I. Dissolution Rate of Borosilicate Glass SON68: A Method of Quantification Based
upon Interferometry and Implications for Experimental and Natural Weathering Rates of Glass. Geochim.
Cosmochim. Acta 2015, 157, 147-163, doi:10.1016/j.gca.2015.02.037.
61. Ma, T.; Jivkov, A.P.; Li, W.; Liang, W.; Wang, Y.; Xu, H.; Han, X. A mechanistic model for long-term nuclear
waste glass dissolution integrating chemical affinity and interfacial diffusion barrier. J. Nucl. Mater. 2017, 486,
70-85, doi:10.1016/j.jnucmat.2017.01.001.
62. Geisler, T.; Dohmen, L.; Lenting, C.; Fritzsche, M.B.K. Real-time in situ observations of reaction and transport
phenomena during silicate glass corrosion by fluid-cell Raman spectroscopy. Nature Materials 2019, 18, 342-
348, doi:10.1038/s41563-019-0293-8.

59
63. Hunter, F.M.I.; Hoch, A.R.; Heath, T.G.; Baston, G.M.N. Report RWM005105, AMEC/103498/02 Issue 2: Review
of Glass Dissolution Models and Application to UK Glasses; AMEC: Didcot, Oxfordshire, UK, 2015.
64. Michalske, T.A.; Freiman, S.W. A Molecular Mechanism for Stress Corrosion in Vitreous Silica. J. Am. Ceram.
Soc. 1983, 66, 284–288, doi:10.1111/j.1151-2916.1983.tb15715.x.
65. Iler, R.K. The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface Properties and Biochemistry of Silica;
Wiley: New York, USA, 1979; pp. 896, ISBN:978-0-471-02404-0.
66. Putnis, C.V.; Ruiz-Agudo, E. The mineral‒water interface: where minerals react with the environment.
Elements 2013, 9, 177–182, doi:10.2113/gselements.9.3.177.
67. Dai, Z.; Shi, F.; Zhang, B.; Li, M.; Zhang, Z. Effect of sizing on carbon fiber surface properties and fibers/epoxy
interfacial adhesion. Appl. Surf. Sci. 2011, 257, 6980–6985, doi:10.1016/j.apsusc.2011.03.047.
68. Yuan, X.; Zhu, B.; Cai, X.; Liu, J.; Qiao, K.; Yu, J. Optimization of interfacial properties of carbon fiber/epoxy
composites via a modified polyacrylate emulsion sizing. Appl. Surf. Sci. 2017, 401, 414-423,
doi:10.1016/j.apsusc.2016.12.234.
69. Plonka, R.; Mäder, E.; Gao, S.L.; Bellmann, C.; Dutschk, V.; Zhandarov, S. Adhesion of epoxy/glass fiber
composites influenced by aging effects on sizings. Composites Part A 2004, 35, 1207–1216,
doi:10.1016/j.compositesa.2004.03.005.
70. Peters, L. Influence of Glass Fibre Sizing and Storage Conditions on Composite Properties. In Durability of
Composites in a Marine Environment 2. Solid Mechanics and Its Applications; Davies, P.; Rajapakse, Y.D.S., Eds.;
Springer: Cham, Switzerland, 2018; Volume 245, pp. 19-31, ISBN:978-3-319-65145-3.
71. Culler, S.R.; Ishida, H.; Koenig, J.L. Hydrothermal stability of γ-Aminopropyltriethoxysilane Coupling Agent on
Ground Silicon Powder and E-Glass Fibers; Technical report, Department of Macromolecular Science: Cleveland,
OH, USA, 1983.
72. Wang, D.; Jones, F.R.; Denison, P. TOF SIMS and XPS study of the interaction of hydrolysed γ-
aminopropyltriethoxysilane with E-glass surfaces. J. Adhes. Sci. Technol. 1992, 6, 79-98,
doi:10.1163/156856192X00070.
73. Wang, D.; Jones, F.R.; Denison, P. Surface analytical study of the interaction between γ-amino propyl
triethoxysilane and E-glass surface. Part I Time-of-flight secondary ion mass spectrometry. J. Mater. Sci. 1992,
27, 36-48, doi:10.1007/BF00553834.
74. Wang, D.; Jones, F.R. Surface analytical study of the interaction between γ-amino propyl triethoxysilane and
E-glass surface. Part II X-ray photoelectron spectroscopy. J. Mater. Sci. 1993, 28, 2481-2488,
doi:10.1007/BF01151683.
75. ASTM D4963/D4963M-2011. Standard Test Method for Ignition Loss of Glass Strands and Fabrics, 2011.
76. Piret, W.; Mason, N.; Luc, P. European Patent EP2540683A1, Glass fibre sizing composition by 3B-Fibreglass
SPRL, 2011.
77. Thomason, J.L. Glass Fibre Sizing: A Review of Size Formulation Patents. Blurb Co: Glasgow, Scotland, UK, 2015;
ISBN:978-0-9573814-3-8.
78. Loewenstein, K.L. Glass science and technology (Book 6), The manufacturing technology of continuous glass fibres.
Elsevier: Amsterdam, Netherlands, 1993, ISBN:978-0444893468.
79. Thomason, J.L.; Adzima, L.J. Sizing up the interphase: an insider’s guide to the science of sizing. Composites
Part A 2001, 32, 313-321, doi:10.1016/S1359-835X(00)00124-X.
80. Plueddemann, E.P. Silane coupling agents, 2nd edition. Plenum Press: New York and London, 1991; ISBN:978-
0-306-43473-0.
81. Wolff, V.; Perwuelz, A.; El Achari, A.; Caze, C.; Carlier, E. Determination of surface heterogeneity by contact
angle measurements on glass fibres coated with different sizings. J. Mater. Sci. 1999, 34, 3821-3829,
doi:10.1023/A:1004604917226.
82. Watson, H.; Mikkola, P.J.; Matisons, J.G.; Rosenholm, J.B. Deposition characteristics of ureido silane ethanol
solutions onto E-glass fibres. Colloids Surf. A 2000, 161, 183-192, doi:10.1016/S0927-7757(99)00336-2.
83. Feresenbet, E.; Raghavan, D.; Holmes, G.A. Influence of silane coupling agent composition on the surface
characterization of fiber and on fiber-matrix interfacial shear strength. J. Adhes. 2003, 79, 643-665,
doi:10.1080/00218460309580.
84. Fagerholm, H.M.; Lindsjö, C.; Rosenholm, J.B.; Rökman, K. Physical characterization of E-glass fibres treated
with alkylphenylpoly(oxyethylene)alcohol. Colloids Surf. 1992, 69, 79-86, doi:10.1016/0166-6622(92)80218-Q.

60
85. Thomason, J.L.; Dwight, D.W. The use of XPS for characterization of glass fibre coatings. Composites Part A
1999, 30, 1401-1413, doi:10.1016/S1359-835X(99)00042-1.
86. Turrión, S.G.; Olmos, D.; González-Benito, J. Complementary characterization by fluorescence and AFM of
polyaminosiloxane glass fibers coatings. Polym. Test. 2005, 24, 301-308,
doi:10.1016/j.polymertesting.2004.11.006.
87. Mai, K.; Mäder, E.; Mühle, M. Interphase characterization in composites with new non-destructive methods.
Composites Part A 1998, 29, 1111-1119, doi:10.1016/S1359-835X(98)00092-X.
88. Joliff, Y.; Belec, L.; Chailan, J.-F. Impact of the interphases on the durability of a composite in humid
environment – a short review. 20th International Conference on Composite Structures ICCS20. Paris, France, 2017.
89. Kim, J.K.; Sham, M.L.; Wu, J. Nanoscale characterization of interphase in silane treated glass fibre composites.
Composites Part A 2001, 32, 607-618, doi:10.1016/S1359-835X(00)00163-9.
90. Riaño, L.; Belec, L.; Chailan, J.-F.; Joliff, Y. Effect of interphase region on the elastic behavior of unidirectional
glass-fiber/epoxy composites. Compos. Struct. 2018, 198, 109-116, doi:10.1016/j.compstruct.2018.05.039.
91. Crank, J. The mathematics of diffusion, 2nd edition. Clarendon Press: Oxford, United Kingdom, 1975; ISBN:978-
0-19-853411-6.
92. Gagani, A.I.; Krauklis, A.E.; Echtermeyer, A.T. Anisotropic fluid diffusion in carbon fiber reinforced
composite rods: Experimental, analytical and numerical study. Marine Struct. 2018, 59, 47-59,
doi:10.1016/j.marstruc.2018.01.003.
93. Gagani, A.I.; Krauklis, A.E.; Echtermeyer, A.T. Orthotropic fluid diffusion in composite marine structures.
Experimental procedure, analytical and numerical modelling of plates, rods and pipes. Composites Part A 2018,
115, 196-205, doi:10.1016/j.compositesa.2018.09.026.
94. Bonniau, P.; Bunsell, A.R. Water Absorption by Glass Fibre Reinforced Epoxy Resin. In Composite Structures;
Marshall, I.H., Ed.; Springer: Dordrecht, Netherlands, 1981; pp. 92–105, ISBN:978-94-009-8122-5.
95. Augl, J.M.; Berger, A.E. The Effect of Moisture on Carbon Fiber Reinforced Epoxy Composites. 1. Diffusion; Technical
Report NSWC/WOL/TR-76-7, White Oak Laboratory: Silver Spring, Maryland, USA, 1976.
96. Bertolino, V.; Cavallaro, G.; Lazzara, G.; Merli, M.; Milioto, S.; Parisi, F.; Sciascia, L. Effect of the Biopolymer
Charge and the Nanoclay Morphology on Nanocomposite Materials. Ind. Eng. Chem. Res. 2016, 55, 7373-7380,
doi:10.1021/acs.iecr.6b01816.
97. Ciprioti, S.V.; Tuffi, R.; Dell’Era, A.; Poggetto, F.D.; Bollino, F. Thermal Behavior and Structural Study of
SiO2/Poly(ε-caprolactone) Hybrids Synthesized via Sol-Gel Method. Materials 2018, 11, 275-285,
doi:10.3390/ma11020275.
98. Mijović, J.; Zhang, H. Local Dynamics and Molecular Origin of Polymer Network−Water Interactions as
Studied by Broadband Dielectric Relaxation Spectroscopy, FTIR, and Molecular Simulations. Macromolecules
2003, 36, 1279-1288, doi:10.1021/ma021568q.
99. Philippe, L.V.S.; Lyon, S.B.; Sammon, C.; Yarwood, J. Validation of Electrochemical Impedance
Measurements for Water Sorption into Epoxy Coatings Using Gravimetry and Infra-Red Spectroscopy.
Corros. Sci. 2008, 50, 887-896, doi:10.1016/j.corsci.2007.09.008.
100. Cotugno, S.; Mensitieri, G.; Musto, P.; Sanguigno, L. Molecular Interactions in and Transport Properties of
Densely Crosslinked Networks: a Time-Resolved FT-IR Spectroscopy Investigation of the Epoxy/H2O System.
Macromolecules 2005, 38, 801-811, doi:10.1021/ma040008j.
101. Weir, M.D.; Bastide, C.; Sung, C.S.P. Characterization of Interaction of Water in Epoxy by UV Reflection
Spectroscopy. Macromolecules 2001, 34, 4923-4926, doi:10.1021/ma0017900.
102. Mijović, J.; Zhang, H. Molecular Dynamics Simulation Study of Motions and Interactions of Water in a
Polymer Network. J. Phys. Chem. B 2004, 108, 2557-2563, doi:10.1021/jp036181j.
103. MacQueen, R.C.; Granata, R.D. A Positron Annihilation Lifetime Spectroscopic Study of the Corrosion
Protective Properties of Epoxy Coatings. Prog. Org. Coat. 1996, 28, 97-112, doi:10.1016/0300-9440(95)00557-9.
104. Muroga, S.; Hikima, Y.; Ohshima, M. Near-Infrared Spectroscopic Evaluation of the Water Content of Molded
Polylactide Under the Effect of Crystallization. Appl. Spectrosc. 2017, 71, 1300-1309,
doi:10.1177/0003702816681011.
105. Genkawa, T.; Watari, M.; Nishii, T.; Ozaki, Y. Development of a Near-Infrared/Mid-Infrared Dual-Region
Spectrometer for Online Process Analysis. Appl. Spectrosc. 2012, 66, 773-781, doi:10.1366/11-06499.
106. Ozaki, Y. Near-Infrared Spectroscopy – Its Versatility in Analytical Chemistry. Anal. Sci. 2012, 28, 545-563,
doi:10.2116/analsci.28.545.

61
107. Zhou, J. Transient analysis on hygroscopic swelling characterization using sequentially coupled moisture
diffusion and hygroscopic stress modeling method. Microelectron. Reliab. 2008, 48, 805–810,
doi:10.1016/j.microrel.2008.03.027.
108. Hahn, H.T.; Kim, R.Y. Swelling of Composite Laminates. In Advanced Composite Materials: Environmental
Effects; ASTM STP 658; American Society for Testing and Materials: Philadelphia, PA, USA, 1978, pp. 98–120.
109. Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Prediction of Orthotropic Hygroscopic Swelling of Fiber-
Reinforced Composites from Isotropic Swelling of Matrix Polymer. Journal of Composites Science 2019, 3(1), 10-
23, doi:10.3390/JCS3010010.
110. Shirangi, M.H.; Michel, B. Mechanism of moisture diffusion, hygroscopic swelling, and adhesion degradation
in epoxy molding compounds. In Moisture Sensitivity of Plastic Packages of IC Devices, Fan, X.J.; Suhir, E., Eds.;
Springer, New York, NY, USA, 2010; pp. 29–69, ISBN:978-1-4419-5719-1.
111. Obeid, H.; Clément, A.; Fréour, S.; Jacquemin, F.; Casari, P. On the identification of the coefficient of moisture
expansion of polyamide-6: Accounting differential swelling strains and plasticization. Mech. Mater. 2018, 118,
1–10, doi:10.1016/j.mechmat.2017.12.002.
112. Cairns, D.S.; Adams, D.F. Moisture and thermal expansion properties of unidirectional composite materials
and the epoxy matrix. J. Reinf. Plast. Compos. 1983, 2, 239–255, doi:10.1177/073168448300200403.
113. Ashton, J.E.; Halpin, J.C.; Petit, P.H. Primer on Composite Materials Analysis; Technomic Pub. Co.: Stamford,
CT, USA, 1969; ISBN:978-0-8776-2754-8.
114. Schapery, R.A. Thermal expansion coefficients of composite materials based on energy principles. J. Compos.
Mater. 1968, 2, 380–404, doi:10.1177/002199836800200308.
115. Coran, A.Y.; Boustany, K.; Hamed, P. Unidirectional fiber-polymer composites: Swelling and modulus
anisotropy. J. Appl. Polym. Sci. 1971, 15, 2471–2485, doi:10.1002/app.1971.070151014.
116. Daniels, B.K. Orthotropic swelling and simplified elasticity laws with special reference to cord-reinforced
rubber. J. Appl. Polym. Sci. 1973, 17, 2847–2853, doi:10.1002/app.1973.070170921.
117. Fan, Y.; Gomez, A.; Ferraro, S.; Pinto, B.; Muliana, A.; La Saponara, V. Diffusion of water in glass fiber
reinforced polymer composites at different temperatures. J. Compos. Mater. 2018, 53, 1097–1110,
doi:10.1177/0021998318796155.
118. Jacquemin, F.; Fréour, S.; Guillén, R. Analytical modeling of transient hygro-elastic stress concentration—
Application to embedded optical fiber in a non-uniform transient strain field. Compos. Sci. Technol. 2006, 66,
397–406, doi:10.1016/j.compscitech.2005.07.019.
119. Meng, M.; Rizvi, M.J.; Le, H.R.; Grove, S.M. Multi-scale modelling of moisture diffusion coupled with stress
distribution in CFRP laminated composites. Compos. Struct. 2016, 138, 295–304,
doi:10.1016/j.compstruct.2015.11.028.
120. Sinchuk, Y.; Pannier, Y.; Gueguen, M.; Tandiang, D.; Gigliotti, M. Computed-tomography based modeling
and simulation of moisture diffusion and induced swelling in textile composite materials. Int. J. Solids Struct.
2018, 154, 88–96, doi:10.1016/j.ijsolstr.2017.05.045.
121. Sinchuk, Y.; Pannier, Y.; Gueguen, M.; Gigliotti, M. Image-based modeling of moisture-induced swelling nd
stress in 2D textile composite materials using a global-local approach. Proc. Inst. Mech. Eng., Part C: J. Mech.
Eng. Sci. 2018, 232, 1505–1519, doi:10.1177/0954406217736789.
122. Weitsman, Y.J. Fluid Effects in Polymers and Polymeric Composites; Springer: New York, NY, USA, 2012;
ISBN:978-1-4614-1059-1.
123. Sar, B.E.; Fréour, S.; Davies, P.; Jacquemin, F. Accounting for differential swelling in the multi-physics
modelling of the diffusive behaviour of polymers. J. Appl. Math. Mech. 2014, 94, 452–460,
doi:10.1002/zamm.201200272.
124. Fan, Y.; Gomez, A.; Ferraro, S.; Pinto, B.; Muliana, A.; La Saponara, V. The effects of temperatures and
volumetric expansion on the diffusion of fluids through solid polymers. J. Appl. Polym. Sci. 2017, 134, 45151–
45165, doi:10.1002/app.45151.
125. Kappert, E.J.; Raaijmakers, M.J.T.; Tempelman, K.; Cuperus, F.P.; Ogieglo, W.; Benes, N.E. Swelling of 9
polymers commonly employed for solvent-resistant nanofiltration membranes: A comprehensive dataset. J.
Membr. Sci. 2019, 569, 177–199, doi:10.1016/j.memsci.2018.09.059.
126. Ramirez, F.A.; Carlsson, L.A.; Acha, B.A. Evaluation of water degradation of vinylester and epoxy matrix
composites by single fiber and composite tests. J. Mater. Sci. 2008, 43, 5230–5242, doi:10.1007/s10853-008-2766-
z.

62
127. Shirell, C.D.; Halpin, J. Moisture Absorption in Epoxy Composite Materials. In Proceedings of the Composite
Materials: Testing and Design (Fourth Conference), ASTM STP617, Valley Forge, PA, USA, 3–4 May 1977;
American Society for Testing and Materials: Philadelphia, PA, USA, pp. 514–528, doi:10.1520/STP26963S.
128. Echtermeyer, A.T. Integrating Durability in Marine Composite Certification. In Durability of Composites in a
Marine Environment; Davies, P.; Rajapakse, Y.D.S., Eds.; Springer: Dordrecht, The Netherlands, 2014; pp. 179–
194. ISBN:978-94-007-7416-2.
129. International Standard ISO 6721-11:2012(E), Plastics – Determination of dynamic mechanical properties – Part
11: Glass transition temperature, 2012.
130. International Standard ISO 527-1:2012(E), Plastics – Determination of tensile properties – Part 1: General
principles, 2012.
131. International Standard ISO 2078:1993 (revised in 2014), Textile glass – Yarns – Designation, 2014.
132. 3B Fibreglass technical data sheet. HiPer-tex W2020 rovings, Belgium, 2012.
133. 3B Fibreglass data sheet. https://www.3b-fibreglass.com/HiPer-tex (last accessed on 5 July 2019).
134. Emadipour, H.; Chiang, P.; Koenig, J.L. Interfacial strength studies of fibre-reinforced composites. Res
Mechanica 1982, 5, 165-176.
135. ASTM D3171/ D3171-15. Standard Test Methods for Constituent Content of Composite Materials, 2015.
136. Zinck, P.; Gerard, J.F. On the hybrid character of glass fibres surface networks. J. Mater. Sci. 2005, 40(9), 2759-
2760, doi:10.1007/s10853-005-2124-3.
137. Swinehart, D.F. The Beer–Lambert Law. J. Chem. Educ. 1962, 39(7), 333-335, doi: 10.1021/ed039p333.
138. Belec, L.; Nguyen, T.H.; Nguyen, D.L.; Chailan, J.F. Comparative effects of humid tropical weathering and
artificial ageing on a model composite properties from nano- to macro-scale. Compos. Part A 2015, 68, 235-241,
doi:10.1016/j.compositesa.2014.09.028.
139. Down, J.L. The Yellowing of Epoxy Resin Adhesives: Report on High-Intensity Light Aging. Stud. Conserv.
1986, 31, 159–170, doi:10.2307/1506247.
140. Down, J.L. The yellowing of epoxy resin adhesives: Report on natural dark aging. Stud. Conserv. 1984, 29, 63–
76, doi:10.2307/1506076.
141. Coutinho, I.; Ramos, A.M.; Lima, A.M.; Fernandes, F.B. Studies of the degradation of epoxy resins used for
the conservation of glass. In Proceedings of the Holding it all together, Ancient and Modern Approaches to Joining,
Repair and Consolidation, London, UK, 2008, doi:10.13140/RG.2.1.2799.3124.
142. Celina, M.C.; Dayile, A.R.; Quintana, A. A perspective on the inherent oxidation sensitivity of epoxy
materials. Polymer 2013, 54, 3290–3296, doi:10.1016/j.polymer.2013.04.042.
143. Ernault, E.; Richaud, E.; Fayolle, B. Thermal-oxidation of epoxy/amine followed by glass transition
temperature changes. Polym. Degrad. Stab. 2017, 138, 82–90, doi:10.1016/j.polymdegradstab.2017.02.013.
144. Galant, C.; Fayolle, B.; Kuntz, M.; Verdu, J. Thermal and radio-oxidation of epoxy coatings. Prog. Org. Coat.
2010, 69, 322–329, doi:10.1016/j.porgcoat.2010.07.005.
145. Deng, T.; Liu, Y.; Cui, X.; Yang, Y.; Jia, S.; Wang, Y.; Lu, C.; Li, D.; Cai, R.; Hou, X. Cleavage of C–N bonds in
carbon fiber/epoxy resin composites. Green Chem. 2015, 17, 2141–2145, doi:10.1039/C4GC02512A.
146. Li, K.; Wang, K.; Zhan, M.-S.; Xu, W. The change of thermal-mechanical properties and chemical structure of
ambient cured DGEBA/TEPA under accelerated thermo-oxidative aging. Polym. Degrad. Stab. 2013, 98, 2340–
2346, doi:10.1016/j.polymdegradstab.2013.08.014.
147. Zahra, Y.; Djouani, F.; Fayolle, B.; Kuntz, M.; Verdu, J. Thermo-oxidative aging of epoxy coating systems.
Prog. Org. Coat. 2014, 77, 380–387, doi:10.1016/j.porgcoat.2013.10.011.
148. Ernault, E.; Richaud, E.; Fayolle, B. Thermal oxidation of epoxies: Influence of diamine hardener. Polym.
Degrad. Stab. 2016, 134, 76–86, doi:10.1016/j.polymdegradstab.2016.09.030.
149. Sauvant-Moynot, V.; Duval, S.; Grenier, J. Innovative pipe coating material and process for high temperature
fields. Oil Gas Sci. Technol. 2002, 57, 269–279, doi:10.2516/ogst:2002019.
150. Theocaris, P.S. Influence of plasticizer on Poisson’s ratio of epoxy polymers. Polymer 1979, 20, 1149-1154,
doi:10.1016/0032-3861(79)90308-2.
151. Fine, R.A.; Millero, F.J. Compressibility of water as a function of temperature and pressure. J. Chem. Phys.
1973, 59(10), 5529-5536, doi:10.1063/1.1679903.
152. Eastes, W.; Potter, R.M.; Hadley, J.G. Estimating in-vitro glass fibre dissolution rate from composition. Inhal.
Toxicol. 2000, 12, 269–280, doi:10.1080/089583700196149.

63
153. Khawam, A.; Flanagan, D.R. Solid-State Kinetic Models: Basics and Mathematical Fundamentals. J. Phys.
Chem. B 2006, 110, 17315-17328, doi:10.1021/jp062746a.
154. Michalske, T.A.; Bunker, B.C. A Chemical Kinetics Model for Glass Fracture. J. Am. Ceram. Soc. 1993, 76, 2613–
2618, doi:10.1111/j.1151-2916.1993.tb03989.x.
155. Shiue, Y.S.; Matthewson, M.J. Stress dependent activation entropy for dynamic fatigue of pristine silica
optical fibres. J. Appl. Phys. 2001, 89, 4787–4793, doi:10.1063/1.1361245.
156. Criscenti, L.J.; Kubicki, J.D.; Brantley, S.L. Silicate Glass and Mineral Dissolution: Calculated Reaction Paths
and Activation Energies for Hydrolysis of a Q3 Si by H3O+ Using Ab Initio Methods. J. Phys. Chem. A 2006,
110, 198–206, doi:10.1021/jp044360a.
157. ASTM D5229/D5229M-14. Standard Test Method for Moisture Absorption Properties and Equilibrium
Conditioning of Polymer Matrix Composite Materials, ASTM International; West Conshohocken, PA, 2014.
158. Perreux, D.; Choqueuse, D.; Davies, P. Anomalies in moisture absorption of glass fibre reinforced epoxy
tubes. Composites Part A 2002, 33, 147-154, doi: 10.1016/S1359-835X(01)00111-7.
159. Gagani, A.I.; Krauklis, A.E.; Sæter, E.; Vedvik, N.P.; Echtermeyer, A.T. A Novel Method for Testing and
Determining ILSS for Marine Composites. Comp. Struct. 2019, 220, 431-440,
doi:10.1016/j.compstruct.2019.04.040.
160. Thomason, J.L. The interface region in glass-fibre-reinforced epoxy resin composites: 2. Water absorption,
voids and the interface. Composites 1995, 26, 477-485, doi:10.1016/0010-4361(95)96805-G.
161. Mazan, T.; Jørgensen, J.K.; Echtermeyer, A.T. Aging of polyamide 11. Part 2: General multiscale model of the
hydrolytic degradation applied to predict the morphology evolution. J. Appl. Polym. Sci. 2015, 132(41), 42630-
42639, doi:10.1002/app.42630.
162. Mazan, T.; Jørgensen, J.K.; Echtermeyer, A.T. Aging of polyamide 11. Part 3: Multiscale model predicting the
mechanical properties after hydrolytic degradation. J. Appl. Polym. Sci. 2015, 132(46), 42792-42802,
doi:10.1002/app.42792.
163. Fournier, R.O.; Rowe, J.J. The solubility of amorphous silica in water at high temperatures and high pressures.
Am. Mineral. 1977, 62, 1052-1056.
164. Crundwell, F.K. On the Mechanism of the Dissolution of Quartz and Silica in Aqueous Solutions. ACS Omega
2017, 2(3), 1116–1127, doi:10.1021/acsomega.7b00019.
165. Von Damm, K.L.; Edmond, J.M.; Grant, B.; Measures, C.I.; Walden, B.; Weiss, R.F. Chemistry of submarine
hydrothermal solutions at 21°N, East Pacific Rise. Geochim. Cosmochim. Acta 1985, 49, 2197-2220,
doi:10.1016/0016-7037(85)90222-4.
166. Holland, H.D. The chemistry of the atmosphere and oceans; Wiley: New York, USA, 1978; pp. 351, ISBN:978-
0471035091.
167. Wierderhorn, S.M.; Bolz, L.H. Stress corrosion and static fatigue of glass. J. Am. Ceram. Soc. 1970, 53(10), 543-
548, doi:10.1111/j.1151-2916.1970.tb15962.x.
168. Sekine, H.; Beaumont, P.W.R. A physically based micro-mechanical theory and diagrams of macroscopic
stress-corrosion cracking in aligned continuous glass-fiber-reinforced polymer laminates. Compos. Sci.
Technol. 1998, 58, 1659-1665, doi:10.1016/S0266-3538(97)00236-4.
169. Charles, R.J. Static Fatigue of Glass. II. J. Appl. Phys. 1958, 29, 1554-1560, doi:10.1063/1.1722992.

64
APPENDICES

65
66
APPENDIX A

PAPER I

KRAUKLIS A.E., GAGANI A.I., ECHTERMEYER A.T.

NEAR-INFRARED SPECTROSCOPIC METHOD FOR MONITORING WATER CONTENT IN

EPOXY RESINS AND FIBER-REINFORCED COMPOSITES.

MATERIALS (SWITZERLAND), 11(4), 2018, 586-599.

DOI:10.3390/MA11040586

PAPER I

67
68
 materials
Article
Near-Infrared Spectroscopic Method for Monitoring
Water Content in Epoxy Resins and
Fiber-Reinforced Composites
ID
Andrey E. Krauklis * , Abedin I. Gagani and Andreas T. Echtermeyer
Department of Mechanical and Industrial Engineering (past: Department of Engineering Design and Materials),
Norwegian University of Science and Technology, 7491 Trondheim, Norway; [email protected] (A.I.G.);
[email protected] (A.T.E.)
* Correspondence: [email protected]; Tel.: +371-268-10288

Received: 14 February 2018; Accepted: 9 April 2018; Published: 11 April 2018 

Abstract: Monitoring water content and predicting the water-induced drop in strength of
fiber-reinforced composites are of great importance for the oil and gas and marine industries. Fourier
transform infrared (FTIR) spectroscopic methods are broadly available and often used for process
and quality control in industrial applications. A benefit of using such spectroscopic methods over the
conventional gravimetric analysis is the possibility to deduce the mass of an absolutely dry material
and subsequently the true water content, which is an important indicator of water content-dependent
properties. The objective of this study is to develop an efficient and detailed method for estimating
the water content in epoxy resins and fiber-reinforced composites. In this study, Fourier transform
near-infrared (FT-NIR) spectroscopy was applied to measure the water content of amine-epoxy neat
resin. The method was developed and successfully extended to glass fiber-reinforced composite
materials. Based on extensive measurements of neat resin and composite samples of varying water
content and thickness, regression was performed, and the quantitative absorbance dependence on
water content in the material was established. The mass of an absolutely dry resin was identified,
and the true water content was obtained. The method was related to the Beer–Lambert law and
explained in such terms. A detailed spectroscopic method for measuring water content in resins and
fiber-reinforced composites was developed and described.

Keywords: epoxy; composites; water; NIR; spectroscopy

1. Introduction
Epoxy resins are widely used in composite materials (e.g., fiber-reinforced laminates), adhesives
and surface coatings due to their relatively high strength, stiffness, low volatility and shrinkage during
the curing process, as well as good adhesive properties. However, it is known that highly crosslinked
amine-cured epoxy resins are hydrophilic, and their properties can significantly deteriorate upon water
uptake [1–3]. It is well established that water uptake is an important factor in the performance and
durability of epoxy-based composites, which undergo plasticization and swelling stresses [4].
Composite materials are often exposed to water or humid air environments. This is of special
interest for offshore applications, since composites are widely used there due to their relatively light
weight and high corrosion resistance, especially when compared to steel. Furthermore, the deep-water
applications of composites have to be mentioned such as in, for example, risers and tethers [5].
It has been reported that water environments negatively impact the mechanical properties of such
materials [6–12]. It is therefore of high importance to develop a water content monitoring method for
such materials exposed to water environments.

Materials 2018, 11, 586; doi:10.3390/ma11040586 www.mdpi.com/journal/materials

Materials 2018, 11, 586 2 of 14

Among the methods used for measuring the water content, such experimental techniques as
thermogravimetry, differential scanning calorimetry (DSC), ultraviolet (UV) reflection spectroscopy,
attenuated total reflection Fourier transform infrared spectroscopy (ATR-FTIR) and Fourier transform
near-infrared spectroscopy (FT-NIR) have been reported [3,13–20]. The most common is a gravimetric
method, which provides weight gain information upon water uptake or the weight loss upon drying.
A significant drawback of the gravimetric method is the necessity to know the mass of the absolutely
dry material, which in some cases requires long drying times and a complicated setup to obtain
such data. Therefore, an alternative technique for measuring water content in epoxy resins and
fiber-reinforced composites is required. NIR spectroscopy, e.g., FT-NIR, is a promising technique
for water content monitoring [21–23]. Due to the high transmission of NIR light, its applicability to
products with sizes in the mm range is usually possible [21].
It is common to report water content without accounting for moisture initially present in the
material. In reality, some water is already present in the resin uptaken from the air, which has a certain
humidity. The spectroscopic method provides a way to deduce the amount of water present initially
in the material. Combining initial water content with that from water uptake experiments, the true
water content is obtained. True water content and its determination are described further in the work.
It is important to know the true water content when one has to look at strength, stiffness and fatigue
property changes due to water content [24,25]. It was found that true water content at initial conditions
is already 0.63% for the studied epoxy. These conditions are what would usually be denoted as a dry
material. The true water content at saturation is 3.44%. Thus, the initial water content is significant
and should not be neglected. The provided method allows determining and quantifying it.
Recent improvements of NIR spectrometers, resulting in a dramatic increase in their performance,
e.g., acceleration of measuring time, have further lowered the threshold for using this method for
process monitoring [21,22]. NIR has a broad range of practical applications, both in the laboratory and
industry, the evaluation of diffusant content being one of those. For instance, NIR spectroscopy has
long been used in such fields as medicine, food and polymers [21,26–28].
Multiple studies of water content evaluation in polymers have been reported. For example,
Camacho et al. [26] and Kuda-Malwathumullage et al. [27] constructed NIR spectroscopic models for
polyamides (Nylon 6,6), while Muroga et al. developed an NIR model for evaluating the water content
of molded polylactide (PLA) under the effect of crystallization [21]. Studies on the determination
of the water content of polymer/filler nanocomposites at variable relative humidity have been
reported [29,30]. Use of spectroscopic methods for epoxies has also been reported, including use
of FT-NIR to study epoxy network interactions with water molecules [15,17,31–34]. However, despite
the described benefits of NIR spectroscopy, and specifically FT-NIR, and the mentioned relevant studies
on epoxy-water interactions, the authors believe that a sufficiently detailed and simple-in-application
method needs to be reported and discussed, allowing one to evaluate the water content in epoxy resin
and epoxy-based composite samples of varying thickness and water content. Thus, the objective of this
study is to establish such a model and provide the details of the method for its application in practice.
Previous work has only been done on polymers, but not on fiber-reinforced composites. Such work on
composites is seen as novel and important for industry. It has not been reported anywhere else to the
best knowledge of the authors.

2. Experimental Section

2.1. Materials
Amine-cured epoxy: Amine-cured epoxy resin was prepared using reagents Epikote Resin
RIMR 135 (Hexion, International: USA/Europe) and Epikure Curing Agent MGS RIMH 137 (Hexion,
International: USA/Europe) in a ratio of 100:30 by weight. The mixture was degassed in a vacuum
chamber for 0.5 h in order to remove bubbles. The sample mold was prepared using computer
numerical control (CNC) machining. Degassed resin was then molded into rectangular-shaped
Materials 2018, 11, 586 3 of 14

samples, followed by curing at room temperature for 24 h and post-curing in an air oven (Lehmkuhls
Verksteder, Oslo, Norway) at 80 ◦ C for 16 h. After samples were post-cured, the resin was removed
from the mold’s grooves and cut into the desired length with a vertical bandsaw. After cutting epoxy
samples, the precise desired length was obtained by using sandpaper (FEPA P60, grain size 269 μm)
to grind the edges. The prepared mold allows sufficient width control within a tolerance of 5%.
In order to get samples to the right thickness and enable sufficient thickness control, a metal holder
for grinding was prepared and used. The desired thickness was obtained using PHOENIX 2000
(Jean Wirtz, Dusseldorf, Germany) and SiC grinding discs (Struers, Cleveland, OH, USA; FEPA P500,
grain size 30 μm). Exicator grease (Riedel-de Haën, Seelze, Germany) was used to enable sufficient
adhesion of the sample with the holder. The sufficient thickness control, correct length and width
were ensured within a 5% tolerance. Dog bone-shaped epoxy resin samples used in tensile tests were
prepared in a similar way.
Composite laminates: Glass fiber-reinforced epoxy composite laminates were prepared using the
vacuum-assisted resin transfer molding (VARTM) process. The composite laminate plate was turned
into cylinders of a diameter of 20 mm. The cylinders were then cut into discs of a thickness slightly
above 2 mm. The thickness was then adjusted to 2 mm within a 5% tolerance via grinding with a super
fine sandpaper (FEPA P800, grain size 21.8 μm).
DI water: Deionized water (0.5–1.0 MΩ·cm) was used for water uptake measurements, produced
via the water purification system Aquatron A4000 (Cole-Parmer, Vernon Hills, IL, USA).

2.2. Methods
Conditioning of resin and composite samples in water: Water uptake experiments were conducted
using a batch system. A heated DI water (60 ◦ C) bath was used for conditioning the samples. Samples
were weighed using the analytical scales AG204 (±0.1 mg; Mettler Toledo, Columbus, OH, USA) in
order to obtain the mass of the unconditioned samples and placed into the water bath. Samples were
conditioned until equilibrium was achieved. Up to and at the saturation point, the samples were
taken out of the water bath, weighed and analyzed with an FTIR spectrophotometer in order to obtain
spectra at different water contents.
Drying of conditioned resin and composite samples: The drying of saturated samples was
performed in a drying cabinet PK-410 (ESAB, London, UK) at 60 ◦ C in air atmosphere with natural
convection and relative humidity of 13 RH%.
Determination of resin fraction in composites: The density of neat epoxy resin (ρresin ) and glass
fiber (ρ glass ) was 1.1 g/cm3 and 2.54 g/cm3 , respectively. The density of the composite (ρcomp ) was
determined to be 1.97 g/cm3 by measuring the mass and dimensions of a large composite block.
The volume and mass fractions of neat resin were calculated using Equations (1) and (2), respectively.

ρcomp − ρresin
Vfresin = 1 − Vf glass = 1 − (1)
ρ glass − ρresin

ρresin ·Vfresin
m fresin =   (2)
ρresin ·Vf resin + ρ glass · 1 − Vfresin

The volume and mass fraction of resin are Vfresin = 0.394 and m fresin = 0.220, respectively.
FT-IR: Sample characterization with different water content was performed using Fourier
transform infrared spectroscopy (FTIR) in the near-infrared range. Near-infrared spectra were obtained
using the Fourier transform spectrophotometer NIRSystems 6500 (Foss, Eden Prairie, MN, USA)
operated in transmission mode; an optical fiber probe and spectral analysis software Vision (Foss,
Eden Prairie, MN, USA) were used. Spectra were taken in the Vis-NIR wavenumber range of
4000–25,000 cm−1 using 32 scans per spectrum with a resolution of 4 cm−1 . A spectrum of the driest
(in this case, dried) sample was subtracted from the spectrum of the sample of interest. Then, the line
connecting spectrum points at 5400 and 4900 cm−1 was constructed. The slope and the intercept of
Materials 2018, 11, 586 4 of 14

this line were obtained. Subsequently, baseline correction was performed by subtracting the obtained
line from the spectrum of interest.
Tensile tests: To evaluate the effect of water content on the ultimate tensile strength (UTS) of
neat epoxy resin, tensile tests were conducted using the servo hydraulic test machine Instron Model
1342 (Instron, International: USA/UK). Dog bone-shaped epoxy resin samples were used in order to
determine tensile strength; the rate was set to 1 mm/min of controlled displacement.

3. Results and Discussion

3.1. Water Uptake, Drying and Conditioning in Air Experiments of Neat Resin
The definition of water content in neat resin is described by Equation (3):

m(t) − mdried
W (t) ≡ ·100% (3)
mdried

where m(t) is the mass of the wet neat resin; mdried is the mass of the resin after drying.
Four samples of neat resin were used to obtain the water uptake and drying curves, as well as
conditioning in air at room temperature back to initial water content; average values with standard
deviations are reported. The curves are shown in Figure 1.

Figure 1. Water uptake, drying and conditioning in air curves of amine-cured epoxy resin.

3.2. Reversible Drop in Ultimate Tensile Strength of Neat Resin with Water Content
The ultimate tensile strength (UTS) was measured (1) for neat resin at initial conditions
(in equilibrium with water vapor in ambient conditions), (2) for water-saturated resin and (3) for dried
resin air-conditioned back to initial water content. The ultimate tensile strength (UTS) has decreased
down to 80% of the initial dry value (from about 60 down to 48 MPa), as shown in Figure 2. Results
indicate that the UTS of dried epoxy is comparable to that of the initial, indicating the reversibility of
the drop in UTS. This aspect of reversibility is reported in greater detail elsewhere [24].
Materials 2018, 11, 586 5 of 14

Figure 2. Water-induced drop in ultimate tensile strength (UTS) of amine-cured epoxy resin.

3.3. The Method for Monitoring Water Content in Neat Resin

The spectrum of a dried epoxy sample was obtained and is further denoted as the reference
spectrum (the sample with the lowest water content available). The water (diffusant) absorption
band’s maximum was identified to be at about 5200 cm−1 . This band changed depending on the water
content, as shown in Figure 3. Based on study by Falk et al. [35], this band corresponds to a combination
mode of stretching and bending of the water molecule’s OH group. This observation is also consistent
with a recent study by Muroga et al. on spectroscopic evaluation of water content in polylactide (PLA)
and with a novel work on water monitoring methods in PMMA by Wiedemair et al. [21,36]. Thus,
this absorption band is chosen to monitor the water content of the epoxy resin. The wavenumber of
the water absorption band shifts to lower values as the water content increases. True water content
and the corresponding wavenumber values at the absorption peak are reported in Table 1.

Figure 3. Difference of spectra with respect to the dried epoxy of the water absorbance band in epoxy
resin samples of varying water content. The baseline corresponds to the driest sample (in this case, dried).

Table 1. True water content and the corresponding wavenumber values at the absorption peak.

True Water Content, W* (%) Wavenumber (cm−1 )

0.26 5230
0.61 5225
1.68 5219
1.88 5214
2.84 5214
2.94 5208
3.34 5208
Materials 2018, 11, 586 6 of 14

The absorbance at the maximum of the water absorption band was quantified in the identified
range. Absorbance values of the water band are comparable to those reported for studying water
content in other polymeric systems [15,21]. The linear regression analysis (linear least square method)
was then performed to obtain the equation A = f (W ). Based on 40 measurements, for the studied
neat resin, the regression equation (Equation (4)) was obtained (R2 = 0.9466):

A = 0.1248W + 0.0330 = Kslope

0
W+β (4)

where 0.0330 is the intercept β and is due to the undried water content. Since the slope is the same for
the undried water content Wundried water as for the determined water content W, the relationship can be
rewritten in the form of Equation (5).

A = Kslope
0
(W + Wundried water ) (5)

When W is equal to zero, absorbance is present because the samples are not fully dried. Thus,
the intercept Wundried water is due to the presence of water that was not removed during the drying
process. This allows one to determine the amount of water present in the sample after drying, as shown
by Equation (6), and to deduce how much water was in samples at equilibrium at ambient conditions
if needed. Furthermore, this allows one to determine the mass of the absolutely dry material.

β
Wundried water = 0
(6)
Kslope

In this case, the undried water content Wundried water is equal to 0.26%, meaning that even after
drying, there is still a significant amount of water present, which is supported by the observable water
band in the IR spectra of dried samples. In order to calculate the mass of the absolutely dry neat resin
m absolutely dry (Equation (8)), Equations (3) and (5) are used.
 
m(t) − mdried W m
A = Kslope
0
·100% + undried water dried (7)
mdried mdried
 
Wundried water
m absolutely dry = 1− mdried (8)
100
This leads to a definition of the true water content. The true water content shows the amount of
water with respect to the absolutely dry material and is defined as in Equation (9).
 
mwater (t) m(t) − m absolutely dry
W ∗ (t) ≡ ·100% = ·100% (9)
m absolutely dry m absolutely dry

The water content monitoring equation then can be written in the form of Equation (10).

A = Kslope W ∗ (10)

The water content is then recalculated to obtain the true water content W ∗ as defined by Equation (9).
The linear regression for A = f (W ∗ ) gives a similar slope (Kslope is equal to 0.1247) with a zero intercept
to the initial result of A = f (W ) (Kslope
0 is equal to 0.1248). The determination coefficient is R2 = 0.9466,
which indicates that this model equation accounts for 89.61% of the variation in absorbance band
maximum values in the dataset. The remaining 10.39% of variation not explained by the equation is
expected to be due to the sample thickness having a tolerance of 5%, as well as some possible drying in
air during the collection of spectra, since spectra are taken in ambient conditions at room temperature
in air atmosphere. Such low variation even within a 5% thickness tolerance of samples indicates this
method as being precise and efficient for monitoring water content of neat resin. Thus, this method
Materials 2018, 11, 586 7 of 14

can be used as an indicator of the water concentration-dependent drop of the mechanical properties of
the material.

3.4. Extension of the Method to Samples of Varying Thickness

Based on the Beer–Lambert law (Equation (11)), absorbance is dependent on the molar attenuation
(absorption) coefficient (ε), concentration (c) and the path length (l).
 
I0
A = log10 = εcl (11)
I

In our case, the concentration term is the true water content (W ∗ ), and the path length is the
thickness of the sample (δ). Thus, the Beer–Lambert law in our case can be expressed as Equation (12).

A = ε∗ W ∗ δ (12)

Note that in this case, attenuation coefficient ε∗ is not the same as a conventional molar attenuation
coefficient ε, since the concentration term used is of true water content (W ∗ ) and not of the molar
concentration (c); thus, in order to avoid misunderstanding, in this work, the proportionality coefficient
will be denoted as ε∗ . However, calculation of the molar attenuation coefficient will be shown later in
this work, as well. In practice, mass concentrations are easier to imagine and work with. However,
conventionally, attenuation coefficients are given in molar units. Thus, a way to convert between mass
and molar concentrations is provided.
Since the model is as shown in Equation (10), the thickness effect (path length) is contained in
the slope term of the model equation (Equation (9)); thus, the slope term can be divided into the
attenuation coefficient and the thickness. The new slope term is the attenuation coefficient and is
obtained as shown in Equation (13).
Kslope
ε∗ = (13)
δ
In the case of resin, the water content monitoring equation (Equation (9)) then becomes Equation (12).
The model in Equation (12) is validated by using water-saturated resin samples of varying thickness, e.g.,
by setting term W ∗ to a constant value, thus proving the linear dependence of absorbance on sample
thickness (path length), as shown in Figure 4. The determination coefficient is R2 = 0.8480.

Figure 4. Linear dependence of absorbance on the sample thickness of water-saturated neat resin.
Materials 2018, 11, 586 8 of 14

Based on the linear dependence of absorbance on sample thickness, for saturated samples ε∗ Wmax ∗
− 1 −
equals 0.2312, and the attenuation coefficient is then 0.0672 ± 0.0034% ·mm (R = 0.8480). Using
1 2

model equation (Equation (10)), it is known that ε∗ δ2 mm equals 0.1247. From this, taking into account a
5% thickness tolerance, the attenuation coefficient is 0.0624 ± 0.0031%−1 ·mm−1 . The values are within
the standard deviation. Since the attenuation coefficient ε∗ in our case has units of %−1 ·mm−1 , in order
to obtain the molar attenuation coefficient ε, calculation of water molar concentration is required. Using
the definition of true water content (Equation (9)), the relationship between the molar concentration of
the diffusant (water) and the true water content can be written as Equation (14).
 W ∗ (t) [%]
·m absolutely dry [g]
 100
mol Mwater [ mol
g
]
c(t) = (14)
L a·b·δ [mm3 ]·10−6 L
mm3

where a and b are length and width of the sample in mm, respectively.
Since thickness is the path length (l = δ) and εcl = ε∗ W ∗ δ (Equations (11) and (12)), the molar
attenuation coefficient can then be obtained from Equation (15).

ε∗ W ∗
ε= (15)
c

Since ε∗ W ∗ is found to be equal to 0.2312 mm−1 for the resin of interest at full saturation
∗
(Wmax = 3.44%) for a sample of 2 mm in thickness, using Equation (14), the molar concentration is
2.29 ± 0.12 M, and using Equation (15), the molar attenuation coefficient ε is equal to 1.01 ± 0.06 molL·cm .
The low value of the molar attenuation coefficient is explained by the fact that resin media has a
relatively high light attenuation itself. Note that molar attenuation coefficient for water in epoxy media
ε is obtained using the difference spectra with respect to the spectrum of the dried neat resin.
The authors have assessed the method in the thickness range from 1.06–2.24 mm. The authors
expect that the method is applicable also to thinner samples. Extrapolation to samples thicker than
2.24 mm is not completely certain.

3.5. Extension of the Method to Composite Systems

In order to extend the method of monitoring water content to fiber-reinforced composites,
the mass fraction of resin has to be known. For the studied glass-fiber reinforced composites, the resin
mass fraction for samples at initial conditions (m fresin ) was determined as described earlier and
initial
is 0.2198. The resin mass fraction is used in order to deduce diffusant uptake by resin from the
composite spectra. The proposed equation for monitoring the true water content due to exposure of
composites to water media is represented by Equation (16).

A = ε∗ Wresin
∗ ∗
δ = Kslope Wresin (16)

The definition of the water content in the composite material is described by Equation (17).

m(t) − mcompdried
Wcomp ≡ ·100% (17)
mcompdried

The true water content of composites is defined by Equation (18).

∗
mcomp (t) − mcompabsolutely dry
Wcomp ≡ ·100% (18)
mcompabsolutely dry

An assumption is required that fiber is always dry (not absorbing water) and neglecting the
influence of the sizing, since the mass of sizing is negligible compared to resin mass. This is, however,
Materials 2018, 11, 586 9 of 14

not always fully true and is a convenient assumption. The mass of the composite is then as shown in
Equation (19).
mcompdried = mresindried + m f iber (19)

In order to prove that the assumption is valid in this case, that the increase in fiber or sizing mass
is negligible due to water uptake, the water uptake using prepared composite discs was performed
and scaled by the factor of the resin mass fraction of the composite. The comparison of the neat resin
and the scaled composite water uptake graphs is represented in Figure 5.

Figure 5. True water content curves of neat resin and glass fiber-reinforced composite (scaled by the
resin mass fraction).

The shape of the kinetic curve is sample geometry-dependent and is slightly different in this case,
as seen in Figure 5. However, the equilibrium water uptake value is a material property and can
be used for the comparison. As seen in Figure 5, the equilibrium true water content value for a
composite (scaled by resin mass fraction) is only slightly higher than that for the neat resin, resulting
in a difference of 0.12% in true water content between average values. The difference can be explained
due to the sizing also absorbing water. The authors believe that an increase of the true water content
(at equilibrium) for the composite compared to the epoxy is increased due to the interactions of
epoxy/fibers via the formation of the interphase. It is likely that what is measured is a sum of water
content in both resin and the interphase. It has been reported that the interphase might be responsible
for increased water uptake [37].
Thus, the assumption that only resin uptakes water is not completely true, but considering the
relatively low difference, this is a reasonable approximation. Another aspect that might be causing the
difference in the equilibrium water uptake value could also be the presence of voids in the composite,
which is usually greater than in the neat resin due to the specifics of the impregnation process.
Use of such and approximation should be assessed in the case of other fiber-resin systems, since
sizing can have a varying influence on the water uptake in different systems. For instance, in the case
of carbon fiber-vinylester composites, it has been reported that the sizing affects the equilibrium water
uptake of the composite significantly [37]. Thus, it has to be noted that the extension of the method
might have limitations for certain composite systems.
Since the mass of water comes into the resin mass term, the mass fraction of resin is dependent on
the water content and is defined as in Equation (20).
Materials 2018, 11, 586 10 of 14

mresin (t) mresinabsolutely dry + mwater (t)

m fresin (t) ≡ = (20)
mresin (t) + m f iber mresinabsolutely dry + mwater (t) + m f iber

In order to obtain the absolutely dry mass of the composite material, the set of Equations (21)–(23)
is used.  
mcompdried − 1 − m fresin ·mcompinitial
m fresin = initial
(21)
dried mcompdried
 
W
mresinabsolutely dry = m f resin ·mcompdried · 1 − undried water (22)
dried 100
 
mcompabsolutely dry = mresinabsolutely dry + 1 − m fresin ·mcompinitial (23)
initial

The true water content of the resin component in the composite can be calculated using Equation (24).
 
mcomp (t) − 1 − m fresin ·mcompinitial − mresinabsolutely dry
∗
Wresin = initial
·100% (24)
mresinabsolutely dry

∗
Calculated Wresin is then to be used for predicting the water content, using Equation (25),
analogous to the monitoring equation for neat resin (Equation (10)).

∗
A = Kslope Wresin (25)

In order to use the composite spectra for predicting Wresin ∗ , a factor for scaling from composites

down to resin is required. Since composite materials have components of different absorbance,
a normalization procedure is required. The composite absorbs light much more than the neat resin.
There is a necessity to correct for the summary absorbance of the components via the baseline shift.
In order to do so conveniently, two parameters need to be determined for the developed absorbance
band maximum-based model: the maximum absorbance of the neat resin (A peakresin ) and of the
composite (A peakcomp ) at water saturation (at water uptake equilibrium values). These values are
obtained using the difference spectra for respective fully-saturated materials as described earlier.
In this case, the obtained values are A peakresin = 0.47 and A peakcomp = 0.044. Note that due to the higher
absorbance of the composite material, the absorbance of the water band deduced from the difference
spectra is one order of magnitude lower than for the neat resin. In order to extend the model from neat
resin to composites, we introduce the scaling factor, which is denoted as Peak Factor and defined as
shown in Equation (26). In our case, it is equal to 0.094.

A peakcomp
Peak Factor ≡ (26)
A peakresin

The final water content monitoring equation for the composite system can then be written in the
form of Equation (27) if the effect of thickness is known and in the form of Equation (28) if the effect of
thickness is not known.
A peakcomp = ε∗ · Peak Factor ·δ·Wresin
∗
(27)
∗
A peakcomp = Kslope · Peak Factor ·Wresin (28)

For the particular composite system of interest (glass fiber-reinforced amine-cured epoxy),
the water monitoring equation is represented by Equation (29).

∗
A peakcomp = 0.012·Wresin (29)
Materials 2018, 11, 586 11 of 14

The model was validated by the experimental data using composite discs with a varying water
content. The experimental data and predicted values (Equation (29)) are shown in Figure 6.

Figure 6. Validation of the water monitoring model extended to composites.

The fit between the final model and experimental data for composites resulted in a determination
coefficient R2 of 0.9329, meaning that the developed model accounts for 87.03% of the variation in the
absorbance band’s maximum values in the dataset. The best linear regression fit (R2 = 0.9908) has only
a slightly higher slope than the model, as shown in Figure 6.

3.6. Final Remarks on the Results

The developed method is potentially useful also for other polymers and composite systems with
different fiber or resin fractions, as well as for diffusant media other than water. The developed
method is based on the physics behind the Beer–Lambert law and can be used for samples of varying
thickness. It should be noted, however, that the method has its limitations. If the sample is too thick
or non-transparent in the studied radiation wavelength range, the method is not applicable, e.g.,
monitoring of carbon fiber-composite materials is expected to be limited.
The method can be used as an indicator of water-induced property changes in polymers and
composites. While the drop in ultimate tensile strength (UTS) of the neat resin is reversible for
the studied amine-cured epoxy resin, some types of fibers, such as glass fibers, are susceptible to
irreversible degradation. Thus, in composites, the decline of UTS is not only water-concentration
dependent, but also time-dependent. In cases when the material stays in the water environment during
its working conditions, the driving force is always directed towards the drier interior of the material.
In such cases, spectroscopic water content monitoring can provide valuable information about the
interaction time of composite constituents (matrix, fibers, sizing) with the diffusant.
In practice, when the water content distribution in a composite structure is of interest, use of
a transmittance FTIR is very limited. In order to become applicable for such cases, the following
scenario is proposed. A replica of the structure parts, alongside the actual structure, should be
immersed in water media at the same conditions. These replica parts then can be cut into samples of
the thickness applicable for the method and analyzed. While destructive and requiring additional
Materials 2018, 11, 586 12 of 14

expenses, the method can be a useful indicator of long-term water content and distribution in the
composite structures, where such destructive tests might be required only once in tens of years.
Other spectroscopic methods such as Raman, while not as sensitive to water, and reflectance FTIR,
while mostly providing information about the surface of the material, might also be considered in
developing water monitoring methods, especially in cases when composites are non-transparent to the
IR radiation.

4. Conclusions
In this study, a detailed method for estimating and monitoring water content in epoxy resins
and fiber-reinforced composites was developed using the maxima of the absorbance band at about
5200 cm−1 in the NIR combination mode region correlated with the true water content. The method
provides a benefit over the conventional gravimetric analysis providing the possibility to deduce the
mass of an absolutely dry material and subsequently the true water content, which is an important
indicator of water content-dependent properties. Based on extensive measurements of neat resin
and composite samples of varying water content and thickness, regression was performed, and the
quantitative absorbance dependence on water content in the materials was successfully established.
The model equations for monitoring water content in epoxy resin and composite material samples
were obtained and experimentally validated. The model was related to the Beer–Lambert law and
explained in such terms. The details of the method were reported, allowing the use of the method in
practical applications.

Acknowledgments: This work is part of the DNV (Det Norske Veritas) GL (Germanischer Lloyd) led Joint
Industry Project “Affordable Composites” with twelve industrial partners and the Norwegian University of
Science and Technology (NTNU). The authors would like to express their thanks for the financial support by The
Research Council of Norway (Project 245606/E30 in the Petromaks 2 programme). The authors are thankful to
Bjørn Kåre Alsberg for providing us with the equipment for this study and to our colleague and friend Emeric
Mialon for machining the stainless-steel holder for sample grinding.
Author Contributions: Andrey E. Krauklis conceived of and designed the experiments. Andrey E. Krauklis
and Abedin I. Gagani prepared the samples for experiments. Andrey E. Krauklis performed the experiments.
Andrey E. Krauklis analyzed the data. Andrey E. Krauklis wrote the paper. Andreas T. Echtermeyer initiated the
project. Abedin I. Gagani and Andreas T. Echtermeyer revised the work.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Li, L.; Yu, Y.; Wu, Q.; Zhan, G.; Li, S. Effect of Chemical Structure on the Water Sorption of Amine-Cured
Epoxy Resins. Corros. Sci. 2009, 51, 3000. [CrossRef]
2. Lee, M.C.; Peppas, N.A. Water Transport in Epoxy-Resins. Prog. Polym. Sci. 1993, 18, 947.
3. Wang, J.; Gong, J.; Gong, Z.; Yan, X.; Wang, B.; Wu, Q.; Li, S. Effect of Curing Agent Polarity on Water
Absorption and Free Volume in Epoxy Resin Studied by PALS. Nucl. Instrum. Methods Phys. Res. B 2010,
268, 2355. [CrossRef]
4. Morel, E.; Bellenger, V.; Verdu, J. Structure-Water Absorption Relationships for Amine-Cured Epoxy Resins.
Polymer 1985, 26, 1719. [CrossRef]
5. Echtermeyer, A.T. Integrating Durability in Marine Composite Certification. In Durability of Composites
in a Marine Environment; Davies, P., Rajapakse, Y.D.S., Eds.; Springer: Dordrecht, The Netherlands, 2014;
pp. 179–194.
6. Weitsman, Y. Coupled Damage and Moisture-Transport in Fiber-Reinforced, Polymeric Composites. Int. J.
Solids Struct. 1987, 23, 1003. [CrossRef]
7. Weitsman, Y.J.; Elahi, M. Effects of Fluids on the Deformation, Strength and Durability of Polymeric
Composites—An Overview. Mech. Time-Depend. Mater. 2000, 4, 107. [CrossRef]
8. Roy, S. Moisture-Induced Degradation in Long-Term Durability of Polymeric Matrix Composites; Pochiraju, V.K.,
Tandon, P.G., Schoeppner, A.G., Eds.; Springer: Boston, MA, USA, 2012; pp. 181–236.
Materials 2018, 11, 586 13 of 14

9. Apicella, A.; Nicolais, L. Effect of Water on the Properties of Epoxy Matrix and Composite. Adv. Polym. Sci.
1985, 72, 69.
10. Lefebvre, D.R.; Elliker, P.R.; Takahashi, K.M.; Raju, V.R.; Kaplan, M.L. The Critical Humidity Effect in the
Adhesion of Epoxy to Glass: Role of Hydrogen Bonding. J. Adhes. Sci. Technol. 2000, 14, 925. [CrossRef]
11. Guermazi, N.; Elleuch, K.; Ayedi, H.F. The Effect of Time and Aging Temperature on Structural and
Mechanical Properties of Pipeline Coating. Mater. Des. 2009, 30, 2006. [CrossRef]
12. Wu, C.F.; Xu, W.J. Atomistic Simulation Study of Absorbed Water Influence on Structure and Properties of
Crosslinked Epoxy Resin. Polymer 2007, 48, 5440. [CrossRef]
13. Bertolino, V.; Cavallaro, G.; Lazzara, G.; Merli, M.; Milioto, S.; Parisi, F.; Sciascia, L. Effect of the Biopolymer
Charge and the Nanoclay Morphology on Nanocomposite Materials. Ind. Eng. Chem. Res. 2016, 55, 7373.
[CrossRef]
14. Ciprioti, S.V.; Tuffi, R.; Dell’Era, A.; Poggetto, F.D.; Bollino, F. Thermal Behavior and Structural Study of
SiO2 /Poly(ε-caprolactone) Hybrids Synthesized via Sol-Gel Method. Materials 2018, 11, 275. [CrossRef]
15. Mijović, J.; Zhang, H. Local Dynamics and Molecular Origin of Polymer Network−Water Interactions as
Studied by Broadband Dielectric Relaxation Spectroscopy, FTIR, and Molecular Simulations. Macromolecules
2003, 36, 1279. [CrossRef]
16. Philippe, L.V.S.; Lyon, S.B.; Sammon, C.; Yarwood, J. Validation of Electrochemical Impedance Measurements
for Water Sorption into Epoxy Coatings Using Gravimetry and Infra-Red Spectroscopy. Corros. Sci. 2008,
50, 887. [CrossRef]
17. Cotugno, S.; Mensitieri, G.; Musto, P.; Sanguigno, L. Molecular Interactions in and Transport Properties
of Densely Crosslinked Networks: A Time-Resolved FT-IR Spectroscopy Investigation of the Epoxy/H2 O
System. Macromolecules 2005, 38, 801. [CrossRef]
18. Weir, M.D.; Bastide, C.; Sung, C.S.P. Characterization of Interaction of Water in Epoxy by UV Reflection
Spectroscopy. Macromolecules 2001, 34, 4923. [CrossRef]
19. Mijović, J.; Zhang, H. Molecular Dynamics Simulation Study of Motions and Interactions of Water in a
Polymer Network. J. Phys. Chem. B 2004, 108, 2557. [CrossRef]
20. MacQueen, R.C.; Granata, R.D. A Positron Annihilation Lifetime Spectroscopic Study of the Corrosion
Protective Properties of Epoxy Coatings. Prog. Org. Coat. 1996, 28, 97. [CrossRef]
21. Muroga, S.; Hikima, Y.; Ohshima, M. Near-Infrared Spectroscopic Evaluation of the Water Content of Molded
Polylactide under the Effect of Crystallization. Appl. Spectrosc. 2017, 71, 1300. [CrossRef] [PubMed]
22. Genkawa, T.; Watari, M.; Nishii, T.; Ozaki, Y. Development of a Near-Infrared/Mid-Infrared Dual-Region
Spectrometer for Online Process Analysis. Appl. Spectrosc. 2012, 66, 773. [CrossRef] [PubMed]
23. Ozaki, Y. Near-Infrared Spectroscopy—Its Versatility in Analytical Chemistry. Anal. Sci. 2012, 28, 545.
[CrossRef] [PubMed]
24. Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Hygrothermal Aging of Amine Epoxy: Reversible Static and
Fatigue Properties. Polym. Test. 2018, submitted.
25. Rocha, I.B.C.M.; Raijmaekers, S.; Nijssen, R.P.L.; van der Meer, F.P.; Sluys, L.J. Hygrothermal ageing behaviour
of a glass/epoxy composite used in wind turbine blades. Compos. Struct. 2017, 174, 110. [CrossRef]
26. Camacho, W.; Valles-Lluch, A.; Ribes-Greus, A.; Karksson, S. Determination of Moisture Content in Nylon
6,6 by Near-Infrared Spectroscopy and Chemometrics. J. Appl. Polym. Sci. 2003, 87, 2165. [CrossRef]
27. Kuda-Malwathumullage, C.P.S.; Small, G.W. Determination of Moisture Content of Polyamide 66 Directly
from Combination Region Near-Infrared Spectra. J. Appl. Polym. Sci. 2014, 131, 40645. [CrossRef]
28. Zhou, G.X.; Ge, Z.; Dorwart, J.; Izzo, B.; Kukura, J.; Bicker, G.; Wyvratt, J. Determination and Differentiation
of Surface and Bound Water in Drug Substances by Near Infrared Spectroscopy. J. Pharm. Sci. 2003, 92, 1058.
[CrossRef] [PubMed]
29. Cavallaro, G.; Lazzara, G.; Konnova, S.; Fakhrullin, R.; Lvov, Y. Composite films of natural clay nanotubes
with cellulose and chitosan. Green Mater. 2014, 2, 232. [CrossRef]
30. De Rodriguez, N.G.; Thielemans, W.; Dufresne, A. Sisal cellulose whiskers reinforced polyvinyl acetate
nanocomposites. Cellulose 2006, 13, 261. [CrossRef]
31. Musto, P.; Ragosta, G.; Mascia, L. Vibrational Spectroscopy Evidence for the Dual Nature of Water Sorbed
into Epoxy Resins. Chem. Mater. 2000, 12, 1331. [CrossRef]
32. Moy, P.; Karasz, F.E. Epoxy-Water Interactions. Polym. Eng. Sci. 1980, 20, 315. [CrossRef]
Materials 2018, 11, 586 14 of 14

33. Zhou, J.M.; Lucas, J.P. Hygrothermal Effects of Epoxy Resin. Part I: The Nature of Water in Epoxy. Polymer
1999, 40, 5505. [CrossRef]
34. Barrie, J.A.; Sagoo, P.S.; Johncock, P. The Sorption and Diffusion of Water in Epoxy Resins. J. Membr. Sci.
1984, 18, 197. [CrossRef]
35. Falk, M.; Ford, T.A. Infrared Spectrum and Structure of Liquid Water. Can. J. Chem. 1966, 44, 1699. [CrossRef]
36. Wiedemair, V.; Mayr, S.; Wimmer, D.S.; Köck, E.M.; Penner, S.; Kerstan, A.; Steinmassl, P.-A.;
Dumfahrt, H.; Huck, C.W. Novel Molecular Spectroscopic Multimethod Approach for Monitoring Water
Absorption/Desorption Kinetics of CAD/CAM Poly(Methyl Methacrylate) Prosthodontics. Appl. Spectrosc.
2017, 71, 1600. [CrossRef] [PubMed]
37. Ramirez, F.A.; Carlsson, L.A.; Acha, B.A. Evaluation of Water Degradation of Vinylester and Epoxy Matrix
Composites by Single Fiber and Composite Tests. J. Mater. Sci. 2008, 43, 5230. [CrossRef]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
APPENDIX B

PAPER II

KRAUKLIS A.E., ECHTERMEYER A.T.

MECHANISM OF YELLOWING: CARBONYL FORMATION DURING HYGROTHERMAL AGING

IN A COMMON AMINE EPOXY.

POLYMERS (SWITZERLAND), 10(9), 2018, 1017-1031.

DOI:10.3390/POLYM10091017

PAPER II

83
84
 polymers
Article
Mechanism of Yellowing: Carbonyl Formation during
Hygrothermal Aging in a Common Amine Epoxy
Andrey E. Krauklis * and Andreas T. Echtermeyer
Department of Mechanical and Industrial Engineering, Norwegian University of Science and Technology,
7491 Trondheim, Norway; [email protected]
* Correspondence: [email protected]; Tel.: +371-26-810-288

Received: 29 August 2018; Accepted: 12 September 2018; Published: 13 September 2018 

Abstract: Epoxies are often exposed to water due to rain and humid air environments. Epoxy
yellows during its service time under these conditions, even when protected from UV radiation.
The material’s color is not regained upon redrying, indicating irreversible aging mechanisms.
Understanding what causes a discoloration is of importance for applications where the visual aspect
of the material is significant. In this work, irreversible aging mechanisms and the cause of yellowing
were identified. Experiments were performed using a combination of FT-NIR, ATR-FT-IR, EDX,
HR-ICP-MS, pH measurements, optical microscopy, SEM, and DMTA. Such extensive material
characterization and structured logic of investigation, provided the necessary evidence to investigate
the long-term changes. No chain scission (hydrolysis or oxidation-induced) was present in the
studied common DGEBA/HDDGE/IPDA/POPA epoxy, whilst it was found that thermo-oxidation
and leaching occurred. Thermo-oxidation involved evolution of carbonyl groups in the polymeric
carbon–carbon backbone, via nucleophilic radical attack and minor crosslinking of the HDDGE
segments. Four probable reactive sites were identified, and respective reactions were proposed.
Compounds involved in leaching were identified to be epichlorohydrin and inorganic impurities
but were found to be unrelated to yellowing. Carbonyl formation in the epoxy backbone due to
thermo-oxidation was the cause for the yellowing of the material.

Keywords: epoxy; yellowing; mechanism; thermo-oxidation; carbonyl formation; leaching

1. Introduction
Epoxy resins are used worldwide as adhesives, as matrices in composites, as surface coatings,
as casting materials, and as laminating agents for artwork, longboards, and guitars [1–4]. Epoxies are
also used extensively for glass conservation [5,6], as adhesives and gap-fillers [7,8], and in preservation
of outdoor architectural and monumental stone as adhesives and as injection grouts for filling cracks,
as well as consolidants for porous, fragile deteriorated stone [9]. In most of these applications, stability
of color is of importance.
Epoxies are two-component systems consisting of an epoxy compound and a hardener, which react
to form the cured network [8]. In this study, the network of interest contains two epoxy compounds
(DGEBA and HDDGE) and two hardeners (IPDA and POPA). After reaction, a three-dimensional
amine epoxy network is obtained [10]. Epoxy resin family covers a large diversity of polymer
networks by the type of the epoxy compound and hardener employed. All these variations lead
to a different final network formed, and affect its physical and chemical properties [8,11]. In addition,
commercial products often contain such components as plasticizers, diluents, accelerators, and trace
impurities, which can affect the final properties and the yellowing [8,9]. Whilst DGEBA constitutes
more than 75% of the market [12], the number of articles concerning aging of this particular resin

Polymers 2018, 10, 1017; doi:10.3390/polym10091017 www.mdpi.com/journal/polymers

Polymers 2018, 10, 1017 2 of 15

(DGEBA/HDDGE/IPDA/POPA) is very limited [13,14]. Furthermore, yellowing is not investigated in

any of them.
Epoxy resins are often exposed to water due to rain, humidity of the air, and in subsea and
offshore applications [13,15]. In such conditions, water molecules can migrate into the polymer and
may affect its properties, and lead to leaching [1,16]. Epoxies tend to yellow even at room temperature,
even at medium humidity levels, with or without light exposure; it is a common observation [5,6,9].
However, light exposure can cause yellowing too, due to photo-yellowing (via photo-oxidation
mechanism) [5,6,8,9,17]. In one study, it was observed that when exposed to high temperatures epoxy
does not change its color in vacuum, while it does in air [18]. The yellowing phenomenon has often
precluded wider use of epoxies in the various aforementioned applications [5,9]. However, studies on
the yellowing of epoxy resins are few [5]. Furthermore, the identification of the mechanism of epoxy
yellowing would be valuable in choosing suitable conditions, compounds or additives, for increasing
service life of epoxies, i.e., in conservation [5,6].
Experimental evidence in this work was obtained using a combination of FT-NIR, ATR-FT-IR,
EDX, HR-ICP-MS, pH measurements, optical microscopy, SEM, and DMTA. Such extensive material
characterization, following the structured logic of investigation presented in this work, was novel and
provided the necessary evidence to investigate the long-term changes in chemical structure of the
studied resin.
The aim of this paper is to investigate the mechanistic origin of yellowing in a commonly used
DGEBA/HDDGE/IPDA/POPA amine epoxy resin.

2. Theory

2.1. Irreversible Degradation Mechanisms

The hygrothermal process may involve both reversible and irreversible processes [19,20].
Irreversible changes persist even after redrying the material [13]. The resin yellows irreversibly,
indicating that the mechanistic origin of the color change lies among irreversible degradation pathways.
The phenomenon of epoxy yellowing has been attributed to the photo-degradation of the amine
hardener, to the degradation of the amine epoxy network itself via various pathways, to the degradation
of additives or accelerators, and to the presence of impurities [9]. Degradation of the amine epoxy
network may follow such pathways as chain scission, crosslinking between segments, and thermal
and photo-oxidation of the main chains or sidegroups [18,21].
Irreversible aging mechanisms, which have been reported in the literature to occur during
hygrothermal influence on general epoxies are [13,20,22,23]:

1. Hydrolysis (involves chain scission)

2. Thermo-oxidation (might involve chain scission, backbone modifications or crosslinking)
3. Photo-oxidation (might involve chain scission, backbone modifications or crosslinking)
4. Residual curing (additional crosslinking)
5. Leaching (initially present additives, impurities or degradation products)

In this work, photo-oxidative effects were avoided by conducting experiments in the absence
of high-intensity light sources [5,6]. For the studied material, no hydrolysis occurred [14]. To avoid
residual curing, the material was fully cured as was indicated by the total disappearance of exothermal
signal via differential scanning calorimetry (DSC) [4,24]. Two runs were performed: The first run was
at 80 ◦ C for 16 h, and the second heating cycle was performed at 80 ◦ C for 1 h, showing that no further
hardening occurs. Based on this, hydrolysis, photo-oxidation, and residual curing were excluded,
whilst thermo-oxidation and leaching were investigated further, in respect to the yellowing.
Polymers 2018, 10, 1017 3 of 15

2.2. Thermo-Oxidation and Leaching

The chemical mechanism of epoxy thermo-oxidation is convoluted, and the exact degradation
chemistry remains the subject of ongoing work [25]. The process is complex as it involves
oxygen diffusion and consumption, and a radical chain mechanism initiated by hydroperoxide
decomposition [26]. The process seems to obey Arrhenius law, with activation energies around
60–80 kJ/mol, for various epoxies [25,26]. Thermo-oxidation may proceed differently for epoxies with
different structures and flexibility of the chains [27].
Thermo-oxidation of amine epoxies follows a general autoxidation mechanism, in which the
main source of radicals is the decomposition of hydroperoxides [4,25]. Processes often involved
during oxidation are chain scission, carbonyl formation, double bond formation, and amide
formation [4,11,23,28,29].
There is a limited number of articles regarding epoxy leaching [30–34]. Moreover, leaching
behavior of DGEBA/HDDGE/IPDA/POPA epoxy is not investigated in any of them. The leaching
phenomenon may occur due to initially present additives, impurities, or degradation products diffusing
out from the epoxy network into the water environment, which is in contact with the polymer. Often
but not always, it follows Fickian type diffusion [30]. Commonly reported epoxy leaching compounds
are bisphenol A or epichlorohydrin [30–33]. The driving force of this process is expected to be due to
the difference in concentration of these chemicals inside the resin, and in the aqueous environment.

3. Materials and Methods

3.1. Materials
Amine-cured epoxy resin was prepared by mixing reagents Epikote Resin RIMR 135 (Hexion,
Columbus, OH, USA) and Epikure Curing Agent MGS RIMH 137 (Hexion, Columbus, OH, USA)
stoichiometrically, in a ratio of 100:30 by weight. The mixture was degassed in a vacuum chamber
for 0.5 h to remove bubbles. The samples were cured at room temperature for 24 h, and post-cured
in an air oven (Lehmkuhls Verksteder, Oslo, Norway) at 80 ◦ C for 16 h. Full cure was achieved as
described above in Reference [24]. The samples were cast into rectangular moulds and then cut into
40 × 7 × 2 mm3 rectangular samples with a vertical bandsaw. Sample preparation was followed
by grinding with sandpaper (FEPA P60, grain size 269 μm). The sample geometry was chosen in
accordance with standard practice for glass transition temperature determination, as described in
Reference [35]. The dimensions were achieved within 5% tolerance.
Resin and hardener (the epoxy system) consist of the following compounds: 0.63 wt % Bisphenol
A diglycidyl ether (DGEBA; CAS 1675-54-3; number average molecular weight ≤ 700); 0.14 wt %
1,6-hexanediol diglycidyl ether (HDDGE; CAS 16096-31-4); 0.14 wt % poly(oxypropylene)diamine
(POPA; CAS 9046-10-0; molecular weight 230); and 0.09 wt % isophorondiamine (IPDA; CAS 2855-13-2).
Chemical structures of these compounds are shown in Figure 1.

Figure 1. Molecular structures of epoxy and hardener components: (A) DGEBA monomer; (B) DGEBA
oligomer (n = 1–2); (C) HDDGE; (D) POPA; (E) IPDA.
Polymers 2018, 10, 1017 4 of 15

The distilled water (resistivity 0.5–1.0 mΩ·cm) was used for conditioning, produced via the water
purification system Aquatron A4000 (Cole-Parmer, Vernon Hills, IL, USA). pH of distilled water was
determined to be 5.65, being slightly acidic as water is equilibrated with the atmospheric CO2 .

3.2. Experimental Methods

3.2.1. Overview
The hygrothermal conditions provide means to the accelerated aging, to study the effect
of thermo-oxidation and leaching on the yellowing of epoxy. As was mentioned elsewhere in
Reference [36], aging studies are intended to accelerate the degradation chemistry. The experimental
evidence on the irreversible degradation mechanisms was obtained and reported in this work.
The analysis of yellowing and morphological characterization was performed using a combination
of visual inspection, optical microscopy, and scanning electron microscopy (SEM, Tescan, Brno, Czech
Republic). Chemical composition and macromolecular changes were studied using a combination of
FT-NIR, ATR-FT-IR, EDX, HR-ICP-MS, pH measurements, and DMTA.

3.2.2. Conditioning of Resin Samples in Distilled Water

Water uptake experiments were conducted using a batch system. A heated distilled water
(60 ± 1 ◦ C) bath was used for conditioning the samples. Samples were weighed using analytical
scales AG204 (± 0.1 mg; Mettler Toledo, Columbus, OH, USA). Samples were conditioned for a period
of two months. Samples were taken out of the water bath, weighed, and analyzed using a FT-NIR
method [37].

3.2.3. Drying of Conditioned Resin Samples

The drying of saturated samples was performed in a drying cabinet PK-410 (ESAB, London, UK),
at 60 ± 1 ◦ C in air atmosphere, with natural convection and relative humidity of 13 RH%. After that,
samples were reconditioned at ambient conditions in the air to regain their initial water content.

3.2.4. Optical Microscopy

Optical microscopy was performed using a digital microscope RH-2000 (Hirox, Tokyo, Japan),
equipped with lens MXB-2500REZ, with a magnification of 140, and resolution of 1.06 μm.

3.2.5. SEM and EDX

Scanning electron microscopy (SEM) and Energy-dispersive X-ray spectroscopy (EDX)
experiments were performed using Mira/LMU (Tescan, Brno, Czech Republic) in backscattered
electron regime, with working voltage of 15 kV.

3.2.6. FT-NIR
Near-infrared (NIR) spectra were obtained using a Fourier transform spectrophotometer
NIRSystems 6500 (Foss, Eden Prairie, MN, USA) operated in a transmission mode; an optical fiber probe
and spectral analysis software Vision (Foss, Eden Prairie, MN, USA) was used. Spectra were taken in
Vis-NIR wavenumber range of 4000–25,000 cm−1 , using 32 scans per spectrum with a resolution of
4 cm−1 .
FT-NIR spectroscopy was used to determine that the initial and redried epoxy has the same water
content [37].

3.2.7. ATR-FT-IR
Fourier transform infrared (FT-IR) spectra were recorded using Scimitar 800 FT-IR (Varian,
Inc., Palo Alto, CA, USA) in the Attenuated Total Reflectance (ATR) mode via GladiATRTM (Pike
Polymers 2018, 10, 1017 5 of 15

Technologies, Fitchburg, WI, USA). Spectra were obtained at 4 cm−1 resolution, co-adding 50 scans
over a range of wavenumbers from 400 to 4000 cm−1 .

3.2.8. HR-ICP-MS
High resolution inductively coupled plasma mass spectrometry (HR-ICP-MS) analyses were
performed using a double focusing magnetic sector field HR-ICP-MS Finnigan ELEMENT 2 (Thermo
Fisher Scientific, International, Waltham, MA USA). A sample introduction system PrepFAST
(ESI/Elemental Scientific, Omaha, NE, USA) was used. Pretreatment/digestion was done using
UltraClave (Milestone, Milan, Italy). Acidification of samples was performed using ultra-pure grade
HNO3 SubPur (Milestone, Milan, Italy), to avoid adsorption of ions to the wall of the vial.

3.2.9. pH Measurements
pH measurements were performed using standard pH-meter MeterLab PHM210 (Radiometer
analytical, Lyon, France) (pH ± 0.01). IUPAC standard buffer solutions (Radiometer analytical, Lyon,
France) were used for calibration of pH meter.

3.2.10. DMTA
Dynamic Mechanical Thermal Analysis (DMTA) tests, for determination of glass transition
temperature, storage, and loss moduli were conducted using a Netzsch GABO qualimeter Eplexor,
equipped with a 1.5 kN load cell (Netzsch GABO Instruments, Ahlden, Germany) operated in
displacement control with a constant static strain of 0.4%, and a cyclic strain of 0.1% applied with
a frequency of 1 Hz. The temperature sweep range was from 20 up to 120 ◦ C, with a heating rate
of 1 ◦ C/min. The glass transition temperature (Tg ) was determined using DMTA as the crossing of
tangents to the inflection points in the storage modulus curves [13,35].

4. Results

4.1. Discoloration and Changes in Morphology

A difference in color of unaged versus aged epoxy is shown in Figure 2. The discoloration persists
even after drying and reconditioning in air. Thus, the change in color was irreversible.

Figure 2. Visual inspection of an epoxy resin showing discoloration due to hygrothermal aging.
Top: initial (blueish grey); bottom: redried after conditioning (yellow).

Yellowing was not a surface phenomenon, since change in color occurred homogeneously in
the bulk, as was suggested by uniform color in the cross-section of the sample cut in the middle.
This agreed with an observation made in another study [5].
Polymers 2018, 10, 1017 6 of 15

Digital optical and SEM micrographs indicated that there was no observable change in
morphology after hygrothermal aging and reconditioning to the initial water content, as shown in
Figure 3. Surface morphology was studied for both aged and unaged material. It showed no significant
changes. No pores or cavities were observed, suggesting no loss of polymer from the surface.

Figure 3. Digital optical (top) and SEM (bottom) micrographs of initial dry (left) and redried after
hygrothermal aging (right).

4.2. Changes in Chemical Composition

Obtained FT-NIR spectra of initial epoxy and resin redried to initial water content are shown in
Figure 4.

Figure 4. FT-NIR spectra of initial resin and redried epoxy after hygrothermal aging. (Left): visible
light region. (Right): NIR region.

Visible light region spectra indicated clearly, the yellowing and reduced transparency of the aged
epoxy. In the NIR region of the spectra, both initial and redried epoxy after aging were similar
with an exception of the peak at around 4535 cm−1 , which corresponds to the epoxy ring [38].
It was observed that after hygrothermal aging, this peak had reduced dramatically, indicating that
Polymers 2018, 10, 1017 7 of 15

a compound, containing an epoxy ring, either leached or reacted. Since the resin was fully cured,
the authors believe this peak corresponded to the leached unreacted epichlorohydrin, which contained
an epoxy ring in its chemical structure. This claim was further supported with EDX and HR-ICP-MS.
Obtained ATR-FT-IR spectra of the initial dry epoxy and epoxy after hygrothermal process, drying
and conditioning in air to initial water content, are shown in Figure 5.

Figure 5. ATR-FT-IR spectra of initial (bottom) and redried after hygrothermal process resin (top).

Obtained spectra indicated that there was no significant difference observed in chemical structure
of the initial and dried samples, except for the peak at 1736 cm−1 , corresponding to carbonyl groups
(νC=O) [4,8,10,28,39,40].
Elemental analysis via EDX indicated that initial dry epoxy had a higher content of Cl
(0.99 ± 0.08 Cl%) than epoxy, after hygrothermal aging (0.73 ± 0.05 Cl%). This suggested leaching
of chlorine-containing compounds from the resin into water during hygrothermal aging. This could
be due to unreacted epichlorohydrin being released, which comes from the epoxy component.
Moreover, oxygen content in the aged epoxy (23.59 ± 0.42 O%) was lower than in the initial one
(24.35 ± 0.45 O%), indicating most likely a leaching of an epoxy ring-containing compound, such as
unreacted epichlorohydrin.
Elements (Ca, K, Na, Cl, S) were identified to be leached by the resin into contacting water during
hygrothermal aging using HR-ICP-MS. The element release curves are shown in Figure 6, with Cl
release being dominant.

Figure 6. Ca, Cl, K, Na and S release from neat resin during hygrothermal aging at 60 ◦ C.
Polymers 2018, 10, 1017 8 of 15

pH measurement results of the water (50 mL samples) in contact with the resin during
hygrothermal aging are shown in Figure 7. There was a strong initial decrease in pH upon contact
of the dry resin with distilled water. The authors believe this could be related to a release of acidic
impurities from the resin, i.e., HCl.

Figure 7. pH measurements of distilled water samples after contact with the resin.

The changes at macromolecular scale have been investigated by Tg change. The glass transition
temperature (Tg ) of the initial, saturated, and redried epoxy was 81.7, 59.1, and 84.7 ◦ C, respectively.
Tg for saturated epoxy was much lower than for the dry and dried resin, due to plasticization [14].
The Tg of a redried epoxy was slightly higher than that of initial resin. In case of chain scission, a Tg
value would have decreased [41], which was not observed. A slight increase in Tg of the dried material
was likely due to polymer relaxation [13], minor thermo-oxidative crosslinking [7,27], or leaching of
plasticizing compounds. The tensile storage and loss moduli in a temperature sweep range from 20 to
120 ◦ C are presented in Figure 8.

Figure 8. Temperature sweep for glass transition temperature determination for initial (dry); saturated;
and redried. (Left): temperature sweep of tensile storage modulus. (Right): temperature sweep of
tensile loss modulus.

5. Discussion

5.1. Logic of the Investigation

The logic which is followed in this study is shown schematically in Figure 9.
Polymers 2018, 10, 1017 9 of 15

Figure 9. Schematic representation of the logic during investigation.

5.2. Leaching
There are three types of leaching that are potentially possible: (1) Leaching of hardener; (2) leaching
of epoxy compounds; and (3) leaching of impurities or additives.
In theory, it is possible that some amount of unreacted hardener (blue in color), since soluble in
water, would be washed out from crosslinked polymer network, or would be used up in additional
crosslinking. The initial uncured epoxy resin is yellowish in color, which could explain the yellowing
of the product over time [42]. The residual crosslinking, can also cause the decrease in unreacted amine
group concentration causing the change in color [13], but this was not the case since the material was
fully cured. Hardener leaching was not present, as supported by ATR-FT-IR. If the hardener washout
or residual crosslinking was the case, it would show a decrease in peak intensity at wavenumber
corresponding to unreacted amine groups at 3200–3500 cm−1 [43], which was not observed in the
ATR-FT-IR spectra. It should also be noted that any changes to the epoxy material below IR sensitivity
cannot be detected.
Whilst it was found that leaching of hardener is improbable, leaching of chemicals that are
initially present in epoxy resin, such as epichlorohydrin, was found likely, using a combination of EDX,
HR-ICP-MS, and FT-NIR. Based on the HR-ICP-MS data and FT-NIR intensity of the corresponding
peak at 4535 cm−1 , the leached amount of epichlorohydrin was estimated to be 75 μg/g of resin, and
the initial concentration of epichlorohydrin in the resin was estimated to be 137 μg/g.
HR-ICP-MS showed leaching of Cl-containing compounds or chloride ions. It should be noted that
the DGEBA epoxy component itself, is a product of O-alkylation of Bisphenol A and epichlorohydrin.
Upon such reaction, HCl is released. However, the reaction is conducted in the presence of sodium
hydroxide NaOH, meaning that HCl is being neutralized [44]. It is possible that some of the Cl
compounds, i.e., NaCl, unneutralized HCl or unreacted epichlorohydrin, were present in the initial dry
resin. It is likely that these compounds were washed out from the epoxy system during hygrothermal
aging. Furthermore, the pH measurements (Figure 7) indicate a release of acidic compounds from the
resin, i.e., HCl.
To sum up, the leaching of hardener was not observed, whilst the leaching of epoxy compounds,
such as epichlorohydrin, and leaching of impurities, such as HCl and NaCl, was present.
Polymers 2018, 10, 1017 10 of 15

5.3. Thermo-Oxidation
Thermo-oxidation is highly selective [45], and the main source of radicals is the decomposition of
the most thermally unstable chemical species [26]. Each carbon atom in α position of an electronegative
atom, such as O and N, has a decreased dissociation energy and increased reactivity in the radical
chain propagation, thus resulting in “weak points” in the network structure [4,28].
Increase in the intensity of the carbonyl group peak (νC=O; 1730–1740 cm−1 ) [4,8,10,28,39,40],
for aged epoxy, is related to an oxidation process taking place. Carbonyl formation can result from
various pathways [28]. Thermo-oxidation involving chain scission and appearance of double bonds;
carbonyl and amide species have been reported in DGEBA/IPDA and DGEBA/POPA systems [4,21,41].
Appearance of diphenylketones was reported in another work [46]. In this work, the appearance of
amide species (C=O; 1644 cm−1 and N-H; 3325 cm−1 ) [8,43] and diphenylketones (C=O; 1660 cm−1 ) [46]
due to thermo-oxidation was not observed. Double bond formation was not observed, as indicated by
C=C bands close to 1600 cm−1 (1624 cm−1 and 1593 cm−1 ) [28].
Chemical structure of a cured amine epoxy network (not considering the mixing proportions),
is shown in Figure 10. “Weak points” for radical attack in the network were identified and are
shown in Figure 10. In the network, 12 unique sites potentially involved in thermo-oxidation were
found. Furthermore, 8 of these sites were excluded based on experimental evidence and literature.
Sites marked in green were the identified main reactive sites (δ+ DGEBA-II, δ+ POPA-I, δ+ HDDGE-III,
and δ+ HDDGE-IV).

Figure 10. Chemical structure of the studied DGEBA/HDDGE/IPDA/POPA amine epoxy network
(mixing ratios are not considered). Marked sites represent “weak points” for radical attack in the
network. Sites marked in red are excluded based on experimental evidence and literature. Sites marked
in green are the main reactive sites.

Tg can be regarded as a useful parameter revealing chemical changes for polymers [47]. The Tg
of a redried epoxy (84.7 ◦ C) was slightly higher than for the initial material (81.7 ◦ C), indicating that
no chain scission occurred [7,41]. A likely reason of a Tg increase was a combination of polymer
relaxation [13], anti-plasticizing effect of leaching, and a minor crosslinking [7,27] of the HDDGE
segments (site δ+ HDDGE-IV). The thermo-oxidative crosslinking mechanism has been reported for
Polymers 2018, 10, 1017 11 of 15

an amine epoxy with a curing agent DGEBU similar to HDDGE in another work [27]. An analogous
crosslinking mechanism of HDDGE segments is proposed in Scheme 1, involving sites δ+ HDDGE-IV.

Scheme 1. Crosslinking reaction of the HDDGE segments via reactive sites δ+ HDDGE-IV.

The only way to form amides is an oxidation of amino methylenes near to network nodes [28].
There was no increase in intensity of the bands corresponding to amide species (C=O; 1644 cm−1
and N-H; 3325 cm−1 ) [8,43]. This excluded the following “weak points” as potential reactive sites:
δ+ DGEBA-IV, δ+ POPA-II, δ+ HDDGE-I, δ+ IPDA-I, and δ+ IPDA-II. Moreover, the formation of amide
species is linked to chain scission, which was not present, as indicated by Tg measurements.
The linkage between the aromatic rings may also be sensitive to oxidation [10,25]. Carbonyl
formation via acetophenone groups due to radical attack on δ+ DGEBA-I site was not observed,
as indicated by the absence of the increase of the acetophenone “in-chain” group band (1684 cm−1 ) [10]
and diphenylketones (C=O; 1660 cm−1 ) [46].
The formation of carbonyl groups may result from oxidation of the secondary alcohol groups in
cured resin [28,46]. This oxidation process is accompanied by the decrease of the band at 1237 cm−1 ,
representing the characteristic C-O band in secondary alcohol groups [46]. No changes to this band
were observed, which excluded the following “weak points” as potential reactive sites: δ− DGEBA-III
and δ− HDDGE-II. Furthermore, in the hydroxyl domain (≈3800–2500 cm−1 ) [10] no changes were
observed, which further supported this conclusion.
Noteworthy that all three reactive sites (δ+ DGEBA-II, δ+ POPA-I, and δ+ HDDGE-III) involved in
carbonyl formation have similar structures. Whilst δ+ POPA-I contains the polyoxypropylene moiety,
the other two sites (δ+ DGEBA-II and δ+ HDDGE-III) are identical and contain the i-propanol moiety.
δ+ POPA-I site as a polyoxypropylene moiety-containing segment is very susceptible to radical
attack under oxidation, due to low stability of the tertiary C-H bond and the destabilizing effect of the
neighboring ether group [10,41]. The proposed carbonyl formation reaction on this site is shown in
Scheme 2.

Scheme 2. Carbonyl formation involving polyoxypropylene moiety on reactive site δ+ POPA-I.

The δ+ DGEBA-II and δ+ HDDGE-III sites as i-propanol moiety-containing segments are also highly
susceptible to oxidation [4,28]. The carbonyl formation on these two sites follows the same reaction,
as shown in Scheme 3 [28].
Contrary to the DGEBA/IPDA and DGEBA/POPA binary amine epoxies [4,21,42], the results
suggest that a combination of DGEBA/HDDGE/IPDA/POPA stops the oxidative chain scission and
amide formation. It is not clear what exactly causes such change, but it is possible that some of the
identified reactive sites operate as weak “sacrificial” centers, as suggested in another work [17], thus
protecting the remaining structure.
Polymers 2018, 10, 1017 12 of 15

To sum up, the results indicated that there was no chain scission, double bonds and amide groups
were not formed, whilst the evolution of carbonyl groups in the macromolecular backbone and minor
crosslinking of HDDGE segments occurred.

Scheme 3. Carbonyl formation reaction involving i-propanol moiety on reactive sites δ+ DGEBA-II and
δ+ HDDGE-III.

5.4. The Cause of Yellowing

The change in color is irreversible, and is related to the irreversible aging mechanism.
Two irreversible phenomena were identified: (1) Leaching of an epoxy compound epichlorohydrin
and impurities, i.e., HCl and NaCl; (2) Formation of carbonyl groups C=O in the polymeric backbone
due to thermo-oxidation (sites δ+ DGEBA-II, δ+ POPA-I, and δ+ HDDGE-III, shown in Figure 10).
Both impurities and epichlorohydrin are colorless and do not cause discoloration. The oxidative
evolution of carbonyl groups in the resin was the reason for the yellowing. It has been reported for
other polymers that carbonyl formation can cause a change in color [48,49]. Furthermore, the yellowing
phenomenon of a polyurethane resin was found to be linked to the mechanism of carbonyl formation
in the macromolecular backbone, caused by oxidation [49].

5.5. Increasing Epoxy Service Life

Yellowing of the studied epoxy was linked to the formation of carbonyl groups in the
macromolecular backbone. Based on other studies [25,50,51], conclusion can be drawn that carbonyl
formation can be slowed down by using phenolic antioxidants, such as hindered phenols, which are
used as stabilizers for various plastics and rubbers [50,51]. These compounds act as radical scavengers
and can prevent thermo-oxidative yellowing. An effective way of introducing stabilization for epoxies,
when mechanical properties are not concerned, is a co-curing procedure of epoxy with resole, which
has hindered phenol moieties [50,51].

5.6. On the Similarity of Yellowing and Thermo-Oxidation Kinetics

Kinetics of epoxy yellowing and thermo-oxidative carbonyl formation found in literature,
undoubtedly show common trends [6,25,28,41].
Yellowing of epoxies is not linear in time and is characterized by three stages [6]: (1) The induction
period with little or no yellowing [6]; oxidation kinetics also display an induction period with a strongly
auto-accelerated character [28]; (2) The steady state period, during which yellowing is high and
constant [6]; (3) The declining rate period, during which the increase in yellowing is occurring
at a slower rate [6], which is in agreement with another study on thermo-oxidation, stating that
high oxidation levels can result in reduced sensitivity to further oxidation [25]. In the long term,
in some cases, there is a horizontal asymptote in carbonyl formation kinetics, indicating equilibrium or
a saturation phase, but it is not systematically observed [41]. The kinetics have a strong dependence on
temperature and presence of antioxidants [28]; the kinetics follow the Arrhenius principle over a wide
temperature range [25,26].

6. Conclusions
This work discussed the mechanism of yellowing of a DGEBA/HDDGE/IPDA/POPA amine
epoxy. Based on the results of FT-NIR, ATR-FT-IR, EDX, HR-ICP-MS, pH measurements, optical
microscopy, SEM, and DMTA, the following conclusions have been made:
Polymers 2018, 10, 1017 13 of 15

1. Yellowing occurred due to the thermo-oxidative carbonyl formation in the epoxy carbon-carbon
backbone via nucleophilic radical attack. The change in color was irreversible. Morphology was
found to be unaffected.
2. No chain scission (hydrolysis or oxidation-induced) was present, whilst thermo-oxidation and
leaching occurred.
3. Compounds involved in leaching were identified to be epichlorohydrin and inorganic impurities
but were unrelated to yellowing.
4. Four unique reactive sites responsible for thermo-oxidation were found. One reactive site was
involved in minor thermo-oxidative crosslinking of the HDDGE segments, while three other
sites were linked to carbonyl formation. Noteworthy that all three sites involved in carbonyl
formation had similar structures, containing highly reactive polyoxypropylene and i-propanol
moieties. Respective reactions were proposed.
5. It is speculated that yellowing could be prevented or delayed by adding phenolic antioxidants,
such as hindered phenols.

Author Contributions: Conceptualization, A.E.K. and A.T.E.; methodology, A.E.K.; formal analysis, A.E.K.;
investigation, A.E.K.; resources, A.E.K. and A.T.E.; data curation, A.E.K. and A.T.E.; writing–original draft
preparation, A.E.K.; writing–review and editing, A.E.K. and A.T.E.; validation, A.E.K.; visualization, A.E.K.;
supervision, A.T.E.; project administration, A.T.E.; funding acquisition, A.T.E.
Funding: This research was funded by The Research Council of Norway (Project 245606/E30 in the Petromaks
2 programme).
Acknowledgments: This work is part of the DNV GL led Joint Industry Project “Affordable Composites” with
twelve industrial partners and the Norwegian University of Science and Technology (NTNU). The authors would
like to express their thanks for the financial support from The Research Council of Norway (Project 245606/E30 in
the Petromaks 2 programme). Authors are thankful to Ilze Kalnin, a, Kristı̄ne Rugele,
‘
Lı̄ga Stı̄pniece, Søren Heinze,
Abedin I. Gagani, Bjørn Kåre Alsberg, Emeric Mialon, Cristian Torres Rodriguez, Anton G. Akulichev, and Syverin
Lierhagen. Andrey is especially grateful to Oksana V. Golubova.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Maggana, C.; Pissis, P. Water sorption and diffusion studies in an epoxy resin system. J. Polym. Sci. Part B
1999, 37, 1165–1182. [CrossRef]
2. Popineau, S.; Rondeau-Mouro, C.; Sulpice-Gaillet, C.; Shanahan, M.E.R. Free/Bound water absorption in
an epoxy adhesive. Polymer 2005, 46, 10733–10740. [CrossRef]
3. Chiang, C.-L.; Ma, C.-C.M.; Wang, F.-Y.; Kuan, H.-C. Thermo-oxidative degradation of novel epoxy containing
silicon and phosphorous nanocomposites. Eur. Polym. J. 2003, 39, 825–830. [CrossRef]
4. Ernault, E.; Richaud, E.; Fayolle, B. Thermal oxidation of epoxies: Influence of diamine hardener.
Polym. Degrad. Stab. 2016, 134, 76–86. [CrossRef]
5. Down, J.L. The Yellowing of Epoxy Resin Adhesives: Report on High-Intensity Light Aging. Stud. Conserv.
1986, 31, 159–170. [CrossRef]
6. Down, J.L. The yellowing of epoxy resin adhesives: Report on natural dark aging. Stud. Conserv. 1984, 29,
63–76.
7. Coutinho, I.; Ramos, A.M.; Lima, A.M.; Fernandes, F.B. Studies of the degradation of epoxy resins used for
the conservation of glass. In Proceedings of the Holding it all together, Ancient and Modern Approaches to
Joining, Repair and Consolidation, London, UK, 21–22 February 2008. [CrossRef]
8. Tennent, N.H. Clear and Pigmented Epoxy Resins for Stained Glass Conservation: Light Ageing Studies.
Stud. Conserv. 1979, 24, 153–164.
9. Ginell, W.S.; Coffman, R. Epoxy resin-consolidated stone: Appearance change on aging. Stud. Conserv. 1998,
43, 242–248.
10. Mailhot, B.; Morlat-Thérias, S.; Ouahioune, M.; Gardette, J.-L. Study of the Degradation of an Epoxy/Amine
Resin, 1 Photo- and Thermo-Chemical Mechanisms. Macromol. Chem. Phys. 2005, 206, 575–584. [CrossRef]
Polymers 2018, 10, 1017 14 of 15

11. Ernault, E.; Richaud, E.; Fayolle, B. Origin of epoxies embrittlement during oxidative ageing. Polym. Test.
2017, 63, 448–454. [CrossRef]
12. Pham, H.Q.; Marks, M.J. Epoxy resins. In Ullmann’s Encyclopedia of Industrial Chemistry; Wiley-VCH:
Weinheim, Germany, 2005; pp. 155–244.
13. Rocha, I.B.C.M.; Raijmaekers, S.; Nijssen, R.P.L.; van der Meer, F.P.; Sluys, L.J. Hygrothermal ageing behaviour
of a glass/epoxy composite used in wind turbine blades. J. Compos. Struct. 2017, 174, 110–122. [CrossRef]
14. Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Hygrothermal Aging of Amine Epoxy: Reversible Static and
Fatigue Properties. Open Eng. 2018. under review.
15. Startsev, V.O.; Lebedev, M.P.; Khrulev, K.A.; Molokov, M.V.; Frolov, A.S.; Nizina, T.A. Effect of outdoor
exposure on the moisture diffusion and mechanical properties of epoxy polymers. Polym. Test. 2018, 65,
281–296. [CrossRef]
16. Toscano, A.; Pitarresi, G.; Scafidi, M.; Di Filippo, M.; Spadaro, G.; Alessi, S. Water diffusion and swelling
stresses in highly crosslinked epoxy matrices. Polym. Degrad. Stab. 2016, 133, 255–263. [CrossRef]
17. Allen, N.S.; Robinson, P.J.; White, N.J.; Swales, D.W. Photo-oxidative Stability of Electron Beam and UV Cured
Acrylated Epoxy and Urethane Acrylate Resin Films. Polym. Degrad. Stab. 1987, 19, 147–160. [CrossRef]
18. Buch, X.; Shanahan, M.E.R. Thermal and thermo-oxidative ageing of an epoxy adhesive. Polym. Degrad. Stab.
2000, 68, 403–411. [CrossRef]
19. Wang, M. The hygrothermal aging process and mechanism of the novolac epoxy resin. Compos. Part B 2016,
107, 1–8. [CrossRef]
20. Clancy, T.C.; Frankland, S.J.V.; Hinkley, J.A.; Gates, T.S. Molecular modeling for calculation of mechanical
properties of epoxies with moisture ingress. Polymer 2009, 50, 2736–2742. [CrossRef]
21. López-Ballester, E.; Doménech-Carbó, M.T.; Gimeno-Adelantado, J.V.; Bosch-Reig, F. Study of FT-IR
spectroscopy of ageing of adhesives used in restoration of archaeological glass objects. J. Mol. Struct.
1999, 482, 525–531. [CrossRef]
22. Xiao, G.Z.; Shanahan, M.E.R. Irreversible effects of hygrothermal aging on DGEBA/DDA epoxy resin. J. Appl.
Polym. Sci. 1998, 69, 363–369. [CrossRef]
23. Belec, L.; Nguyen, T.H.; Nguyen, D.L.; Chailan, J.F. Comparative effects of humid tropical weathering and
artificial ageing on a model composite properties from nano- to macro-scale. Compos. Part A Appl. Sci. Manuf.
2015, 68, 235–241. [CrossRef]
24. Heinze, S.; NTNU, Trondheim, Norway. Personal communication, 2017.
25. Celina, M.C.; Dayile, A.R.; Quintana, A. A perspective on the inherent oxidation sensitivity of epoxy
materials. Polymer 2013, 54, 3290–3296. [CrossRef]
26. Colin, X.; Verdu, J. Thermal ageing and lifetime prediction for organic matrix composites. Plast. Rubber
Compos. 2003, 32, 349–356. [CrossRef]
27. Ernault, E.; Richaud, E.; Fayolle, B. Thermal-oxidation of epoxy/amine followed by glass transition
temperature changes. Polym. Degrad. Stab. 2017, 138, 82–90. [CrossRef]
28. Galant, C.; Fayolle, B.; Kuntz, M.; Verdu, J. Thermal and radio-oxidation of epoxy coatings. Prog. Org. Coat.
2010, 69, 322–329. [CrossRef]
29. Bellenger, V.; Verdu, J.; Francilette, J.; Hoarau, P.; Morel, E. Infra-red study of hydrogen bonding in
amine-crosslinked epoxies. Polymer 1987, 28, 1079–1086. [CrossRef]
30. Bruchet, A.; Elyasmino, N.; Decottignies, V.; Noyon, N. Leaching of bisphenol A and F from new and old
epoxy coatings: Laboratory and field studies. Water Sci. Technol. 2014, 14, 383–389. [CrossRef]
31. Rajasärkkä, J.; Pernica, M.; Kuta, J.; Lašňák, J.; Šimek, Z.; Bláha, L. Drinking water contaminants from epoxy
resin-coated pipes: A field study. Water Res. 2016, 103, 133–140. [CrossRef] [PubMed]
32. Lipke, U.; Haverkamp, J.B.; Zapf, T.; Lipperheide, C. Matrix effect on leaching of Bisphenol A diglycidyl
ether (BADGE) from epoxy resin based inner lacquer of aluminium tubes into semi-solid dosage forms.
Eur. J. Pharm. Biopharm. 2016, 101, 1–8. [CrossRef] [PubMed]
33. Vermeirssen, E.L.M.; Dietschweiler, C.; Werner, I.; Burkhardt, M. Corrosion protection products as a source
of bisphenol A and toxicity to the aquatic environment. Water Res. 2017, 123, 586–593. [CrossRef] [PubMed]
34. Xiao, G.Z.; Delamar, M.; Shanahan, M.E.R. Irreversible interactions between water and DGEBA/DDA epoxy
resin during hygrothermal aging. J. Appl. Polym. Sci. 1997, 65, 449–458. [CrossRef]
35. Plastics in the Determination of Dynamic Mechanical Properties–Part 11: Glass Transition Temperature; International
Standard ISO 6721-11:2012(E); ISO: Geneva, Switzerland, 2012.
Polymers 2018, 10, 1017 15 of 15

36. Celina, M.C. Review of polymer oxidation and its relationship with materials performance and lifetime
prediction. Polym. Degrad. Stab. 2013, 98, 2419–2429. [CrossRef]
37. Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Near-infrared spectroscopic method for monitoring water
content in epoxy resins and fiber-reinforced composites. Materials 2018, 11, 586. [CrossRef] [PubMed]
38. Lyon, R.E.; Chike, K.E.; Angel, S.M. In situ Cure Monitoring of Epoxy Resins Using Fiber-Optic Raman
Spectroscopy. J. Appl. Polym. Sci. 1994, 53, 1805–1812. [CrossRef]
39. De’Néve, B.; Shanahan, M.E.R. Water absorption by an epoxy resin and its effect on the mechanical properties
and infra-red spectra. Polymer 1993, 34, 5099–5105. [CrossRef]
40. Larché, J.-F.; Bussiére, P.-O.; Thérias, S.; Gardette, J.-L. Photooxidation of polymers: Relating material
properties to chemical changes. Polym. Degrad. Stab. 2012, 97, 25–34. [CrossRef]
41. Zahra, Y.; Djouani, F.; Fayolle, B.; Kuntz, M.; Verdu, J. Thermo-oxidative aging of epoxy coating systems.
Prog. Org. Coat. 2014, 77, 380–387. [CrossRef]
42. Hexion Technical Data Sheet: EPIKOTE RIMR 135 and EPIKURE RIMH 134–137. 2006. Available online:
http://www.hexion.com/en-us/chemistry/epoxy-resins-curing-agents-modifiers/epoxy-tds (accessed
on 1 August 2018).
43. Deng, T.; Liu, Y.; Cui, X.; Yang, Y.; Jia, S.; Wang, Y.; Lu, C.; Li, D.; Cai, R.; Hou, X. Cleavage of C–N bonds in
carbon fiber/epoxy resin composites. Green Chem. 2015, 17, 2141–2145. [CrossRef]
44. Unnikrishnan, K.P. Studies on the Toughening of Epoxy Resins. Ph.D. Thesis, Cochin University of Science
and Technology, Kochi, India, 2006.
45. Rasoldier, N.; Colin, X.; Verdu, J.; Bocquet, M.; Olivier, L.; Chocinski-Arnault, L.; Lafarie-Frenot, M.C. Model
systems for thermo-oxidised epoxy composite matrices. Compos. Part A Appl. Sci. Manuf. 2008, 39, 1522–1529.
[CrossRef]
46. Li, K.; Wang, K.; Zhan, M.-S.; Xu, W. The change of thermal-mechanical properties and chemical structure
of ambient cured DGEBA/TEPA under accelerated thermo-oxidative aging. Polym. Degrad. Stab. 2013, 98,
2340–2346. [CrossRef]
47. Sauvant-Moynot, V.; Duval, S.; Grenier, J. Innovative pipe coating material and process for high temperature
fields. Oil Gas Sci. Technol. 2002, 57, 269–279. [CrossRef]
48. Yousif, E.; Haddad, R. Photodegradation and photostabilization of polymers, especially polystyrene: Review.
SpringerPlus 2013, 2, 398–429. [CrossRef] [PubMed]
49. Rosu, D.; Rosu, L.; Cascaval, C.N. IR-change and yellowing of polyurethane as a result of UV irradiation.
Polym. Degrad. Stab. 2009, 94, 591–596. [CrossRef]
50. Lin, M.-S.; Chiu, C.-C. Protection of epoxy resin against thermo-oxidation via co-curing epoxy/resole (I).
Polym. Degrad. Stab. 2000, 69, 251–253. [CrossRef]
51. Lin, M.-S.; Chiu, C.-C. Protection of epoxy resin against thermo-oxidation via co-curing epoxy/resole (II).
Polym. Degrad. Stab. 2001, 71, 327–329. [CrossRef]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
APPENDIX C

PAPER III

KRAUKLIS A.E., GAGANI A.I., ECHTERMEYER A.T.

HYGROTHERMAL AGING OF AMINE EPOXY: REVERSIBLE STATIC AND FATIGUE

PROPERTIES.

OPEN ENGINEERING (POLAND), 8(1), 2018, 447-454.

DOI:10.1515/ENG-2018-0050

PAPER III

101
102
 Open Eng. 2018; 8:447–454

Research Article Open Access

Andrey E. Krauklis*, Abedin I. Gagani, and Andreas T. Echtermeyer

Hygrothermal Aging of Amine Epoxy:

Reversible Static and Fatigue Properties
https://doi.org/10.1515/eng-2018-0050 are widely used as matrices for composite materials, i.e.
Received August 1, 2018; accepted October 19, 2018 fiber-reinforced polymers (FRP), as well as adhesives, or-
ganic surface coatings and encapsulating agents [1, 4].
Abstract: Fiber-reinforced polymers (FRP) are widely used
FRPs are often exposed to water or humid air environ-
in structural applications. Long-term properties of such
ments, where water molecules can migrate in the poly-
materials exposed to water are of high concern and inter-
meric matrix and modify their physical and mechani-
est, especially for subsea and offshore applications. The
cal properties [1]. Highly crosslinked amine-cured epoxy
objective of this study is to identify the mechanisms and
resins are hydrophilic and their properties can signifi-
to identify whether drop in properties of diamine-cured
cantly deteriorate upon water uptake [2, 5–13]. A degra-
mixed DGEBA-HDDGE is reversible upon drying the ma-
dation of the matrix-dominated properties of epoxy-based
terial to its initial water content. The properties of inter-
glass fiber composites is also expected when they are ex-
est are mechanical strength, elastic properties and fatigue
posed to and saturated with water [14, 15]. Such effects
performance, as well as changes in chemical structure.
are of special interest for offshore and marine industries,
The effect of absorbed water on the properties of the resin
but also for many other industries, such as the renewable
is evaluated, and hygrothermal effects and aging mecha-
energy sector, i.e. wind turbines, where FRPs are widely
nisms are discussed. Furthermore, it is shown experimen-
used [16, 17]. The number of articles concerning aging of
tally that the tension fatigue S-N curve of a wet epoxy resin
this particular mixed resin (DGEBA/HDDGE/IPDA/POPA)
can be estimated by shifting the S-N curve of a dry material
is very limited [17].
proportionally to a reduction in static tensile strength due
Water has a double effect on the polymer networks de-
to hygrothermal effects.
pending on its concentration [12]. At the beginning of the
Keywords: epoxy; hygrothermal aging; plasticization; hygrothermal process, absorbed water causes the relax-
strength; fatigue ation of residual stress and the acceleration of additional
crosslinking, both of which may be responsible for the in-
crease of the mechanical properties [12, 14, 18]. Tg can be
regarded as the most useful parameter revealing material
1 Introduction
degradation [19]. The more flexible the polymer chains,
the lower the Tg [12]. A decrease in Tg is generally at-
Epoxy resins are well known for their relatively high
tributed to plasticization and deterioration (i.e. chain scis-
strength, stiffness, low volatility, chemical resistance, and
sion), while an increase in Tg is derived from additional
low shrinkage on curing [1, 2]. Glycidyl ether derivatives of
crosslinking [14]. The degradation of the tensile strength
bisphenol A, i.e. DGEBA, are the most widely used epoxy
can be attributed to the plasticization and deterioration of
resins in structural applications and constitute more than
the resin [14, 20]. The percentage reduction in the tensile
75% of epoxy resins sold worldwide [3]. These epoxy resins
strength of epoxies is related to the hydrophilicity of the
resin blend, which may be measured by Hoy’s solubility
parameter for hydrogen bonding [21]. Fatigue life reduc-
*Corresponding Author: Andrey E. Krauklis: Department of Me- tion is still not completely understood, and while some au-
chanical and Industrial Engineering (past: Department of Engineer- thors report significant degradation, others do not observe
ing Design and Materials), Norwegian University of Science and noticeable changes after water absorption [17].
Technology, 7491 Trondheim, Norway, Tel: +371 26 810 288, E-mail: Hygrothermal process may induce both reversible and
[email protected]
irreversible changes in the epoxy [14, 22]. Irreversible
Abedin I. Gagani, Andreas T. Echtermeyer: Department of Me-
chanical and Industrial Engineering (past: Department of Engineer- changes persist after re-drying the material [17]. It seems
ing Design and Materials), Norwegian University of Science and that chemical processes involved in hygrothermal aging of
Technology, 7491 Trondheim, Norway

Open Access. © 2018 Andrey E. Krauklis et al., published by De Gruyter. This work is licensed under the Creative Commons
Attribution-NonCommercial-NoDerivs 4.0 License. Authenticated | [email protected]
Download Date | 11/25/18 9:16 PM
448 | Andrey E. Krauklis, Abedin I. Gagani, and Andreas T. Echtermeyer

epoxies are generally irreversible and persist even after the before pouring. The specimens were cured at room temper-
material is dried [17]. Irreversible damage in epoxies may ature for 24 h and post-cured in an air oven (Lehmkuhls
occur due to susceptibility of the polymer to hydrolysis, Verksteder, Norway) at 80 ◦ C for 16 h. The specimens were
oxidation, and change of the effective average crosslinked removed from the moulds’ grooves and cut into the desired
molecular weight. For some epoxy systems, water at el- length with a vertical bandsaw. After cutting the epoxy
evated temperatures can attack the crosslinked network, specimens were ground using sandpaper (FEPA P60, grain
causing chain scission and the leaching of segments [23]. size 269 μm). The specified dimensions from ISO 6721 and
Crosslinking reactions can also continue in an epoxy over ISO 527 [26, 27] were achieved within 5% tolerance.
time resulting in increased stiffness, since water causes an
acceleration of the resin-hardener reaction [22, 24].
V.O. Startsev et al studied the effect of outdoor expo-
sure on epoxies and assessed reversible effects of mois-
ture. They have found that strength of the studied epoxies
decreased by 20-40% due to plasticization [25].
Plasticization, due to water uptake, is commonly
placed in the physical aging category. Authors believe that
this is not fully correct, and, instead, offer a distinction be-
tween hygrothermal aging and hygrothermal effects. Ag- Figure 1: Geometry of dogbone specimens used for static tensile
and fatigue tests. The placement of strain gauges is indicated.
ing can be defined as processes which cause changes in
material properties and that are directly time-dependent.
Processes which are also dependent on other terms, such The placement of strain gauges as shown in Figure 1
as concentration (e.g. related to mass transport phenom- allowed to measure strains in both the direction of the ap-
ena), are termed as hygrothermal effects, where other plied load and the direction normal to it, thus enabling the
terms might or might not be time-dependent themselves. calculation of Poisson’s ratio.
The aim of this paper is to investigate whether
static and fatigue properties of a commonly used
DGEBA/HDDGE/IPDA/POPA amine epoxy resin are
reversible or not after re-drying the material. 3 Experimental
3.1 Conditioning of resin specimens in
2 Materials distilled water and drying

The epoxy specimens were tested dry, saturated in water

Amine-cured epoxy resin was prepared using reagents
and re-dried. Specimens were conditioned in a heated dis-
Epikote Resin RIMR 135 (Hexion) and Epikure Curing Agent
tilled water (60 ± 1 ◦ C) bath for a period of two months
MGS RIMH 137 (Hexion) in a ratio of 100:30 by weight (sto-
(equilibrium was achieved after a few weeks; see Fig-
ichiometrically). The epoxy value of the resin is 0.54-0.60
ure 2). The water content was measured by FT-NIR spec-
equivalent/100 g. The amine value of the hardener is 400-
troscopy [28] and by the more widely used weight gain
600 mg KOH/g [24]. The mixture was degassed in a vacuum
method [29]. Specimens were weighed using analytical
chamber for 0.5 h in order to remove bubbles. The epoxy
scales AG204 (± 0.1 mg; Mettler Toledo, USA). Both meth-
system consists of the following compounds: bisphe-
ods gave about the same results, but the FT-NIR method
nol A diglycidyl ether (DGEBA), 1,6-hexanediol diglycidyl
has the advantage of providing a true water content using
ether (HDDGE), poly(oxypropylene)diamine (POPA) and
the method described in another paper [28]. The true water
isophorondiamine (IPDA).
content shows the amount of water in respect to the abso-
The resin was casted into moulds to make rectan-
lutely dry material and is defined in Equation 1.
gular DMTA (40 × 7 × 2 mm3 ) and dogbone-shaped
(200 × 30 × 2 mm3 with 20 mm width in the most narrow m water (t)
W * (t) ≡ · 100%
part) specimens according to ISO 6721 and ISO 527 [26, 27]. m absolutely dry
 
The geometry of dogbone specimens equipped with strain m (t) − m absolutely dry
= · 100% (1)
gauges (gauge length 6 mm; Tokyo Sokki Kenkyujo Co., m absolutely dry
Ltd., Japan) is shown in Figure 1. The resin was degassed

Authenticated | [email protected]
Download Date | 11/25/18 9:16 PM
 Hygrothermal Aging of Amine Epoxy | 449

where m water is a moisture mass, m absolutely dry is a mass of 3.3 Mechanical testing
resin with 0% moisture content, m is a total mass of resin
with moisture. Tensile tests were conducted using a servo hydraulic
Already at the initial stage, the ‘dry’ material has a test machine Instron model 1342 (Instron, International:
true water content of about 0.63% [28], meaning that it has USA/UK). the displacement rate was set to 1 mm/min. Fa-
some initial water present being in equilibrium with wa- tigue tests were done using the same equipment. The test-
ter vapour in the air. The evolution of true water content ing frequency was chosen in order to keep a constant strain
in time during hygrothermal conditioning, drying and re- rate of 0.05%/min. Tests were performed at R ratio of 0.1.
conditioning in air atmosphere to initial water content is The temperature during the tests was about 23 ◦ C (room
presented in Figure 2. temperature). Tensile tests were performed with 4 speci-
mens for each configuration (dry, wet and dried). Average
values and experimental scatter were reported for ultimate
tensile strength, Young’s modulus and Poisson’s ratio for
each group. In fatigue, between 11 and 13 dogbone spec-
imens were used for obtaining S-N curves for each case
(dry, wet and dried).

4 Results and discussion

4.1 Change of mechanical properties due to
Figure 2: Evolution of true water content of amine-cured epoxy resin water uptake
during water uptake, drying and conditioning in air.
In order to investigate the influence of the hygrothermal
process and water content on the mechanical properties,
The drying of saturated specimens was performed in a
static tensile and tension fatigue tests were performed.
drying cabinet PK-410 (ESAB, UK) at 60 ± 1 ◦ C in air atmo-
sphere with natural convection and relative humidity of 13
RH%.
4.1.1 Reversibility of mechanical properties in static
tension

3.2 Glass transition temperature Static tensile test results are illustrated in stress-strain
determination curves shown in Figure 3. The dry material has an ulti-
mate tensile strength (UTS) of 60.5 ± 2.7 MPa. All scatter
Dynamic Mechanical Thermal Analysis (DMTA) tests is given as one standard deviation. After the hygrothermal
for determination of glass transition temperature (Tg ) aging process, the UTS decreased to 48.5 ± 3.3 MPa, result-
were conducted using Netzsch GABO qualimeter Eplexor ing in a relative decrease of about 20% on average. Results
equipped with a 1.5 kN load cell (Netzsch GABO Instru- indicate that the material regained its initial strength and
ments, Germany) operated in tension in displacement con- Young’s modulus after redrying (and conditioning in air to
trol with a constant static strain of 0.4% and a cyclic strain its initial water content). It should be noted, that dried ma-
of 0.1% applied with a frequency of 1 Hz. The temperature terial has a slightly higher strength (66.4 ± 3.0 MPa) than
sweep range was from 20 up to 120 ◦ C with a heating rate the initial (60.5 ± 2.7 MPa), although the effect is within the
of 1 ◦ C/min. Tg is determined as the point of inflection in standard deviations. A possible explanation to this phe-
the temperature sweep of the storage modulus plot [26]. nomenon could be residual crosslinking or polymer relax-
Evaporation of water for the wet specimens was avoided ation [17]. It should be noted that not only the UTS of the
by conducting measurements in the water environment, dry and re-dried specimens were the same, but also the en-
ensuring that the specimens have been saturated during tire shape of the stress strain curve was very similar.
the whole process. Poisson’s ratios are reported in Figure 5. The Poisson’s
ratio was calculated as the ratio of strains measured via
strain gauges in directions normal to applied load and in

Authenticated | [email protected]
Download Date | 11/25/18 9:16 PM
450 | Andrey E. Krauklis, Abedin I. Gagani, and Andreas T. Echtermeyer

Figure 3: Stress-strain curves of dry, conditioned and dried epoxy

Figure 4: Placement and failure of the specimen in the test machine.
specimens.

the direction of tension. Results show that Poisson’s ra-

tio increases for saturated epoxy but returns to the initial
values after drying to the initial water content, indicating
that this effect is reversible. An increase in Poisson’s ratio
of epoxy due to absorbed water is consistent with litera-
ture [30]. The authors believe that such increase in Pois-
son’s ratio might be due to the absorbed almost incom-
pressible water or the reduction of glass transition temper-
ature (Tg ) due to the plasticizing action [30,31]. Theocaris
has shown that the Poisson’s ratio of epoxy increases with
higher plasticizer content even at room temperature below
Figure 5: Poisson’s ratio of initial (dry), wet and dried epoxy speci-
Tg [30]. The Tg of dry, saturated and re-dried epoxy was
mens.
81.7, 59.1 and 84.7 ◦ C, respectively.
The strain to failure (ϵ max ) is identical for dry and
dried epoxy, being 3.71 ± 0.10% and 3.69 ± 0.22% respec-
tively, while being much higher for saturated epoxy (4.78 ±
0.51%). Dogbone specimens failed in the middle as shown
in Figure 4. It was a brittle fracture and shattering often
occurred for dry and re-dried specimens, while failure was
less brittle for saturated epoxy. On the micro level failure
is believed to occur due to a combination of crosslink bond
breakage and disentanglement of macromolecular chains.
The specimen placement in the machine along with the
failure are shown in Figure 4.
Mechanical and elastic properties of dry, wet and
Figure 6: S-N curves of initial (dry), saturated and dried epoxy speci-
dried epoxy are shown in Table 1. The average values mens in log-log scales.
of ultimate tensile strength (UTS), Young’s modulus (E),
strain to failure (ϵ max ) and Poisson’s ratios are reported
along with the statistics indicating experimental scatter 4.1.2 Reversibility of mechanical properties in tension
(Table 1). fatigue

Tension fatigue results are shown in Figure 6 and Figure 7

in log S-log N and linear S-N scales, respectively, where
maximum stress S is in MPa and number of cycles to failure
N is unitless. The R ratio was 0.1. The results of the fatigue

Authenticated | [email protected]
Download Date | 11/25/18 9:16 PM
 Hygrothermal Aging of Amine Epoxy | 451

Table 1: Mechanical and elastic properties of dry, wet and dried epoxy.

Material UTS [MPa] E [MPa] ϵ max [%] Poisson’s ratio [-]

Dry 60.5 ± 2.7 2908.8 ± 126.6 3.71 ± 0.10 0.35 ± 0.02
Dried 66.4 ± 3.0 3237.5 ± 126.3 3.69 ± 0.22 0.36 ± 0.01
Wet 48.5 ± 3.3 2588.9 ± 252.3 4.78 ± 0.51 0.39 ± 0.01

swelling-induced microcracking and formation of microc-

racks through absorption/desorption cycles. Multiple ag-
ing mechanisms and hygrothermal effects may also com-
bine.
Results of this study indicate that plasticization (hy-
grothermal effect) is fully reversible for the studied epoxy.
No chain scission was observed. The Tg of dry, sat-
urated and re-dried epoxy was 81.7, 59.1 and 84.7 ◦ C, re-
spectively. The Tg of saturated specimen decreased due to
plasticization. There was only a slight increase in Tg after
re-drying compared to the dry epoxy, indicating that no
Figure 7: S-N curves of initial (dry), saturated and dried epoxy speci- major changes in chemical structure have occurred dur-
mens in linear scales. ing hygrothermal aging. In case of chemical degradation
(chain scission) a Tg value would have decreased, which
is opposite to what was observed. A slight increase in Tg
tests show that the S-N curve of the wet epoxy shifted down
of the dried material can be explained due to additional
by 20% without a change in slope. The slope was obtained
crosslinking taking place or polymer relaxation [14, 17].
by linear regression of the log-log curve with the number
of cycles being the dependent variable. Regression equa-
tions are shown in Table 2.
The fact that the slope of the S-N curve did not change, 6 Conclusions
and the drop is the same as for the static data indicates that
fatigue of wet epoxy can be simply predicted by measuring Exposure of the amine-cured epoxy to water caused the
static strength changes of wet specimens. material to swell. The absorption of water had detrimental,
The S-N curve of the re-dried epoxy was the same but reversible effects on static strength and cyclic fatigue
(within experimental scatter) as the original S-N curve of properties. Similar values of dry and re-dried specimen Tg
the dry material. This effect indicates reversibility, as was serve as an indicator that no chemical degradation (chain
already observed for the static data. scission) has occurred.
The ultimate tensile strength of the resin decreased by
20% relative to the initial dry value due to plasticization
by water. The strength of dried epoxy is comparable to that
5 Aging mechanisms and of the initial dry epoxy, indicating the reversibility in me-
hygrothermal effects chanical properties.
Poisson’s ratio increased for conditioned epoxy, but
For epoxy-amine networks [17], two possible major hy- returned to the initial values after drying, also indicating
grothermal degradation mechanisms might occur: plas- that the effect of water conditioning is reversible. The in-
ticization (hygrothermal effect) and chain scission (hy- crease of Poisson’s ratio is believed to be due to the ab-
grothermal aging). Other failure mechanisms may also sorbed almost incompressible water or due to the reduc-
happen [14, 15, 17, 22, 32–34], but are considered of mi- tion of glass transition temperature (Tg ) due to the plas-
nor importance to this study, such as thermo-oxidation, ticizing action. The results of fatigue tests indicated that
photo-oxidation, additional crosslinking (residual cur- the S-N curve of a wet epoxy also shifted by 20% without a
ing), secondary crosslinking between the epoxy and wa- change in slope. The S-N curves of dry and dried material
ter molecules, leaching of additives, polymer relaxation,

Authenticated | [email protected]
Download Date | 11/25/18 9:16 PM
452 | Andrey E. Krauklis, Abedin I. Gagani, and Andreas T. Echtermeyer

Table 2: Linear regression of fatigue data.

Linear models Regression equation Determination coefficient R2

Dry (linear) log N = -11.5207 log S + 21.5115 0.89
Dried (linear) log N = -11.3895 log S + 21.3121 0.82
Wet (linear) log N = -11.9332 log S + 21.0585 0.95
Wet (linear predicted from static data) log N = -11.5207 log S + 20.3952 0.90

were identical, further proving the reversibility in mechan- M.E.R., Free/bound water absorption in an epoxy adhesive,
ical properties. Polymer, 2005, 46(24), 10733–10740.
It is shown experimentally that the tension fatigue S-N [5] Wang J., Gong J., Gong Z., Yan X., Wang B., Wu Q., Li S., Effect
of curing agent polarity on water absorption and free volume in
curve of a wet epoxy resin can be estimated by shifting the
epoxy resin studied by PALS, Nucl. Instrum. Methods Phys. Res.
S-N curve of a dry material proportionally to a reduction in B., 2010, 268(14), 2355–2361.
static tensile strength due to hygrothermal effects. [6] Weitsman Y., Coupled damage and moisture-transport in fiber-
reinforced, polymeric composites, Int. J. Solid Struct., 1987,
Acknowledgement: This paper is part of the DNV GL led 23(7), 1003–1025.
[7] Weitsman Y.J., Elahi M., Effects of fluids on the deformation,
Joint Industry Project “Affordable Composites” with twelve
strength and durability of polymeric composites – an overview,
industrial partners and the Norwegian University of Sci- Mech. Time-Depend. Mater., 2000, 4(2), 107–126.
ence and Technology (NTNU). The authors would like to [8] Roy S., Moisture-induced degradation, in Long-Term Durability
express their thanks for the financial support by The Re- of Polymeric Matrix Composites, 2012, Eds.: Pochiraju K., Tan-
search Council of Norway (Project 245606/E30 in the Petro- don G., Schoeppner G., Springer: Boston, MA.
maks 2 programme). Authors are thankful to Erik Sæter, [9] Apicella A., Nicolais L., Effect of water on the properties of epoxy
matrix and composite, Adv. Polym. Sci., 1985, 72, 69–77.
Carl-Magnus Midtbø, Emeric Mialon, Anton G. Akulichev
[10] Lefebvre D.R., Elliker P.R., Takahashi K.M., Raju V.R., Kaplan
and Lars H. Østengen who contributed to this work to M.L., The critical humidity effect in the adhesion of epoxy to
some extent. Andrey is especially grateful to Oksana V. glass: role of hydrogen bonding, J. Adhes. Sci. Technol., 2000,
Golubova. 14, 925–937.
[11] Guermazi N., Elleuch K., Ayedi H.F., The effect of time and ag-
ing temperature on structural and mechanical properties of
pipeline coating, Mater. Design., 2009, 30, 2006–2010.
List of symbols [12] Wu C.F., Xu W.J., Atomistic simulation study of absorbed wa-
W* Moisture content in the resin (w%)
ter influence on structure and properties of crosslinked epoxy
mwater Moisture mass in the resin (g)
resin, Polymer, 2007, 48, 5440–5448.
mabsolutelydry Mass of an absolutely dry resin specimen (g)
Measured mass of a resin specimen (g)
[13] Li L., Yu Y., Wu Q., Zhan G., Li S., Effect of chemical structure
m
Tg Glass transition temperature (◦ C) on the water sorption of amine-cured epoxy resins, Corros. Sci.,
t Time (s) 2009, 51, 3000–3006.
ϵ max Linear strain to failure (%) [14] Wang M., The hygrothermal aging process and mechanism of
E Young’s modulus (MPa) the novolac epoxy resin, Composites Part B, 2016, 107, 1–8.
S Stress (MPa) [15] Halpin J.C., Effects of environmental factors on composite mate-
N Number of cycles before specimen fails in fatigue (-) rials, Technical Report AFML-TR67-423, 1969, Air Force Materials
UTS Ultimate tensile strength (MPa) Laboratory, US, Ohio.
DMTA Dynamic mechanical testing analysis
[16] Echtermeyer A.T., Integrating durability, in Durability of Com-
FT-NIR Fourier transform near infrared spectroscopy
posites in a Marine Environment, 2014, Eds.: Davies P., Ra-
japakse Y.D.S., Springer: Dordrecht.
[17] Rocha I.B.C.M., Raijmaekers S., Nijssen R.P.L., van der Meer
F.P., Sluys L.J., Hygrothermal ageing behaviour of a glass/epoxy
References composite used in wind turbine blades, J. Compos. Struct.,
2017, 174, 110–122.
[1] Maggana C., Pissis P., Water sorption and diffusion studies in [18] Xian G., Karbhari V.M., Segmental relaxation of water-aged am-
an epoxy resin system, J. Polym. Sci., Part B: Polym. Phys., 1999, bient cured epoxy, Polym. Degrad. Stab., 2007, 92(9), 1650–
37(11), 1165–1182. 1659.
[2] Lee M.C., Peppas N.A., Water transport in epoxy-resins, Prog. [19] Sauvant-Moynot V., Duval S., Grenier J., Innovative pipe coating
Polym. Sci., 1993, 18(5), 947–961. material and process for high temperature fields, Oil Gas Sci.
[3] Pham H.Q., Marks M.J., Epoxy resins, in Ullmann’s encyclopedia Technol., 2002, 57(3), 269–279.
of industrial chemistry, 2005, Wiley-VCH. [20] Chen Y., Davalos J.F., Ray I., Kim H.-Y., Accelerated aging tests for
[4] Popineau S., Rondeau-Mouro C., Sulpice-Gaillet C., Shanahan

Authenticated | [email protected]
Download Date | 11/25/18 9:16 PM
 Hygrothermal Aging of Amine Epoxy | 453

evaluations of durability performance of FRP reinforcing bars for [27] International Standard ISO 527-1:2012(E), Plastics – Determina-
concrete structures, Compos. Struct., 2007, 78(1), 101–111. tion of tensile properties – Part 1: General principles, 2012.
[21] Yiu C.K.Y., King N.M., Pashley D.H., Suh B.I., Carvalho R.M., Car- [28] Krauklis A.E., Gagani A.I., Echtermeyer A.T., Near-infrared spec-
rilho M.R.O., Tay F.R., Effect of resin hydrophilicity and water troscopic method for monitoring water content in epoxy resins
storage on resin strength, Biomaterials, 2004, 25(26), 5789– and fiber-reinforced composites, Materials, 2018, 11(4), 586-
5796. 599.
[22] Clancy T.C., Frankland S.J.V., Hinkley J.A., Gates T.S., Molecu- [29] International Standard ISO 62-2008(E), Plastics – Determina-
lar modeling for calculation of mechanical properties of epoxies tion of water absorption, 2008.
with moisture ingress, Polymer, 2009, 50(12), 2736–2742. [30] Theocaris P.S., Influence of plasticizer on Poisson’s ratio of
[23] Xiao G.Z., Shanahan M.E.R., Irreversible effects of hygrothermal epoxy polymers, Polymer, 1979, 20, 1149-1154.
aging on DGEBA/DDA epoxy resin, J. Appl. Polym. Sci., 1998, [31] Fine R.A., Millero F.J., Compressibility of water as a function of
69(2), 363–369. temperature and pressure, J. Chem. Phys., 1973, 59(10), 5529-
[24] Technical data sheet, EPIKOTE RIMR 135 and EPIKURE RIMH 134– 5536.
137, 2006, Hexion. [32] Xiao G.Z., Shanahan M.E.R., Swelling of DGEBA/DDA epoxy
[25] Startsev V.O., Lebedev M.P., Khrulev K.A., Molokov M.V., Frolov resin during hygrothermal ageing, Polymer, 1998, 39(14), 3253–
A.S., Nizina T.A., Effect of outdoor exposure on the moisture 3260.
diffusion and mechanical properties of epoxy polymers, Polym. [33] Belec L., Nguyen T.H., Nguyen D.L., Chailan J.F., Comparative
Test., 2018, 65, 281-296. effects of humid tropical weathering and artificial ageing on a
[26] International Standard ISO 6721-11:2012(E), Plastics – Determi- model composite properties from nano- to macro-scale, Com-
nation of dynamic mechanical properties – Part 11: Glass tran- posites Part A, 2015, 68, 235–241.
sition temperature, 2012. [34] Alfrey, Jr. T., Gurnee E.F., Lloyd W.G., Diffusion in glass polymers,
J. Polym. Sci. Part C, 1966, 12(1), 249–261.

Authenticated | [email protected]
Download Date | 11/25/18 9:16 PM
454 | Andrey E. Krauklis, Abedin I. Gagani, and Andreas T. Echtermeyer

A Appendix Fatigue raw data

Table 3: Fatigue raw data. S represents maximum stress in MPa. The R ratio was 0.1.

DRY DRIED WET

S S S
Specimen N [-] Specimen N [-] Specimen N [-]
[MPa] [MPa] [MPa]
1 21 56.0 1 158 50.1 1 139 38.4
2 53 48.9 2 53 48.0 2 311 36.0
3 1515 42.9 3 2210 43.5 3 286 35.4
4 489 40.7 4 4986 42.0 4 974 34.5
5 1455 40.7 5 1155 40.5 5 1524 33.8
6 4392 39.9 6 1798 37.5 6 1169 33.2
7 2711 39.6 7 3171 33.0 7 686 32.1
8 1451 38.3 8 22265 30.0 8 9219 27.8
9 5896 36.1 9 54185 28.5 9 10911 27.5
10 2643 34.1 10 14538 28.0 10 3543 26.9
11 13183 31.6 11 109947 27.0 11 7710 26.0
12 111785 28.5 12 138162 21.6
13 148408 22.5 13 381758 19.8

Authenticated | [email protected]
Download Date | 11/25/18 9:16 PM
APPENDIX D

PAPER IV

KRAUKLIS A.E., GAGANI A.I., ECHTERMEYER A.T.

PREDICTION OF ORTHOTROPIC HYGROSCOPIC SWELLING OF FIBER-REINFORCED

COMPOSITES FROM ISOTROPIC SWELLING OF MATRIX POLYMER.

JOURNAL OF COMPOSITES SCIENCE (SWITZERLAND), 3(1), 2019, 10-23.

DOI:10.3390/JCS3010010

PAPER IV

111
112
Article
Prediction of Orthotropic Hygroscopic Swelling of
Fiber-Reinforced Composites from Isotropic
Swelling of Matrix Polymer
Andrey E. Krauklis * , Abedin I. Gagani and Andreas T. Echtermeyer
Department of Mechanical and Industrial Engineering (past: Department of Engineering Design and Materials),
Norwegian University of Science and Technology, 7491 Trondheim, Norway; [email protected] (A.I.G.);
[email protected] (A.T.E.)
* Correspondence: [email protected] or [email protected]; Tel.: +371-268-10288

Received: 12 December 2018; Accepted: 8 January 2019; Published: 12 January 2019 

Abstract: Swelling in fiber-reinforced composites is anisotropic. In this work, dealing with glass fiber
epoxy composite immersed in distilled water, swelling coefficients are obtained in each direction
experimentally. Swelling behaviour in the fiber direction was constrained by the non-swelling fibers
and was close to null, while swelling in the transverse directions was found to occur freely—similar to
the unconstrained polymer. An analytical method for predicting anisotropic swelling in composites
from the swelling of the matrix polymer is reported in this work. The method has an advantage that
it is simple to use in practice and requires only a swelling coefficient of the matrix polymer, elastic
constants of the matrix and fibers, and a known fiber volume fraction of the composite. The method
was validated using finite element analysis. Good agreement was obtained and is reported between
experimental hygroscopic swelling data, analytical and numerical results for composite laminates,
indicating the validity of this predictive approach.

Keywords: epoxy; composites; hygroscopic; swelling; hygrothermal; finite element analysis

1. Introduction
Fiber-reinforced composite (fiber-reinforced polymer; FRP) laminates are used for structural
applications in marine, offshore and oil and gas industries due to their light weight and corrosion
resistance [1–3]. Composites offshore have been implemented in such applications as risers, tethers,
repair patches and ship hulls [4–8]. In these applications, FRPs are exposed to water and experience
subsequent water-induced or hygroscopic swelling [1–3].
One of the main effects of water on the mechanical property deterioration of polymers is swelling,
even more so if the polymer is not affected by hydrolysis or chain scission, such as the epoxy in
this study [1,9]. Swelling is a specific response accompanying moisture diffusion in polymers and
polymer-based composites [2]. Susceptibility of polymers to swelling results in a two-fold effect on
FRPs: on the one hand, it causes a decrease in mechanical strength of the polymeric matrix [9], while on
the other, it results in swelling stresses when the hygroscopic swelling is restrained [10].
In FRPs, the matrix is constricted by fibers, and as a result, this affects the swelling
behaviour. What complicates the phenomenon even more is the orthotropic nature of swelling
of composites—fibers, such as glass or carbon, do not swell, while the polymer does [3,10].
Such incompatible swelling behaviour in FRPs leads to swelling stresses at the interfaces,
which may lead to microcrack formation, especially under transient conditions (non-uniform moisture
content distribution) [3,11]. Hygroscopic swelling may affect the mechanical properties of FRPs
significantly [2,9,12,13]. Thus, it is important to know not only the moisture diffusion behaviour,

J. Compos. Sci. 2019, 3, 10; doi:10.3390/jcs3010010 www.mdpi.com/journal/jcs

J. Compos. Sci. 2019, 3, 10 2 of 14

but also the swelling behaviour, in order to properly characterize the FRP material property change,
i.e., strength or modulus, resulting from moisture absorption. The focus of this work is on swelling
and its orthotropic nature in FRPs.
The amount of hygroscopic strain found from dimensional change is normally assumed to be
linearly proportional to the moisture concentration as follows (Equation (1)) [14]:

ε h = βW (1)

where ε h is the hygroscopic strain, β is the coefficient of hygroscopic expansion (CHE), and W is the
moisture concentration.
Linear strain behaviour has been observed experimentally with increasing moisture concentration
for both composites and polymers [13,14]. For orthotropic laminates, three CHEs (β x , β y , β z ) are needed
in order to predict swelling. Strains in longitudinal (along-the-fibers) direction are often assumed to be
null for composites with moisture-insensitive fibers, i.e., glass or carbon [15].
Quantification of the orthotropic CHEs can be performed experimentally using samples with
different fiber orientations. However, it is a time-consuming and tedious process that also tends to
involve quite high experimental scatters. The industrial interest lies in the reduction of testing time
and testing-related expenses. Thus, a modelling approach to swelling of FRPs due to the effects of
water (and also other liquids, such as oil) is of interest [16].
Various studies have been performed on swelling of FRPs [10,13,17–25] and, more recently,
on hygroscopic swelling in textile composites [26,27]. The works available in the literature have
addressed several aspects of hygroscopic swelling in composites, from the nature of swelling in
polymeric matrix [3,10,28,29], to the influence of swelling on the fluid diffusion in polymers [13,16,23,
30,31], to the development of micromechanical models to predict transverse swelling [17,26,27].
Ashton et al. [18] suggested the use of thermal expansion theory when dealing with hygroscopic
swelling: in particular the Halpin-Tsai equation [19] for fluid diffusion and Schapery equation [20] for
hygroscopic swelling of composites. Coran et al. [21] and Daniels [22] analyzed swelling of reinforced
rubbers by means of thermodynamic theory of elasticity, confirming an orthotropic swelling behaviour
and relating the elastic constants to the swelling constants of the rubber composite. Fan et al. [23]
modelled a coupled diffusion and swelling of fiber-reinforced composites. Meng et al. [25] developed a
multiscale model for coupled moisture diffusion and swelling in FRPs by means of finite element (FE)
analysis. The model enabled evaluation of the fiber-matrix interfacial stresses [25]. Sinchuk et al. [26]
developed a realistic voxel-model simulating fluid diffusion and hygroscopic swelling in a textile
composite using a mesoscale approach of modelling the orthotropic tows and the resin-rich areas.
In a later work, Sinchuk et al. [27] developed a hierarchical multiscale model for the prediction of
hygroscopic swelling-induced stresses in textile composites.
An interesting opportunity would be the possibility to predict the orthotropic swelling constants
(CHEs) of the composite from the CHE of the matrix polymer, which is isotropic. The matrix properties
are easy to measure. Furthermore, they also may be found in literature for various polymers [28,32].
However, in some cases, the interfacial effects may not be negligible [33], many of the moisture-related
properties of composites are known to be traceable to those of the matrix material [12,13]. Swelling
strains of a composite and a matrix polymer should also be related to each other through a proper
analysis [13,34]. Since swelling in polymers does not follow the ideal mixing law [2], i.e., the volume
increase of the polymer is not equal to the volume of the absorbed water, and it is necessary to perform
swelling experiments for the matrix polymer itself, or to find polymer CHE in the literature [28,32].
The composite swelling can then be analytically or numerically predicted from the swelling of the
matrix polymer, as shown in this work.
Few works have investigated the anisotropic nature of swelling in unidirectional FRPs. In this
work, the authors developed an analytical model based on linear elasticity. Both the analytical model
and a finite element analysis employing a periodic representative volume element (RVE) were able to
predict the orthotropic hygroscopic swelling of fiber-reinforced composites from the isotropic swelling
J. Compos. Sci. 2019, 3, 10 3 of 14

of the polymer used as a matrix. The model was validated with experiments on glass fiber-reinforced
epoxy composites, using the matrix isotropic CHE to predict directional swelling of the composite,
and yielded a good agreement. To the best knowledge of the authors, this is the first micromechanical
model that predicts the anisotropic swelling of composites from isotropic swelling of the polymer.

2. Materials and Methods

2.1. Materials
A typical glass fiber epoxy used for marine and oil and gas applications was selected for this
study. HiPer-TexTM fabrics weaved by 3B Fibreglass (Birkeland, Norway) were used as a glass fiber
reinforcement with an average fiber diameter of 17 μm. The density of glass (ρ f ) was 2.54 g/cm3 .
HexionTM (Columbus, OH, USA) epoxy resin RIMR135TM and amine hardener RIMH137TM were used
for preparing the matrix polymer by mixing in a stoichiometric ratio of 100:30 by weight. The resin
and the curing agent consisted of the following compounds: bisphenol A diglycidyl ether (DGEBA),
1,6-hexanediol diglycidyl ether (HDDGE), poly(oxypropylene)diamine (POPA) and isophorondiamine
(IPDA). The density of the polymer (ρm ) was 1.1 g/cm3 . Before pouring, the mixture was degassed
in a vacuum chamber for 0.5 h in order to remove bubbles. Degassed resin was moulded into
rectangular-shaped moulds, followed by curing at room temperature for 24 h and post-curing in
an air oven (Lehmkuhls Verksteder, Norway) at 80 ◦ C for 16 h. After samples were post-cured,
the polymer was cut into rectangular bars and then further cut into plates using a vertical bandsaw.
The final dimensions of plates were 25 mm × 25 mm × 2 mm. The desired thickness was obtained
using PHOENIX 2000 (Jean Wirtz, Germany) and SiC grinding discs (Struers, Cleveland, OH, USA;
FEPA P500, grain size 30 μm). The sufficient thickness control, correct length and width was ensured
within a 5% tolerance.
Glass fiber-reinforced epoxy composite laminates were prepared using vacuum assisted resin
transfer moulding (VARTM). The same epoxy resin was used as for preparation of polymer samples.
The composite laminate was cut into rectangular bars and subsequently into plates with different fiber
orientations (C1 and C3 as shown in Figure 1). The dimensions of plates were 25 mm × 25 mm × 2
mm. The thickness was adjusted to 2 mm within 5% tolerance via grinding with a super fine sandpaper
(FEPA P800, grain size 21.8 μm). The specified dimensions were achieved within 5% tolerance.

Figure 1. Matrix polymer and composite plates used for swelling measurements.

Distilled water (0.5–1.0 MΩ·cm) was used for conditioning matrix polymer and composite
samples, produced via water purification system Aquatron A4000 (Cole-Parmer, Vernon Hills, IL, USA).

2.2. Methods
Fiber volume fractions were obtained by density measurements. The polymer and composite
plates were exposed to water. The resulting changes in plate dimensions were measured using digital
J. Compos. Sci. 2019, 3, 10 4 of 14

microscopy, allowing swelling-induced linear strains to be obtained at various water contents for
matrix polymers (ε m ) and for composites in 3 different directions (ε x , ε y , ε z ).
The density of matrix polymer (ρm ) and glass fiber (ρ f ) was 1.1 g/cm3 and 2.54 g/cm3 , respectively.
The density of the composite (ρcomposite ) was determined to be 1.97 g/cm3 by measuring mass and
dimensions of a large composite block. The volume and mass fractions of matrix polymer were
calculated using Equations (2) and (3), respectively.

ρcomposite − ρm
Vf = (2)
ρ f − ρm

ρ f · Vf
mf =   (3)
ρ m · 1 − Vf + ρ f · Vf

The volume and mass fraction of the fibers were Vf = 0.606 and m f = 0.780, respectively. The void
content is ignored, because it was very low (less than 0.02 %—calculated from true water content [12])
and could be neglected. The true water content of a composite should be calculated according to [12].
If the void content is high, it should be considered.
Water uptake experiments were conducted using a batch system. A heated distilled water
(60 ± 1 ◦ C) bath was used for conditioning the polymer and composite plates. Samples were
conditioned for a period of about two months up to reaching equilibrium. During this time, samples
were taken out of the water bath at various times and weighed using analytical scales AG204 (± 0.1 mg;
Mettler Toledo, Columbus, OH, USA). The moisture content (true water content) was calculated from
experimental gravimetric data using a method described in more detail in another work that allows to
obtain the mass of an absolutely dry plate [12]. The definition of moisture content used in this work is
given in Equation (4):
mwater m plate − mdry plate
W= ·100% = ·100% (4)
mdry plate mdry plate
where m plate is the mass of epoxy or composite plate sample; mdry plate is the mass of an absolutely dry
epoxy or composite plate sample; mwater is the mass of moisture uptaken by the plate.
The moisture content in composite plates was scaled by the mass matrix fraction as shown in
Equation (5), since fibers do not take up moisture:

W
Wm = (5)
1 − mf

This operation is useful for easier comparison of swelling-induced strains between the matrix
polymer and the composite laminates, since it is only the polymer part which is affected by swelling
in composites.
Optical microscopy was performed using a digital microscope RH-2000 (Hirox, Tokyo, Japan)
equipped with lens MXB-2500REZ with a magnification of 140 and resolution of 1.06 μm. Changes in
the length and width of the plates were measured using the edge dimensions of the plates. The strains
and swelling coefficients were obtained using Equations (6) and (7) for matrix polymer and composites,
respectively:
l − l0
εm = = βmW (6)
l0
l − l0
εi = = β i Wm (7)
l0
where i stands for x, y and z for respective swelling directions. Composite C1 was used to obtain
swelling coefficients in the direction parallel to fibers (β x ) and transverse to fibers (β y ), while composite
C3 also provided transverse-to-fibers swelling (β z ) which was similar to and consistent with β y .
J. Compos. Sci. 2019, 3, 10 5 of 14

3. Model

3.1. Swelling Analytical Model

The swelling behaviour of a fiber-reinforced polymer composite can be estimated using a simple
scheme where fiber and matrix are schematized as rectangular, in a cubic cell having unit dimensions
1 × 1 × 1, shown in Figure 2.

Figure 2. Schematic fiber-reinforced composite (cubic unit cell with dimensions 1 × 1 × 1).

The RVE represented in Figure 2 is the result of periodicity and symmetry considerations.
In fact, the RVE can be imagined as half model of a RVE comprised by matrix surrounded by fibers.
Consequently, localized differential deformations and strains may arise in the RVE, but the global
deformed shape does not bend, due to the symmetry along its lateral plane.
Composite transverse swelling, which occurs in direction y in Figure 2, can be predicted as a
serial connection of fiber and matrix, hence:
 
ε y = Vf ε f + 1 − Vf ε m (8)

where ε y is the composite transverse swelling strain, Vf the fiber volume fraction, ε f is the fiber swelling
strain and ε m is the matrix swelling strain. For many engineering reinforcements (carbon fibers, glass
fibers, etc.) the swelling is null, ε f = 0. In these cases, Equation (8) can be simplified as follows:
 
ε y = 1 − Vf ε m (9)

The transverse swelling coefficient is defined in Equation (10) [35]:

 
εy 1 − Vf ε m  W
βy = = = 1 − Vf βm (10)
Wc Wc Wc

where W is the moisture content in the matrix and Wc the moisture content in the composite.
Composite axial swelling, which occurs in direction x in Figure 2, can be predicted employing
a parallel connection model of fiber and matrix. The swelling of the matrix in this case is strongly
constrained by the stiffness of the fibers. The constrained swelling strain in the matrix ε m generates a
stress equal to:
σm = Em ε m (11)

where σm is the stress in the matrix and Em is the stiffness of the matrix.
The load transferred from the matrix to the fibers, L can be estimated as:

L = σm Am = σm A(1 − Vf ) = σ f A f = σ f AVf (12)

J. Compos. Sci. 2019, 3, 10 6 of 14

where A is the cubic cell lateral surface area (in the x − z plane), Am = A(1 − Vf ) is the matrix part of
the cubic cell lateral surface area and A f = AVf is the fiber part of the cubic cell lateral surface area,
as shown in Figure 2.
From Equation (12) it is possible to estimate the stress transferred to the fiber, σ f :

1 − Vf
σ f = σm (13)
Vf

Finally, the composite axial strain, ε x , can be predicted as equal to the fiber axial strain, ε f , as the
axial swelling of the cubic cell is governed by its behaviour:

σf σm 1 − Vf
εx = ε f = = (14)
Ef E f Vf

where E f is the fiber stiffness.

The transverse swelling coefficient is defined as [35]:

εx σm 1 − Vf Em ε m 1 − Vf Em β m W 1 − Vf
βx = = = = (15)
Wc Wc E f Vf Wc E f Vf E f Wc Vf

3.2. Finite Element Model

The swelling behaviour of a fiber-reinforced polymer composite can be estimated using a
three-dimensional periodic RVE comprised of matrix and randomly placed fibers.
An RVE was modelled in AbaqusTM , (Abaqus Inc., Johnston, RI, USA) having dimensions 110 μm
× 110 μm × 10 μm. The fibers radius is 9 μm, as measured experimentally. These RVE dimensions
enable having several fibers in the RVE, predicting the composite properties with a good fidelity [36,37].
The thickness of the RVE enables axial swelling to be captured accurately while reducing the total
number of elements required for the analysis. A sensitivity analysis was performed employing a
thicker RVE (110 μm × 110 μm × 55 μm), the results deviation was below 1%. The elements used are
C3D6 for the fibers (6-node linear triangular prism) and C3D8R for the matrix (8-node linear brick,
reduced integration). The element size chosen was 1.5 μm. A mesh sensitivity analysis was also
performed in order to verify the convergence of the results.
The mesh of the model is shown in Figure 3. The constituents of the composite are modelled
as linear elastic. The elastic constants of the matrix polymer are Em = 2.908 GPa and νm = 0.35 [9].
The elastic constants of the fibers are E f = 72.4 GPa and ν f = 0.22 [35].

Figure 3. Finite element model of the representative volume element (RVE): fiber and matrix properties
are also shown.
J. Compos. Sci. 2019, 3, 10 7 of 14

In order to account for the deviations in the swelling coefficients due to the random fiber
distribution in the RVE, five RVEs having the same fiber volume fraction were modelled, as shown
in Figure 4. This allows the deviation in swelling coefficient due to the fiber random arrangement to
be quantified.

Figure 4. Finite element model of the RVE.

Hygrothermal swelling was simulated in AbaqusTM using a heat-mass transfer laws analogy,
defining a coefficient of hygrothermal expansion β instead of the thermal expansion coefficient α and a
concentration field W ( x, y, z) instead of the temperature field T ( x, y, z).

ε = αΔT ↔ ε = βΔW (16)

During the simulation, periodic boundary conditions were applied on the surfaces of the RVE.
→ → →
These are applied as relative displacements of opposite faces of the RVE, vectors u x , uy and uz .
→ → →
u ( L x , y, z) − u (0, y, z) = u x
→ → →
u x, Ly , z − u ( x, 0, z) = uy (17)
→ → →
u ( x, y, Lz ) − u ( x, y, 0) = uz

→
where x, y and z are Cartesian coordinates, u ( x, y, z) the displacement vector at a point having
coordinates (x, y, z) and L x , Ly and Lz the dimensions of the RVE in directions x, y and z, respectively.
Displacements on one node of the RVE were constrained in order to avoid rigid translations.
Fluid saturation was modelled using a predefined field of initially dry and later saturated material.
The homogenized strain components of the RVE were obtained as:

1
ε ij = ε ij dV (18)
V V

where ε ij is the homogenized strain component of the RVE and ε ij and V are the strain component and
the volume of each point in the RVE, respectively.
The swelling coefficients are defined as [35]:

ε ij
β ij = (19)
ΔW
where W is the moisture content.
J. Compos. Sci. 2019, 3, 10 8 of 14

4. Results

4.1. Swelling Measurements

Hygroscopic strains were measured after various exposure times to water for both matrix polymer
and composite samples (C1 and C3). The experimentally obtained evolution of linear strains with
increasing water contents for the polymer and composites, is shown in Figures 5 and 6, respectively.

Figure 5. Experimental data and linear regression of hygroscopic swelling of epoxy polymer.

Figure 6. Experimental data and linear regression of orthotropic swelling for composites.

Linear regression of experimental data was performed via the least squares approach to obtain a
coefficient of hygroscopic expansion of the matrix polymer β m and for composite plates in directions
transverse (β y , β z ) and parallel to fibers (β x ). The determination coefficient R2 obtained for the matrix
J. Compos. Sci. 2019, 3, 10 9 of 14

polymer via linear regression of experimental data was 0.9534 with an average β m of 0.332. The linear
regression included 112 observations with a standard deviation of 0.021 for the β m . The average slope
for composite experimental data in the transverse-to-fibers directions was found to be 0.571 with a
determination coefficient of R2 = 0.9750. A similar value of 0.6 was reported by Tsai for Kevlar/epoxy
composite [15]. The linear regression included 129 observations with a standard deviation of 0.063 for
the slope. The average slope β x for experimental data obtained in the parallel-to-fibers direction was
found to be non-zero and equal to 0.060 with a determination coefficient of R2 = 0.3110. The linear
regression included 77 observations with a standard deviation of 0.018 for the slope. The experimental
scatter is reported with one standard deviation in Table 1.

4.2. Analytical and Finite Element Predictions

Experiments validate both analytical and numerical models. Both numerical and analytical
results are shown in Figure 7. They show a good fit with the experimental swelling data of
composites. The analytically predicted coefficient 0.546 (both β y and β z ) of hygroscopic swelling
in transverse-to-fibers directions (y and z) is close to the numerically predicted values of 0.552 and
0.573 for β y and β z , respectively, and all predicted values fall within the experimental scatter of 0.571 ±
0.063 (both β y and β z ).
In the fiber direction, both analytically and numerically obtained coefficients of hygroscopic
swelling β x are close, 0.036 and 0.044, respectively. Both values are slightly higher than the experimental
value obtained via linear regression 0.060 ± 0.018. Such discrepancy is reasonable considering the
low determination coefficient in the linear regression of experimental data (0.3110) and low strains
involved resulting in larger experimental data scatter.
Numerical FE and analytical prediction as well as experimental results are shown in Table 1 and
Figure 7.

Figure 7. Numerical and analytical fit of swelling with the experimental data for composites. Analytical
results are shown in red lines; global averaged out numerical results are shown in blue lines, while a
numerical scatter is shown with blue sectors indicating scatter on the local scale due to various
fiber arrangements.
J. Compos. Sci. 2019, 3, 10 10 of 14

Table 1. Coefficients of hygroscopic expansion. Experimental scatter is reported with one standard
deviation for the coefficient of hygroscopic expansion (CHE). * indicates an input parameter for
the models.

CHE Experimental Analytical Numerical

βm 0.332 ± 0.021 0.332 * 0.332 *
βX 0.060 ± 0.018 0.036 0.044
βY 0.569 ± 0.066 0.546 0.552
βZ 0.576 ± 0.059 0.546 0.573

The numerical scatter was obtained using five RVEs of the same fiber volume fraction with
random fiber distribution to assess swelling on the local scale. On the global scale, these effects are
averaged out. The predicted numerical scatter 0.552 ± 0.036 and 0.573 ± 0.049 for numerically obtained
β y and β z , respectively.
It is possible to notice that axial swelling coefficients obtained numerically, β x , have quite low
scatters on a local scale due to various fiber arrangements, while transverse coefficients, β y and
β z , have higher scatters. This suggests that axial swelling is not strongly influenced by the fiber
arrangement, while transverse swelling is sensible to the fiber distribution.
Variability of numerically predicted CHE values due to random fiber orientation is shown in
Figures 8 and 9.

Figure 8. Swelling coefficients in transverse directions y and x for the five RVEs used in the simulations.

Figure 9. Swelling coefficient in axial direction, z, for the five RVEs used in the simulations.
J. Compos. Sci. 2019, 3, 10 11 of 14

Figure 10 shows the hygrothermal strains in RVE1. For the transverse directions, y and z,
the swelling strains arise mainly in the matrix and are negligible in the fibers, since these do not
swell. For the axial direction, however, it is possible to notice a very homogeneous strain field in
the whole RVE, matrix and fibers are displaced together. This observation shows the validity of the
assumption in the analytical section, where axial swelling is modelled as a parallel connection of fiber
and matrix.

Figure 10. Swelling coefficient in axial direction, z, for the five RVEs used in the simulations.

5. Discussion
From the experimentally obtained hygroscopic swelling data (Figure 5), it can be deduced that
the isotropic coefficient of hygroscopic swelling of the polymer matrix β m is 0.322. For composite
plates hygroscopic swelling occurred freely in the transverse-to-the-fibers direction. As expected,
obtained values of β y (0.569 ± 0.066) and β z (0.576 ± 0.059) for both composite plates C1 and C3 were
very similar. These values can be predicted from the matrix polymer (β m = 0.322), considering that
fibers are not affected by swelling and the volume fraction of the matrix in these composites was 0.394.
Hygroscopic strains determined in the direction of the fibers are considered to be consistent with the
hypothesis that hygroscopic strains in the fiber direction in composites are very low, but nevertheless
they were found to be non-zero.
Furthermore, analytical and numerical approaches show a good agreement. The analytical model
predicts the coefficient of hygroscopic swelling well within the experimental scatter. The experimental
scatter includes errors due to (1) weighing, (2) determination of dimensional change and (3) variations
between the samples. The experimental fit curve is a mean of all these considerations. Analytical
results are based on the mean values of Vf , β m , E f , Em . It fits the experimental mean curve fairly well,
and is clearly within the experimental scatter.
The analytical model, however, does not consider the effect of fiber distribution for a known fiber
fraction. The effect of fiber distribution on the coefficient of hygroscopic swelling can be accessed via
FE simulations and the numerical scatter can be obtained. However, when a simplified and faster
approach is needed, the analytical model is able provide a good prediction of hygroscopic swelling of
composites using the properties of the polymer (β m and Em ) and fibers (E f ). The observation that the
simple analytical model gives practically the same results on swelling strains as experiments and more
sophisticated FE analysis shows as a first approximation that the stiff fibers do not restrain expansion
of the matrix transverse to the fibers. Furthermore, the effect of fiber distribution is important mostly
on the local scale and would average out on the global scale. Such global results along with local
numerical scatters are shown and compared with the experimental and analytical results in Figure 7.
FE results were very similar to experimental and analytical results.
To the best understanding of the authors, there is no limitation to apply the model at various
humidity levels. This is due to the fact that swelling behaviour is predicted from the microconstituent
properties. The water uptake behaviour of a polymer matrix may be affected by different humidity
levels. This would affect the concentration of water in the polymer, but to the best knowledge of the
J. Compos. Sci. 2019, 3, 10 12 of 14

authors this does not change the method of predicting the swelling of the composite from the swelling
of the polymer.
The imperfections of the fiber–matrix interface are neglected in this model. While this is an
assumption, the good agreement between experimental, numerical and analytical results indicates that
this effect is negligible for the material system studied. For a more advanced model, there should be
additional studies concerning this aspect.
In addition, the authors believe that it would be of interest to develop analytical models based on
the same approach shown here not only for the fiber-reinforced composites, but also for other novel
material systems such as polymer/clay composites [38] and multilayer nanocomposites of halloysite
and biopolymers [39], that also possess a high mismatch in elastic properties of constituent materials.

6. Conclusions
Orthotropic water-induced or hygroscopic swelling of fiber-reinforced composite laminates can
be predicted from the isotropic swelling data of the polymer (the matrix material of a composite).
Swelling in composites shows a strong anisotropy, where swelling in the fiber direction is close to null,
being constricted by non-expanding fibers, while swelling transverse to fibers behaves similarly to that
of the unconstrained polymer. Good agreement was obtained and is reported between experimental
swelling data, and analytical and numerical results for composite laminates, indicating the validity
of this predictive approach. The analytical approach for predictions is the simplest to use, and good
agreement with finite element analysis validates the use of this analytical tool for applications.

Author Contributions: Conceptualization, A.E.K., A.I.G. and A.T.E.; methodology, A.E.K. and A.I.G.; formal
analysis, A.E.K. and A.I.G.; investigation, A.E.K.; resources, A.E.K. and A.T.E.; data curation, A.E.K., A.I.G. and
A.T.E.; writing—original draft preparation, A.E.K. and A.I.G.; writing—review and editing, A.E.K., A.I.G. and
A.T.E.; validation, A.E.K. and A.I.G.; visualization, A.E.K. and A.I.G.; supervision, A.T.E.; project administration,
A.T.E.; funding acquisition, A.T.E.
Funding: This research was funded by the Research Council of Norway (Project 245606/E30 in the Petromaks
2 programme).
Acknowledgments: This work is part of the DNV GL led Joint Industry Project “Affordable Composites” with
19 industrial partners and the Norwegian University of Science and Technology (NTNU). The authors would like
to express their thanks for the financial support from The Research Council of Norway (Project 245606/E30 in the
Petromaks 2 programme). Andrey is especially grateful to Oksana V. Golubova.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Krauklis, A.; Echtermeyer, A. Mechanism of yellowing: Carbonyl formation during hygrothermal aging in a
common amine epoxy. Polymers 2018, 10, 1017. [CrossRef]
2. Xiao, G.Z.; Shanahan, M.E.R. Swelling of DGEBA/DDA epoxy resin during hygrothermal ageing. Polymer
1998, 39, 3253–3260. [CrossRef]
3. Toscano, A.; Pitarresi, G.; Scafidi, M.; Di Filippo, M.; Spadaro, G.; Alessi, S. Water diffusion and swelling
stresses in highly crosslinked epoxy matrices. Polym. Degrad. STable 2016, 133, 255–263. [CrossRef]
4. Grabovac, I.; Whittaker, D. Application of bonded composites in the repair of ships structures—A 15-year
service experience. Compos. Part A 2009, 40, 1381–1398. [CrossRef]
5. McGeorge, D.; Echtermeyer, A.T.; Leong, K.H.; Melve, B.; Robinson, M.; Fischer, K.P. Repair of floating
offshore units using bonded fibre composite materials. Compos. Part A 2009, 40, 1364–1380. [CrossRef]
6. Gustafson, C.G.; Echtermeyer, A. Long-term properties of carbon fibre composite tethers. Int. J. Fatigue. 2006,
28, 1353–1362. [CrossRef]
7. Salama, M.M.; Stjern, G.; Storhaug, T.; Spencer, B.; Echtermeyer, A. The First Offshore Field Installation for a
Composite Riser Joint. OTC-14018-MS. In Proceedings of the Offshore Technology Conference, Houston, TX,
USA, 6–9 May 2002; Offshore Technology Conference: Houston, TX, USA, 2002. [CrossRef]
J. Compos. Sci. 2019, 3, 10 13 of 14

8. Echtermeyer, A.T.; Gagani, A.I.; Krauklis, A.E.; Mazan, T. Multiscale Modelling of Environmental
Degradation—First Steps. In Durability of Composites in a Marine Environment 2. Solid Mechanics and Its
Applications; Davies, P., Rajapakse, Y.D.S., Eds.; Springer: Cham, Switzerland, 2018; Volume 245, pp. 135–149,
ISBN 978-3-319-65145-3.
9. Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Hygrothermal aging of amine epoxy: Reversible static and
fatigue properties. Open Eng. 2018, 8, 447–454. [CrossRef]
10. Ibarra, L.; Chamorro, C. Short fiber–elastomer composites. Effects of matrix and fiber level on swelling and
mechanical and dynamic properties. J. Appl. Polym. Sci. 1991, 43, 1805–1819. [CrossRef]
11. Zhou, J. Transient analysis on hygroscopic swelling characterization using sequentially coupled moisture
diffusion and hygroscopic stress modeling method. Microelectron. Reliab. 2008, 48, 805–810. [CrossRef]
12. Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Near-Infrared spectroscopic method for monitoring water
content in epoxy resins and fiber-reinforced composites. Materials 2018, 11, 586. [CrossRef]
13. Hahn, H.T.; Kim, R.Y. Swelling of Composite Laminates. In Advanced Composite Materials: Environmental
Effects; ASTM STP 658; American Society for Testing and Materials: Philadelphia, PA, USA, 1978; pp. 98–120.
14. Shirangi, M.H.; Michel, B. Mechanism of moisture diffusion, hygroscopic swelling, and adhesion degradation
in epoxy molding compounds. In Moisture Sensitivity of Plastic Packages of IC Devices; Fan, X.J., Suhir, E., Eds.;
Springer: New York, NY, USA, 2010; pp. 29–69, ISBN 978-1-4419-5719-1.
15. Tsai, S.W. Composites Design, 4th ed.; Think Composites: Dayton, OH, USA, 1988; ISBN 0-9618090-2-7.
16. Obeid, H.; Clément, A.; Fréour, S.; Jacquemin, F.; Casari, P. On the identification of the coefficient of moisture
expansion of polyamide-6: Accounting differential swelling strains and plasticization. Mech. Mater. 2018,
118, 1–10. [CrossRef]
17. Cairns, D.S.; Adams, D.F. Moisture and thermal expansion properties of unidirectional composite materials
and the epoxy matrix. J. Reinf. Plast. Compos. 1983, 2, 239–255. [CrossRef]
18. Ashton, J.E.; Halpin, J.C.; Petit, P.H. Primer on Composite Materials Analysis; Technomic Pub. Co.: Stamford,
CT, USA, 1969; ISBN 978-0-8776-2754-8.
19. Halpin, J.C. Effects of Environmental Factors on Composite Materials; Technical report AFML-TR-67–423;
Air Force Materials Laboratory: Dayton, OH, USA, 1969.
20. Schapery, R.A. Thermal expansion coefficients of composite materials based on energy principles. J. Compos.
Mater. 1968, 2, 380–404. [CrossRef]
21. Coran, A.Y.; Boustany, K.; Hamed, P. Unidirectional fiber-polymer composites: Swelling and modulus
anisotropy. J. Appl. Polym. Sci. 1971, 15, 2471–2485. [CrossRef]
22. Daniels, B.K. Orthotropic swelling and simplified elasticity laws with special reference to cord-reinforced
rubber. J. Appl. Polym. Sci. 1973, 17, 2847–2853. [CrossRef]
23. Fan, Y.; Gomez, A.; Ferraro, S.; Pinto, B.; Muliana, A.; La Saponara, V. Diffusion of water in glass fiber
reinforced polymer composites at different temperatures. J. Compos. Mater. 2018, 1–14. [CrossRef]
24. Jacquemin, F.; Fréour, S.; Guillén, R. Analytical modeling of transient hygro-elastic stress
concentration—Application to embedded optical fiber in a non-uniform transient strain field. Compos.
Sci. Technol. 2006, 66, 397–406. [CrossRef]
25. Meng, M.; Rizvi, M.J.; Le, H.R.; Grove, S.M. Multi-scale modelling of moisture diffusion coupled with stress
distribution in CFRP laminated composites. Compos. Struct. 2016, 138, 295–304. [CrossRef]
26. Sinchuk, Y.; Pannier, Y.; Gueguen, M.; Tandiang, D.; Gigliotti, M. Computed-tomography based modeling
and simulation of moisture diffusion and induced swelling in textile composite materials. Int. J. Solids Struct.
2018, 154, 88–96. [CrossRef]
27. Sinchuk, Y.; Pannier, Y.; Gueguen, M.; Gigliotti, M. Image-based modeling of moisture-induced swelling nd
stress in 2D textile composite materials using a global-local approach. Proc. Inst. Mech. Eng. Part C J. Mech.
Eng. Sci. 2018, 232, 1505–1519. [CrossRef]
28. Weitsman, Y.J. Fluid Effects in Polymers and Polymeric Composites; Springer: New York, NY, USA, 2012;
ISBN 978-1-4614-1059-1.
29. Weitsman, Y.J.; Elahi, M. Effects of fluids on the deformation, strength and durability of polymeric
composites—An overview. Mech. Time-Depend. Mater. 2000, 4, 107–126. [CrossRef]
30. Sar, B.E.; Fréour, S.; Davies, P.; Jacquemin, F. Accounting for differential swelling in the multi-physics
modelling of the diffusive behaviour of polymers. J. Appl. Math. Mech. 2014, 94, 452–460. [CrossRef]
J. Compos. Sci. 2019, 3, 10 14 of 14

31. Fan, Y.; Gomez, A.; Ferraro, S.; Pinto, B.; Muliana, A.; La Saponara, V. The effects of temperatures and
volumetric expansion on the diffusion of fluids through solid polymers. J. Appl. Polym. Sci. 2017, 134,
45151–45165. [CrossRef]
32. Kappert, E.J.; Raaijmakers, M.J.T.; Tempelman, K.; Cuperus, F.P.; Ogieglo, W.; Benes, N.E. Swelling of 9
polymers commonly employed for solvent-resistant nanofiltration membranes: A comprehensive dataset.
J. Membr. Sci. 2019, 569, 177–199. [CrossRef]
33. Ramirez, F.A.; Carlsson, L.A.; Acha, B.A. Evaluation of water degradation of vinylester and epoxy matrix
composites by single fiber and composite tests. J. Mater. Sci. 2008, 43, 5230–5242. [CrossRef]
34. Shirell, C.D.; Halpin, J. Moisture Absorption in Epoxy Composite Materials. In Proceedings of the Composite
Materials: Testing and Design (Fourth Conference), ASTM STP617, Valley Forge, PA, USA, 3–4 May 1977;
American Society for Testing and Materials: Philadelphia, PA, USA, 1979; pp. 514–528. [CrossRef]
35. Agarwal, B.D.; Broutman, L.J. Analysis and Performance of Fiber Composites, 2nd ed.; John Wiley & Sons, Inc.:
Hoboken, NJ, USA, 1990; ISBN 978-0-4715-1152-6.
36. González, C.; Llorca, J. Mechanical behavior of unidirectional fiber-reinforced polymers under transverse
compression: Microscopic mechanisms and modeling. Compos. Sci. Technol. 2007, 67, 2795–2806. [CrossRef]
37. Yang, L.; Yan, Y.; Ma, J.; Liu, B. Effects of inter-fiber spacing and thermal residual stress on transverse failure
of fiber-reinforced polymer–matrix composites. Comput. Mater. Sci. 2013, 68, 255–262. [CrossRef]
38. Yamabe, K.; Goto, H. Synthesis and surface observation of montmorillonite/polyaniline composites. J. Comp.
Sci. 2018, 2, 15. [CrossRef]
39. Bertolino, V.; Cavallaro, G.; Milioto, S.; Parisi, F.; Lazzara, G. Thermal properties of multilayer
nanocomposites based on halloysite nanotubes and biopolymers. J. Comp. Sci. 2018, 2, 41. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
APPENDIX E

PAPER V

KRAUKLIS A.E., ECHTERMEYER A.T.

LONG-TERM DISSOLUTION OF GLASS FIBERS IN WATER DESCRIBED BY DISSOLVING

CYLINDER ZERO-ORDER KINETIC MODEL: MASS LOSS AND RADIUS REDUCTION.

OPEN CHEMISTRY (POLAND), 16(1), 2018, 1189-1199.

DOI:10.1515/CHEM-2018-0133

PAPER V

127
128
 Open Chem., 2018; 16: 1189–1199

Research Article Open Access

Andrey E. Krauklis* Andreas T. Echtermeyer

Long-Term Dissolution of Glass Fibers in Water

Described by Dissolving Cylinder Zero-Order Kinetic
Model: Mass Loss and Radius Reduction
https://doi.org/10.1515/chem-2018-0133
received August 16, 2018; accepted October 1, 2018.
1 Introduction
Abstract: Glass fibers are degraded when they are Glass fibers are widely used as reinforcement in structural
exposed to water. In this work, a model is developed that composite materials. They are produced from raw
uses zero-order kinetics for predicting a decreasing glass materials, which are virtually unlimited in supply [1]. The
fiber radius. The model is used to describe experimental fibers have high hardness and such desirable reinforcement
test results of almost half a year long-term dissolution properties as high strength and stiffness in fiber direction
of R-glass fibers. The model is able to predict both mass [1,2]. It is known that, glass fibers are hydrophilic and are
loss and radius reduction kinetics using the same four susceptible to a relatively slow degradation when exposed
parameters: initial fiber radius (r0), rate constants for to water environments [3]. Hydrolytic degradation of glass
both short-term degradation (‫ܭ‬଴ூ ) and steady-state fibers can significantly reduce mechanical strength and
degradation (‫ܭ‬଴ூூ ) and the time when steady-state leads to corrosion-induced defects [4].
kinetics are reached (tst). All parameters can be easily The term degradation is used in the present work to
determined from initial radius measurements and mass denote all processes which lead to or affect the mass loss of
loss evolution in time. Elements released and detected the glass material by interacting with water. This includes
during degradation were Na, K, Ca, Mg, Fe, Al, Si and Cl. the complex nature of several processes occurring in
Rate constants were obtained for individual ion release parallel, such as dissolution of glass matrix constituents,
and for the total mass loss. The contribution of Si to the gel layer formation, neoformation of solid reaction
total mass loss was the largest (56.1% by mass). It governed products, alkaline and alkaline earth ion exchange. Some
the dissolution process. The kinetics of radius reduction of these reactions occur while maintaining the glassy state,
are also reported. The radius reduction was found to be while others lead to leaching of the reaction products into
linear with time during the steady-state dissolution. The the water environment [5].
zero-order kinetic constant and the density of the glass The long-term environmental degradation of glass
describe the rate (proportionality) of the dissolution. materials, described by chemical kinetics, has been
extensively studied with respect to nuclear waste
Keywords: glass fibers; model; dissolution; kinetics; ion applications [5,6]. More recently, there has been an
release. increasing interest in environmental degradation of
composite laminates, where R-glass is often used as
reinforcement [7-10]. However, studies on environmental
degradation of composites are usually concerned with
deterioration of the mechanical properties, and the
kinetics of chemical degradation tend to be overlooked
[8,10]. A number of studies exist that explain general
mechanisms of environmental degradation of glass
*Corresponding author: Andrey E. Krauklis, Department of Mechanical materials using various approaches that are based on
and Industrial Engineering, Norwegian University of Science and surface reactions, chemical affinity and diffusion [3,11-
Technology, 7491 Trondheim, Norway, E-mail: [email protected] 14]. A process-driven approach was described by Geisler-
Andreas T. Echtermeyer: Department of Mechanical and Industrial Wierwille et al., involving the congruent dissolution of
Engineering, Norwegian University of Science and Technology, 7491
the glass coupled to the precipitation and growth of an
Trondheim, Norway

Open Access. © 2018 Andrey E. Krauklis, Andreas T. Echtermeyer, published by De Gruyter. This work is licensed under the
Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. Authenticated | [email protected]
Download Date | 11/23/18 12:05 PM
1190 Andrey E. Krauklis, Andreas T. Echtermeyer

amorphous silica layer at an inwardly moving reaction or contracting cylinder models [5,16,18-20]. Much of the
interface [12]. More recently, a combination of chemical confusion related to the modelling of degradation of
affinity for controlling the distribution of Si among glass fibers is due to the fact that dissolution is not the
different alteration phases and the diffusion barrier for only process involved in degradation. The discrepancies
the release rate of glass modifiers was implemented by Ma between models are, however, mostly dominant in the
et al. [14]. Dissolution experiments in existing studies are short-term [5]. The whole degradation versus time curve
mainly performed with bulk silicate glasses, and fibers are is not linear as long as the complex processes take place.
not studied often. The model reported in this work is similar in concept to an
Few studies exist on the kinetics of glass fiber existing solid-state model called the contracting cylinder
dissolution. Mišíková et al. have studied the E-glass [20] or shrinking cylinder [16], which relates the evolution
fiber leaching kinetics in distilled water at different of conversion to time. However, the novel model presented
temperatures [15]. Bashir et al. studied the kinetics of in this work differentiates between the complex short-
the dissolution of E-glass fibres in alkaline solutions by term and dissolution-dominated long-term processes.
immersing single fibers and measuring the diameter Furthermore, the novel model describes both dissolution
change [16]. They concluded that the rate limiting step and radius reduction kinetics without the necessity for
was either the diffusion of hydroxide ions through the introducing additional terms such as a conversion factor.
solution or the glass fibre etching itself [16]. Krauklis Rate constants are often given in relation to the surface
and Echtermeyer presented a model that predicts Si loss area of a material, thus describing dissolution behaviour
kinetics during long-term hygrothermal aging of fiber- from a unit of the material’s surface area [14]. Surface
reinforced composites [17]. Glass fibers formulated with area-based constants are used throughout this work.
significant amounts of various metal oxides, such as
R-glass (often used in composites), are rarely studied [16].
Thus, this work is concerned with the dissolution kinetics
of R-glass fibers in water.
3 Experimental
The degradation is divided into two distinct kinetic
regions: (1) short-term non-steady-state and (2) long-term 3.1 Materials
steady-state degradation. In the short-term region, the
degradation is complex and involves such processes as Boron-free and fluorine-free HiPer-tex W2020 glass
ion exchange, gel formation and dissolution [5]. When the fiber (GF) bundles were used. These are classified as
long-term steady-state is reached, the dissolution becomes high strength, high modulus R-glass (defined by an
dominant [5,18] and the degradation follows zero-order international standard [21]). The density is 2.54 g/cm3; the
reaction kinetics. Such kinetics depend on the glass mean filament diameter is 17 ± 2 μm [22].
surface area in contact with water, which is proportional Deionized water (0.5-1.0 MΩ·cm) was used for
to the fiber radius. As the dissolution continues, the radius conditioning of the glass fibers. It was produced by
decreases resulting in the mass loss deceleration. using the water purification system Aquatron A4000
The aim of this work is twofold: (1) to study the long- (Cole-Parmer, USA). The pH of the deionized water was
term dissolution behaviour of R-glass fibers in water and determined to be 5.65, which is lower than neutral due to
(2) to develop a model that describes the mass loss and dissolved CO2 (in equilibrium).
fiber radius reduction kinetics during dissolution.

3.2 Determination of number of fibers in the

2 Theory – dissolution kinetic sample
models The amount of fibers is denoted as n. The length of fibers is
l = 0.1 m; the initial mass of a fiber bundle is m0=5.8 g; the
Different models are used for describing dissolution mean initial radius is r0=8.5∙ 10-6 m. The amount of fibers
kinetics of various materials such as zero-order, first-order, was calculated using Equation 1:
advection/dispersion/reaction, GRAAL (Glass Reactivity
݉଴
with Allowance for the Alteration Levels), Noyes- ݊ൌ  (1)
ߩ௚௟௔௦௦ ߨ݈‫ݎ‬଴ଶ
Whitney, Hixson-Crowell, Hopfenberg and Korsmeyer-
Peppas models, as well as similar in concept shrinking

Authenticated | [email protected]
Download Date | 11/23/18 12:05 PM
 Long-Term Dissolution of Glass Fibers in Water Described by Dissolving Cylinder Zero-Order Kinetic... 1191

The amount of fibers in the sample n was determined to field ICP-MS Finnigan ELEMENT 2 (Thermo-Scientific),
be 100602. equipped with a sample introduction system PrepFAST
(ESI/Elemental Scientific) and a pretreatment/digestion
UltraClave (Milestone). Acidification of samples
3.3 Determination of surface area of the was performed using ultra-pure grade HNO3 SubPur
sample (Milestone) to avoid adsorption of ions to the wall of the
sample vials.
Since a surface reaction is of interest, it is important to The benefit of HR-ICP-MS versus gravimetric analysis
identify the specific surface area of the glass material. is that the model can be applied to obtain the rate constant
The external surface area of glass fibers was evaluated (K0) of each ion separately, as well as for the total mass
from geometrical considerations as a product of number, loss (mdissolved). The data obtained from the HR-ICP-MS
circumference and length of fibers. The specific surface experiments are in the form of mass concentration at
area of a studied sample can be calculated using each time point (non-cumulative) c (g/L) and need to be
Equation 2: converted to the mdissolved form by using Equation 3:

ܵ଴ ʹߨ݈݊‫ݎ‬଴ ௧
ܵ଴௦௣௘௖௜௙௜௖ ൌ ൌ  (2) ݉ௗ௜௦௦௢௟௩௘ௗ ൌ ܸ௪௔௧௘௥ න ܿ ݀‫ݐ‬ (3)
݉଴ ݉଴ ଴

Specific surface area ܵ଴௦௣௘௖௜௙௜௖ was determined to be 0.09 where Vwater is the volume of a water sample in the HR-ICP-
(m2/g). MS measurement. Vwater used for experiments was 50 mL.
Equation 3 is valid for each individual ion release and for
the total mass loss.
3.4 Dissolution experiments

Dissolution in water experiments of glass fibers were 3.5 Radius reduction measurements.
conducted using a batch system. Basically, infinite
water availablity conditions were ensured by using R-glass fibers were immersed in deionized water for
fresh deionized water for each separate measurement. prolonged periods of time, carefully dried with a paper
The ion release with time (dissolution) was obtained tissue, and the fiber diameter was measured using a
as the cumulative result of the individual ion release digital microscope RH-2000 (Hirox, Japan). The radii of
measurements. around five fibers were measured at each time, and for
The dissolution of ions was accelerated by each fiber, measurements of radii were made at three
conditioning the fibers at an elevated temperature of different locations, similarly as was done in another work
60 °C, since the dissolution reaction is an exothermic [16].
reaction [5,6,18]. The temperature of 60°C is often used Ethical approval: The conducted research is not
for conditioning fiber-reinforced composites, and this related to either human or animal use.
fact provided an additional justification for choosing
this particular temperature. The water´s temperature was
controlled via PID-controlled heating, giving an accuracy
of ± 1°C. Two-stage heating system was used in order to
4 Results and Discussion
ensure that there is no contact of the sample water with
other potential ion release sources, such as the heating 4.1 Elements released and respective
element itself. chemical reactions
Samples were weighed using analytical scales AG204
(± 0.1 mg; Mettler Toledo, USA) before the experiments. The HR-ICP-MS method identified the following elements
Experiments were performed with 3 parallels. as being released during glass fiber degradation in
The concentration of the dissolved ions in the water deionized water: Na, K, Ca, Mg, Fe, Al, Si, Cl. Release of B
bach was measured by high resolution inductively and Zn was found to be insignificant at less than 0.8 ppm,
coupled plasma mass spectrometry (HR-ICP-MS). Analyses if the elements were present at all.
were performed using a double focusing magnetic sector

Authenticated | [email protected]
Download Date | 11/23/18 12:05 PM
1192 Andrey E. Krauklis, Andreas T. Echtermeyer

Figure 1: Left: Schematic representation of a fiber bundle and geometrical dimensions. Right: micrograph of a fiber bundle.

Chemical reactions involved in the degradation of the The number of fibers is n (-); the initial radius of the fibers
studied glass fibers for the identified released elements is r0 (m); the length of fibers is l (m).
are shown in Chemical reactions (a)-(l) [5,18,23]: The model involves the following assumptions. As a
simplification, this model is deterministic and all fibers
ሺ‫ ݅ܵ ؠ‬െ ܱܰܽሻ ൅ ‫ܪ‬ଶ ܱ ՜ ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ܱ‫ ି ܪ‬൅ ܰܽା  ሺܽሻ are assumed to have the same initial radius, which is
r0; and the cross-sectional surface area at the end of the
ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܭ‬ሻ ൅ ‫ܪ‬ଶ ܱ ՜ ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ܱ‫ ିܪ‬൅ ‫ ܭ‬ା  ሺܾሻ
fibres is assumed to be negligible in calculations of the
ሺ‫ ݅ܵ ؠ‬െ ܱሻଶ ‫ ܽܥ‬൅ ‫ܪ‬ଶ ܱ ՜ ʹሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ʹܱ‫ ܪ‬൅ ‫ܽܥ‬ ି ଶା
 ሺܿሻ surface area. The length of the long fibers l is assumed to
ሺ‫ ݅ܵ ؠ‬െ ܱሻଶ ‫ ݃ܯ‬൅ ‫ܪ‬ଶ ܱ ՜ ʹሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ʹܱ‫ ܪ‬൅ ‫ ݃ܯ‬ ି ଶା
ሺ݀ሻ be constant during the whole dissolution process. During
the whole degradation process, the density of the glass
ሺ‫ ݅ܵ ؠ‬െ ܱ െ ‫ ݈ܣ‬ൌሻ ൅ ‫ܪ‬ଶ ܱ ՞ ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ሺൌ ‫ ݈ܣ‬െ ܱ‫ܪ‬ሻ ሺ݁ሻ
material stays constant (ρglass).
ሺ‫ ݅ܵ ؠ‬െ ܱሻଶ ‫ ݁ܨ‬൅ ‫ܪ‬ଶ ܱ ՜ ʹሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ʹܱ‫ ି ܪ‬൅ ‫ ݁ܨ‬ଶା  ሺ݂ሻ Dissolution is a surface reaction. In a general case,
the rate of the dissolution is dependent on the constant
ሺ‫ ݅ܵ ؠ‬െ ܱሻଷ ‫ ݁ܨ‬൅ ‫ܪ‬ଶ ܱ ՜ ͵ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ͵ܱ‫ ି ܪ‬൅ ‫ ݁ܨ‬ଷା  ሺ݃ሻ
describing the rate of the reaction (K0), the glass surface
ሺ‫ ݅ܵ ؠ‬െ ܱ െ ܵ݅ ‫ؠ‬ሻ ൅ ܱ‫ ି ܪ‬՞ ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ሺ‫ ݅ܵ ؠ‬െ ܱ ି ሻ ሺ݄ሻ area exposed to water (S), the availability of water (‫ܥ‬ுమ ை )
and the order of the reaction (nreaction). The global model
ሺ‫ ݅ܵ ؠ‬െ ܱ ି ሻ ൅ ‫ܪ‬ଶ ܱ ՞ ሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ൅ ܱ‫  ି ܪ‬ሺ݅ሻ
(general case) can then be mathematically expressed as
Equation 4.
Combining chemical reactions (h) and (i), a summary
reaction can be written as Chemical reaction (j): ߲݉ ௡
ൌ ‫ܭ‬଴ ܵ‫ܥ‬ுమೝ೐ೌ೎೟೔೚೙
ை  (4)
߲‫ݐ‬
ሺ‫ ݅ܵ ؠ‬െ ܱ െ ܵ݅ ‫ؠ‬ሻ ൅ ‫ܪ‬ଶ ܱ ՞ ʹሺ‫ ݅ܵ ؠ‬െ ܱ‫ܪ‬ሻ ሺ݆ሻ
In case of infinite availability of water, the rate of
ܱܵ݅ଶ ൅ ʹ‫ܪ‬ଶ ܱ ՜ ‫ܪ‬ସ ܱܵ݅ସ  ሺ݇ሻ reaction becomes independent of the reactant (water)
ுమை concentration, and the reaction order nreaction becomes 0. In
‫݈ܥ݁ܯ‬௫ ሱۛሮ ሺ‫ ݁ܯ‬௫ା ሻ ൅ ‫ ି ݈ܥݔ‬ ሺ݈ሻ
infinite water availability conditions, the surface reaction
can be well-described with zero-order kinetics [16,24],
which can then be represented by a differential Equation 5:
4.2 Zero-order kinetic model of a dissolving
cylinder
߲݉ (5)
ൌ ‫ܭ‬଴ ܵ
A schematic representation of a fiber bundle and ߲‫ݐ‬
important dimensions for the model are shown in Figure 1.

Authenticated | [email protected]
Download Date | 11/23/18 12:05 PM
 Long-Term Dissolution of Glass Fibers in Water Described by Dissolving Cylinder Zero-Order Kinetic... 1193

where m(g) is a total cumulative mass dissolved after time

ͳ ʹ‫ܭ‬଴
t (s), K0 (g/m2∙s) is a zero-order reaction kinetic constant ξ‫ ݖ‬ൌ ቆ ‫ ݐ‬൅ ܿ ᇱ ቇ (12)
ʹ ߩ௚௟௔௦௦
and S (m2) is the glass surface area in contact with water.
As the reaction proceeds, the radius of the fibers is
‫ܭ‬଴
reduced and the total surface area (S) is decreased, thus ξ‫ ݖ‬ൌ ‫ ݐ‬൅ ܿ ᇱᇱ  (13)
ߩ௚௟௔௦௦
leading to a decrease in the rate of mass loss as seen
in Equation 5. The overall ion release rate decreases
proportionally to the decrease in total surface area or a where c’ and c’’ are arbitrary constants after integration.
decrease in fiber radius. Substituting r2 back into the equation:
‫ܭ‬଴
ඥ‫ ݎ‬ଶ ൌ ‫ ݐ‬൅ ܿ ᇱᇱ  (14)
ܵሺ‫ݐ‬ሻ ʹߨ݈݊‫ ݎ‬ሺ‫ݐ‬ሻ ‫ݎ‬ሺ‫ݐ‬ሻ ߩ௚௟௔௦௦
ൌ ൌ  (6)
ܵ଴ ʹߨ݈݊‫ݎ‬଴ ‫ݎ‬଴
‫ܭ‬଴
േ‫ ݎ‬ൌ ‫ ݐ‬൅ ܿ ᇱᇱ  (15)
ௌሺ௧ሻ ௥ሺ௧ሻ ߩ௚௟௔௦௦
It can be seen that ௌ ൌ ௥ . Thus, the ion release rate
బ బ
decelerates linearly with a decrease in fiber radius. The
mass loss has to slow down as the available surface area It can be seen that the radius reduction is linear with
is reduced. Thus, the cumulative mass loss kinetic curve time, where the proportionality is given by the zero-order
deviates from a linear dissolution. kinetic constant K0 (g/m2∙s) and the density of the glass
The volume of a single fiber is πr2 l, where l is the (g/m3). A linear radius reduction of fibers with time was
cylinder length and r is the cylinder radius. For n fibers, previously experimentally observed in another work [16].
the volume is πr2 l and mass is ߩ௚௟௔௦௦ ݊ߨ‫ ݎ‬ଶ ݈ . The surface Since the radius reduction depends on the initial radius r0
area of a single fiber is 2nπrl. For n fibers it is 2nπrl. of the fibers (m), the arbitrary constant c’’ (m) is equal to
Substituting mass and surface area expressed in such the initial radius.
terms into Equation 5, The kinetic model equation for fiber radius reduction
then becomes as shown in Equation 16:

߲ߩ௚௟௔௦௦ ݊ߨ‫ ݎ‬ଶ ݈ ‫ܭ‬଴

ൌ ‫ܭ‬଴ ή ʹ݊ߨ‫݈ݎ‬ (7) ‫ ݎ‬ൌ ‫ݎ‬଴ െ ‫ݐ‬ (16)
߲‫ݐ‬ ߩ௚௟௔௦௦
Returning to the mass loss kinetics (Equation 5),
߲‫ ݎ‬ଶ ʹ‫ܭ‬଴
ൌ ‫ݎ‬ (8)
߲‫ݐ‬ ߩ௚௟௔௦௦
߲݉
ൌ ‫ܭ‬଴ ܵ ൌ ‫ܭ‬଴ ή ʹ݊ߨ‫݈ݎ‬ (5)
߲‫ݐ‬
Substituting then r2 with the following z = r2,
Substituting r for the radius reduction kinetic Equation 28,
the final mass loss kinetic model equation in differential
߲‫ݖ‬ ʹ‫ܭ‬଴
ൌ ξ‫ݖ‬ (9) form is obtained (Equation 17):
߲‫ߩ ݐ‬௚௟௔௦௦

ଶ௄బ ߲݉ ‫ܭ‬଴ଶ
where is a constant. Dividing both sides by ξ‫ݖ‬, ൌ ʹ݊ߨ݈ ቆ‫ݎ‬଴ ‫ܭ‬଴ െ ‫ݐ‬ቇ (17)
ఘ೒೗ೌೞೞ ߲‫ݐ‬ ߩ௚௟௔௦௦
߲‫ݖ‬
߲‫ ݐ‬ൌ ʹ‫ܭ‬଴  (10) Integrating the obtained Equation 17 over time t, the
ξ‫ߩ ݖ‬௚௟௔௦௦ integral model equation is obtained, describing the
cumulative mass loss (ion release):
Both sides are then integrated with respect to t,
௧
߲‫ݖ‬ ‫ܭ‬଴ଶ
ʹ‫ܭ‬଴ ݉ ௗ௜௦௦௢௟௩௘ௗ ൌ න ʹ݊ߨ݈ ቆ‫ݎ‬଴ ‫ܭ‬଴ െ ‫ݐ‬ቇ ݀‫ݐ‬ (18)
න ߲‫ ݐ݀ ݐ‬ൌ න ݀‫ݐ‬ (11) ଴ ߩ௚௟௔௦௦
ξ‫ݖ‬ ߩ௚௟௔௦௦

Integrating and solving for ξ‫ݖ‬,

Authenticated | [email protected]
Download Date | 11/23/18 12:05 PM
1194 Andrey E. Krauklis, Andreas T. Echtermeyer

Figure 2: Separation of mass loss or cumulative ion release curves into (I) short-term non-steady-state and (II) long-term steady-state regions.

Definite integral solution is then the following (Equation Equations 20 and 22 by the integration and by the mass
19): conservation principle, respectively, has proven the
mathematical consistency of the model. The fact that the
ͳ ‫ܭ‬଴ଶ model is adequate and physical was demonstrated by its
݉ௗ௜௦௦௢௟௩௘ௗ ൌ െ ή ʹ݊ߨ݈ ή ‫ ݐ‬ቆ ‫ ݐ‬െ ʹ‫ݎ‬଴ ‫ܭ‬଴ ቇ (19) ability to explain the experimentally observed dissolution
ʹ ߩ௚௟௔௦௦
phenomena as described later.

The final mass loss model kinetic equation in the integral

form is then the following (Equation 20): 4.3 The model extended to short-term and
long-term degradation.
‫ܭ‬଴ଶ
݉ௗ௜௦௦௢௟௩௘ௗ ൌ ݊ߨ݈ ቆʹ‫ݎ‬଴ ‫ܭ‬଴ ‫ ݐ‬െ ‫ ݐ‬ଶ ቇ (20) The reaction kinetics described above apply for a
ߩ௚௟௔௦௦
dissolution-dominated process. As shown in the Chemical
reactions (a)-(l), various competing reactions happen
The solution of Equation 20 is then checked using an simultaneously. Initially these reactions happen at
alternative approach. Based on the mass conservation independent rates, later one process becomes limiting
principle, mass loss in integral form can be also written and dominates the behaviour. Therefore, the degradation
as Equation 21: process should be divided into a short-term non-steady-
state process and a long-term steady-state process since
݉ௗ௜௦௦௢௟௩௘ௗ ൌ ݉଴ െ ݉ ൌ ߩ௚௟௔௦௦ ݊ߨ݈ ሺ‫ݎ‬଴ଶ െ ‫ ݎ‬ଶ ሻ (21) the subprocesses involved follow individual kinetics and
are likely interdependent contributing to the complexity
Combining Equations 16 and 21, of the short-term region. In the non-steady-state region
the process is rather complex, involving in addition to
ଶ dissolution also other competing leaching mechanisms
‫ܭ‬଴ such as ion exchange and gel formation, and thus
݉ௗ௜௦௦௢௟௩௘ௗ ൌ ߩ௚௟௔௦௦ ݊ߨ݈ ൭‫ݎ‬଴ଶ െ ቆ‫ݎ‬଴ െ ‫ݐ‬ቇ ൱ ൌ
ߩ௚௟௔௦௦ cannot be properly described with only one kinetic
‫ܭ‬଴ଶ (22) equation [5,18]. The complexity and varying rates of the
ൌ ݊ߨ݈ ቆʹ‫ݎ‬଴ ‫ܭ‬଴ ‫ ݐ‬െ ‫ ݐ‬ଶ ቇ competing processes in the short-term make this stage
ߩ௚௟௔௦௦
of the degradation to be non-linear non-steady-state. In
The obtained Equation 22 is the same mass loss the long-term hygrothermal degradation of glass fibers,
kinetic equation in the integral form as Equation 20. however, dissolution kinetics become dominant [5,18]
The use of two alternative mathematical ways to obtain and the process becomes then steady-state following the

Authenticated | [email protected]
Download Date | 11/23/18 12:05 PM
 Long-Term Dissolution of Glass Fibers in Water Described by Dissolving Cylinder Zero-Order Kinetic... 1195

Table 1: Model parameters and fit to experimental data (total mass loss mdissolved).

‫ܭ‬଴ூ (g/m2∙s) ‫ܭ‬଴ூூ (g/m2∙s) tst (h) R2 R2

(t ≤ tst) (t > tst)
3.00∙10-9 6.68∙10-10 166 0.8069 0.9653

Table 2: Model parameters for individual released ions and fit to experimental data.

‫ܭ‬଴ூ (g/m2∙s) R2 R2
‫ܭ‬଴ூூ (g/m2∙s)
(t ≤ tst) (t > tst)

Na 6.80∙10-11 1.80∙10-11 0.5326 0.9578

K 4.85∙10-11 9.80∙10-12 0.5583 0.9211
Ca 8.72∙10-10 9.70∙10-11 0.7256 0.9310
Mg 4.85∙10-10 1.02∙10-10 0.7164 0.9898
Fe 2.20∙10-12 4.40∙10-13 0.9261 0.8963
Al 1.45∙10-10 4.68∙10-11 0.7596 0.9272
Si 1.10∙10-9 3.80∙10-10 0.9177 0.9781
Cl 7.80∙10-11 2.28∙10-11 0.6605 0.9572

Figure 3: Experimental data and predicted ion release and total mass loss kinetics.

zero-order kinetics. In the long-term degradation, kinetic exchange, gel formation) taking place in the short-term
equations described above become physical. non-steady-state should be studied and implemented, and
When long-term degradation is of most interest, such their influence on each other should be understood.
as in this work, the shape of the non-linear short-term part The fiber radius decreases at a constant rate (linearly)
is not important, an approximate solution can then be in time. However, the induction period before the linear
used to describe also the short-term non-steady-state stage regime is often observed [16], which can be explained
with dissolving cylinder zero-order kinetic model. In case with the complexity of the short-term non-steady-state
when short-term degradation is of most importance, kinetic process and, to the best knowledge of the authors, was not
models for each individual subprocesses (dissolution, ion explained before.

Authenticated | [email protected]
Download Date | 11/23/18 12:05 PM
1196 Andrey E. Krauklis, Andreas T. Echtermeyer

Table 3: Predicted dissolution times for various mass and radius

loss values at 60°C.

Time in water Mass loss (%) Radius loss (%)

(years)
0.5 0.10 0.05
2.5 0.50 0.25
5.1 1.00 0.50
25.9 5.00 2.53
52.5 10.00 5.13
300.2 50.00 29.30
512.6 75.00 50.02
1025.1 100.00 100.00
Figure 4: Comparison of ion release rates in the steady-state (values
have been calculated from the rate constants in the steady-state).

The final model equations combining approximated

non-steady-state short-term and physically-correct steady-
state long-term dissolution kinetics are proposed.
The radius reduction kinetic model based on Equation
16 is shown in Equation 23:

‫ܭ‬଴ூ
‫ۓ‬ ‫ ݐ‬൑ ‫ݐ‬௦௧ ǣ‫ ݎ‬ൌ ‫ݎ‬଴ െ ‫ݐ‬
ۖ ߩ௚௟௔௦௦
(23)
‫۔‬ ‫ܭ‬଴ூூ
ۖ‫ ݐ‬൐ ‫ݐ‬௦௧ ǣ‫ ݎ‬ൌ ‫ݎ‬௧ೞ೟ െ ሺ‫ ݐ‬െ ‫ݐ‬௦௧ ሻ
‫ە‬ ߩ௚௟௔௦௦

Figure 5: Modelled radius reduction in R-glass fibers after treatment

in water at various times at 60°C. The mass loss kinetic model based on Equation 20 is
shown in Equation 24:
‫ܭ‬଴ூଶ ଶ
‫ۓ‬ ‫ ݐ‬൑ ‫ݐ‬௦௧ ǣ݉ௗ௜௦௦௢௟௩௘ௗ ൌ ݊ߨ݈ ቆʹ‫ݎ‬଴ ‫ܭ‬଴ூ ‫ ݐ‬െ ‫ ݐ‬ቇ
ۖ ߩ௚௟௔௦௦
‫۔‬ ‫ܭ‬଴ூூଶ
ۖ‫ ݐ‬൐ ‫ݐ‬௦௧ ǣ݉ௗ௜௦௦௢௟௩௘ௗ ൌ ݉ௗ௜௦௦௢௟௩௘ௗ೟ೞ೟ ൅ ݊ߨ݈ ቆʹ‫ݎ‬௧ೞ೟ ‫ܭ‬଴ூூ ሺ‫ ݐ‬െ ‫ݐ‬௦௧ ሻ െ ሺ‫ ݐ‬െ ‫ݐ‬௦௧ ሻଶ ቇ
‫ە‬ ߩ௚௟௔௦௦

(24)

where ‫ܭ‬଴ூ and ‫ܭ‬଴ூூ are the rate constants (g/m2∙s) for the
short-term non-steady-state and long-term steady-state
regions, respectively; rtst(m) and ݉ௗ௜௦௦௢௟௩௘ௗ೟ (g) are the
ೞ೟
fiber radius and lost mass after time tst (s), when steady-
state is reached.

4.4 Dissolution of R-glass fibers explained

Figure 6: Digital microscope images of R-glass fibers from left to with the model
right: untreated; after 69 h in water at 60°C; after 3194 h in water at
60°C.
Experimentally mdissolved was measured as a sum of all ions’
release quantified with HR-ICP-MS cumulatively over time.
The whole process can be divided into two regions Alternatively, a gravimetric analysis over time can be used
as shown in Figure 2 and can be described via four to obtain mdissolved. Both kinetic parameters ‫ܭ‬଴ூ and ‫ܭ‬଴ூூ are
parameters: ‫ݎ‬଴ ǡ ‫ݐ‬௦௧ ǡ ‫ܭ‬଴ூ ǡ ‫ܭ‬଴ூூ. determined from obtained mdissolved – t curves by using the

Authenticated | [email protected]
Download Date | 11/23/18 12:05 PM
 Long-Term Dissolution of Glass Fibers in Water Described by Dissolving Cylinder Zero-Order Kinetic... 1197

described methodology in this work for each individual total mass loss, respectively, while all other elements
element, as well as for the total mass loss. contribute less than 7% individually. Once the steady-
Ions released from the studied glass material are the state is reached, it can be assumed that the reactions do
following: Na, K, Ca, Mg, Fe, Al, Si, Cl. The total mass of not have any effect on each other and proceed at constant
all ions released is the cumulative mass loss, as shown by independent rates. It can be speculated that in the steady-
Equation 25: state some equilibrium composition of glass (different
from the bulk composition) is obtained in the outer layers
݉ௗ௜௦௦௢௟௩௘ௗ ൌ ݉ௗ௜௦௦௢௟௩௘ௗಿೌ ൅ ݉ௗ௜௦௦௢௟௩௘ௗ಼ ൅ ݉ௗ௜௦௦௢௟௩௘ௗ಴ೌ ൅ ݉ௗ௜௦௦௢௟௩௘ௗಾ೒ ൅ ݉ௗ௜௦௦௢௟௩௘ௗಷ೐ that are contact with water, allowing elements to dissolve
at some limiting rate and proportionally to their content in
൅ ݉ௗ௜௦௦௢௟௩௘ௗಲ೗ ൅ ݉ௗ௜௦௦௢௟௩௘ௗೄ೔ ൅ ݉ௗ௜௦௦௢௟௩௘ௗ಴೗ 
(25) the outer layers of the glass.
Radius reduction in R-glass fibers after treatment in
The total mass loss is described by four parameters water at various times at 60°C predicted with the model is
( ‫ݎ‬଴ ǡ ‫ݐ‬௦௧ ǡ ‫ܭ‬଴ூ ǡ ‫ܭ‬଴ூூ). For each ion, individual ‫ܭ‬଴ூ and ‫ܭ‬଴ூூ values shown in Figure 5. The fiber radius decrease could not be
have to be obtained, while tst and r0 for each individual ion captured with a digital microscope. Digital micrographs of
release are the same and are equivalent to that for the total fibers, untreated and treated in water are shown in Figure
mass loss. 6.
From the cumulative data of ions released altogether The lack of accuracy is reasonable, taking into
(mdissolved), the kinetic model parameters (‫ܭ‬଴ூ and ‫ܭ‬଴ூூ ) were account that the radius reduction predicted by the model
obtained using non-linear regression for each separate ion in 3194 h time is only 0.0036 μm (0.04% radius loss) as
as well as for the total mass loss curve. The Generalized seen in Figure 5, and the microscope resolution is 0.16 μm.
Reduced Gradient (GRG) non-linear regression method Furthermore, a statistical error has to be considered since
was used by minimizing the sum of squares of the the radius is not perfectly equal in different locations
differences between modelled and experimental values. of the fibers (as seen by discrepancy in measurements
The obtained parameters and model fit to experimental reported in Figure 6).
data are systematized in Table 1. Obtained values are With the reaction rates obtained in this work, the
reasonable (similar order of magnitude) compared with full dissolution of the studied R-glass fibers would take
dissolution rates of other glass fibers studied in literature 8.98∙106 h or 1025.1 years as reported in Table 3. The radius
[25]. loss after 25 years, a typical design lifetime, would be
Using the obtained tst, these parameters are included 2.45%. These values are for 60°C. For lower temperatures,
in the model for each individual ion release (Na, K, Ca, the radius reduction would be even less [5,6,18]. The effect
Mg, Fe, Al, Si, Cl). ‫ܭ‬଴ூ and ‫ܭ‬଴ூூ are obtained for each ion of temperature on dissolution kinetics and the effect of
by regression of experimental data for each individual ion cracks in the fibers of this particular glass material will be
cumulative release curve. The same method of non-linear studied in the future.
regression as described above was used in each case. The
rate constants ‫ܭ‬଴ூ and ‫ܭ‬଴ூூ for the identified released ions
are systematized in Table 2. It should be noted that the R2 5 Conclusions
values in the non-steady-state (t ≤ tst) are fairly low, while
the fit is good in the steady-state region (t > tst). As was In this study, the long-term (3194 h) dissolution of R-glass
mentioned earlier, this is because the model can only be fibers in water was studied experimentally. A dissolving
used as a mathematical approximation in the non-steady- cylinder zero-order kinetic model was developed and
state, while it is physical in the steady-state long-term. successfully used to describe the kinetics of mass loss and
Experimental data and modelled curves for individual the fiber’s radius reduction during dissolution in water.
ion release and total mass loss are shown in Figure 3. Elements released during degradation were
Experimentally mass loss was obtained from HR-ICP-MS determined to be Na, K, Ca, Mg, Fe, Al, Si and Cl. The
measurements for separate ions being released. total material loss and release of each separate ion was
Rates of ion release during steady-state for the studied modelled using the developed kinetic equations, and rate
material are compared in Figure 4. constants were obtained and reported. The contribution
The contribution from Si to the total mass loss is the of Si to the total mass loss was the largest (56.1% by mass)
largest (56.1% by mass) and seems to govern the dissolution and governed the dissolution process.
process. Ca and Mg are released at approximately similar The model differentiates between the complex short-
rates to each other and contribute 14.3 and 15.1% to the term and dissolution-dominated long-term processes.

Authenticated | [email protected]
Download Date | 11/23/18 12:05 PM
1198 Andrey E. Krauklis, Andreas T. Echtermeyer

Furthermore, the novelty of the model is the ability to bundle tensile strength, Compos. Sci. Technol., 2005, 65,
describe both dissolution and radius reduction kinetics 129–136.
[5] Grambow B., Müller R., First-order dissolution rate law and the
without the necessity for introducing additional terms
role of surface layers in glass performance assessment, J. Nucl.
such as a conversion factor. The model is able to predict Mater., 2001, 298(1-2), 112-124.
both mass loss and radius reduction kinetics using [6] Grambow B., A General Rate Equation for Nuclear Waste Glass
the same four parameters, which can be determined Corrosion, Mat. Res. Soc. Symp. Proc., 1985, 44, 15-27.
experimentally. The methodology provided in this work [7] Krauklis A.E., Gagani A.I., Echtermeyer A.T., Near-Infrared
offers the guidelines to obtain the required parameters to Spectroscopic Method for Monitoring Water Content in Epoxy
Resins and Fiber-Reinforced Composites, Materials, 2018,
use the model in practice. The developed model is useful
11(4), 586-599.
for both monitoring the radius of degrading fibers and for [8] Stamenović M.R., Putić S.S., Rakin M.B., Medjo B., Čikara D.,
monitoring the time evolution of the loss of material. Effect of alkaline and acidic solutions on the tensile properties
The radius reduction was found to be linear with time of glass–polyester pipes, Mater. Des., 2011, 32(4), 2456-2461.
during dissolution when a steady-state is reached. The [9] Krauklis A.E., Echtermeyer A.T., Mechanism of Yellowing:
Carbonyl Formation during Hygrothermal Aging in a Common
zero-order kinetic constant and the density of the glass
Amine Epoxy, Polymers, 2018, 10(9), 1017-1031.
describe the rate (proportionality) of the dissolution. [10] Amaro A.M., Reis P.N.B., Neto M.A., Louro C., Effects of alkaline
Radius reduction predicted by the model in 3194 h time and acid solutions on glass/epoxy composites, Polym. Degrad.
was only 0.0036 μm (0.04% radius loss) at 60°C and could Stab., 2013, 98(4), 853-862.
not be captured with a microscope. The full dissolution of [11] Delage F., Ghaleb D., Dussossoy J.L., Chevallier O., Vernaz E.,
the studied R-glass fibers would take 8.98∙106 h or 1025.1 A mechanistic model for understanding nuclear waste glass
dissolution, J. Nucl. Mater., 1992, 190, 191-197.
years. The radius loss after 25 years, a typical design
[12] Geisler-Wierwille T., Nagel T.J., Kilburn M.R., Janssen A.,
lifetime, would be 2.45%. Icenhower J., Fonseca R.O.C., Grange M.L., Nemchin A.A., The
Mechanism of Borosilicate Glass Corrosion Revisited, Geochim.
Acknowledgments: This work is part of the DNV GL Cosmochim. Acta, 2015, 158, 112-129.
led Joint Industry Project “Affordable Composites” with [13] Icenhower J., Steefel C.I., Dissolution Rate of Borosilicate Glass
SON68: A Method of Quantification Based upon Interferometry
twelve industrial partners and the Norwegian University
and Implications for Experimental and Natural Weathering
of Science and Technology (NTNU). The authors would Rates of Glass, Geochim. Cosmochim. Acta, 2015, 157, 147-163.
like to express their thanks for the financial support by [14] Ma T., Jivkov A.P., Li W., Liang W., Wang Y., Xu H., Han X.,
The Research Council of Norway (Project 245606/E30 in A mechanistic model for long-term nuclear waste glass
the Petromaks 2 programme). The authors are thankful dissolution integrating chemical affinity and interfacial
to Abedin I. Gagani, Syverin Lierhagen and Melanie diffusion barrier, J. Nucl. Mater., 2017, 486, 70-85.
[15] Mišíková L., Liška M., Galusková D., CORROSION OF E-GLASS
Shebel. The first author is especially grateful to Oksana V.
FIBERS IN DISTILLED WATER, Ceram. Silikaty, 2007, 51(3), 131-
Golubova. 135.
[16] Bashir S.T., Yang L., Liggat J.J., Thomason J.L., Kinetics of
Conflict of interest: Authors declare no conflict of dissolution of glass fibre in hot alkaline solution, J. Mater. Sci.,
interest. 2018, 53(3), 1710-1722.
[17] Krauklis A.E., Echtermeyer A.T., Dissolving Cylinder Zero-Order
Kinetic Model for Predicting Hygrothermal Aging of Glass Fiber
Bundles and Fiber-Reinforced Composites, International Glass
References [18]
Fiber Symposium, 2018.
Report RWM005105, AMEC/103498/02 Issue 2: Review of glass
dissolution models and application to UK glasses, Didcot,
[1] Wallenberger F.T., Commercial and Experimental Glass Fibers,
2015.
in: Fiberglass and Glass Technology, Eds: Wallenberger F.T.,
[19] Costa P., Sousa Lobo J.M., Review: Modeling and comparison of
Bingham P.A., Springer, US, 2010.
dissolution profiles, Eur. J. Pharm. Sci., 2001, 13, 123-133.
[2] Steinmann W., Saelhoff A.-K., Essential Properties of Fibres
[20] Khawam A., Flanagan D.R., Solid-State Kinetic Models: Basics
for Composite Applications, in: Fibrous and Textile Materials
and Mathematical Fundamentals, J. Phys. Chem. B, 2006, 110,
for Composite Applications, Textile Science and Clothing
17315-17328.
Technology, Eds.: Rana S., Fangueiro R., Springer, Singapore,
[21] International Standard ISO 2078:1993 (revised in 2014), Textile
2016.
glass – Yarns – Designation, 2014.
[3] Tournié A., Ricciardi P., Colomban P., Glass Corrosion
[22] 3B Fibreglass technical data sheet. HiPer-tex W2020 rovings,
Mechanisms: A Multiscale Analysis, Solid State Ionics, 2008,
Belgium, 2012.
179(38), 2142-2154.
[23] Li H., Gu P., Watson J., Meng J., Acid corrosion resistance and
[4] Brown E.N., Davis A.K., Jonnalagadda K.D., Sottos N.R., Effect
mechanism of E-glass fibers: boron factor, J. Mater. Sci., 2013,
of surface treatment on the hydrolytic stability of E-glass fiber
48(8), 3075-3087.

Authenticated | [email protected]
Download Date | 11/23/18 12:05 PM
 Long-Term Dissolution of Glass Fibers in Water Described by Dissolving Cylinder Zero-Order Kinetic... 1199

[24] Čornaja S., Fizikālā ķīmija. Elektroķīmija. Kinētika, RTU

Izdevniecība, Rīga, 2008, (in Latvian).
[25] Eastes W., Potter R.M., Hadley J.G., Estimating in-vitro glass
fiber dissolution rate from composition, Inhal. Toxicol., 2000,
12, 269-280.

Authenticated | [email protected]
Download Date | 11/23/18 12:05 PM
APPENDIX F

PAPER VI

KRAUKLIS A.E., ECHTERMEYER A.T.

DISSOLVING CYLINDER ZERO-ORDER KINETIC MODEL FOR PREDICTING

HYGROTHERMAL AGING OF GLASS FIBER BUNDLES AND FIBER-REINFORCED
COMPOSITES.

4TH INTERNATIONAL GLASS FIBER SYMPOSIUM (GERMANY), PUBLISHER: MAINZ, G,

AACHEN, 2018, 66–72.

ISBN: 978-3-95886-249-4

Not included due to copyright restrictions

PAPER VI

141
142
APPENDIX G

PAPER VII

KRAUKLIS A.E., GAGANI A.I., VEGERE K., KALNINA I., KLAVINS M., ECHTERMEYER A.T.

DISSOLUTION KINETICS OF R-GLASS FIBRES: INFLUENCE OF WATER ACIDITY,

TEMPERATURE, AND STRESS CORROSION.

FIBERS (SWITZERLAND), 7(3), 2019, 22-40, IN A SPECIAL ISSUE: ADVANCES IN GLASS FIBERS.

DOI:10.3390/FIB7030022

PAPER VII

151
152
 fibers
Article
Dissolution Kinetics of R-Glass Fibres: Influence of
Water Acidity, Temperature, and Stress Corrosion
Andrey E. Krauklis 1, * , Abedin I. Gagani 1 , Kristine Vegere 2 , Ilze Kalnina 2 , Maris Klavins 3
and Andreas T. Echtermeyer 1
1 Department of Mechanical and Industrial Engineering (past: Department of Engineering Design and
Materials), Norwegian University of Science and Technology, 7491 Trondheim, Norway;
[email protected] (A.I.G.); [email protected] (A.T.E.)
2 Institute of General Chemical Engineering, Faculty of Materials Science and Applied Chemistry, Riga
Technical University, P. Valdena str. 3/7, Riga LV-1048, Latvia; [email protected] (K.V.);
[email protected] (I.K.)
3 Department of Environmental Science, University of Latvia, Riga LV-1004, Latvia; [email protected]
* Correspondence: [email protected] or [email protected]; Tel.: +371-268-10288

Received: 11 February 2019; Accepted: 5 March 2019; Published: 12 March 2019 

Abstract: Glass fibres slowly degrade due to dissolution when exposed to water. Such environmental
aging results in the deterioration of the mechanical properties. In structural offshore and marine
applications, as well as in the wind energy sector, R-glass fibre composites are continuously exposed
to water and humid environments for decades, with a typical design lifetime being around 25 years
or more. During this lifetime, these materials are affected by various temperatures, acidity levels,
and mechanical loads. A Dissolving Cylinder Zero-Order Kinetic (DCZOK) model was able to
explain the long-term dissolution of R-glass fibres, considering the influence of the pH, temperature,
and stress corrosion. The effects of these environmental conditions on the dissolution rate constants
and activation energies of dissolution were obtained. Experimentally, dissolution was measured using
High Resolution Inductively Coupled Plasma Mass Spectrometry (HR-ICP-MS). For stress corrosion,
a custom rig was designed and used. The temperature showed an Arrhenius-type influence on the
kinetics, increasing the rate of dissolution exponentially with increasing temperature. In comparison
with neutral conditions, basic and acidic aqueous environments showed an increase in the dissolution
rates, affecting the lifetime of glass fibres negatively. External loads also increased glass dissolution
rates due to stress corrosion. The model was able to capture all of these effects.

Keywords: glass; fibres; model; dissolution; kinetics; water; environmental; aging; stress corrosion

1. Introduction
Glass fibres (GFs) are the most common fibrous reinforcement material used in fibre-reinforced
composites (fibre-reinforced polymers; FRPs) [1]. The GFs possess desirable reinforcement material
properties, such as high hardness, strength, and stiffness [2,3]. Various types of GFs exist, such as E,
ECR, R, and S-glass, listed in the order of their increasing mechanical strength. Glass fibre-reinforced
composites (GFRPs) are often used in structural applications in marine and offshore industries, as well
as in the wind energy sector [4–11]. In these applications, GFRPs are continuously exposed to water
and humid environments for decades, with a typical design lifetime being around 25 years or more [8].
When exposed to such environments, hydrolytic degradation of the glass material occurs [12,13],
leading to a reduction of the mechanical strength and the initiation of corrosion-induced defects in
fibres and composites [14]. The degradation of glass fibres due to environmental attacks can severely
affect the performance of GFRPs [1]. Aqueous environments clearly have a negative effect on the

Fibers 2019, 7, 22; doi:10.3390/fib7030022 www.mdpi.com/journal/fibers

Fibers 2019, 7, 22 2 of 18

fatigue properties of GFRPs and such environments act primarily to reduce the fibre strength [15].
The fact that even water may corrode glass fibres has been known for many years [15]. In the long-term,
hydrolytic degradation occurs mainly via the glass dissolution mechanism [12]. Thus, understanding
the dissolution behaviour of glass is necessary to predict the deterioration of the composite’s physical
and mechanical properties [16].
The long-term environmental degradation of glass materials has been previously studied, mainly
with respect to nuclear waste applications [17,18]. Most of the existing works on glass dissolution have
been performed with bulk silicate glass and fibres are not studied often [12]. Recently, there has been
an increasing interest in the environmental aging of FRPs, where R-glass and E-glass are often used as
reinforcement [4,12,16,19–21]. Even so, very few studies exist on the kinetics of GF dissolution (mostly
on E-glass) [22,23] and even less on R-glass [12,16].
The degradation of GFs in water can be divided into two kinetic regions (Phases I and II). Phase I
is the short-term non-steady-state. Phase II is the long-term steady-state [12,16]. Phase I is complex
and involves such processes as ion exchange, gel formation, and dissolution [12,17]. However, once the
long-term kinetic region is reached, glass dissolution becomes a dominant process [12,17,24]. For the
studied GFs, this transition to steady-state occurs in about a week (166 h) [12,16]. This time is very
negligible compared to the typical design lifetime of about 25 years or more for the structures in which
these materials are implemented [8]. Thus, as in the case of structural applications, when the long-term
degradation is mainly of interest, dissolution is the governing process, which has to be understood
and well-predictable.
During glass-water interactions, several chemical reactions may occur, shown in the chemical
reactions (1)–(11) [12,17,24,25]:

(≡ Si − ONa) + H2 O → (≡ Si − OH ) + OH − + Na+ (1)

(≡ Si − OK ) + H2 O → (≡ Si − OH ) + OH − + K + (2)
− 2+
(≡ Si − O)2 Ca + H2 O → 2(≡ Si − OH ) + 2OH + Ca (3)
− 2+
(≡ Si − O)2 Mg + H2 O → 2(≡ Si − OH ) + 2OH + Mg (4)

(≡ Si − O − Al =) + H2 O ↔ (≡ Si − OH ) + (= Al − OH ) (5)

(≡ Si − O)2 Fe + H2 O → 2(≡ Si − OH ) + 2OH − + Fe2+ (6)

− 3+
(≡ Si − O)3 Fe + H2 O → 3(≡ Si − OH ) + 3OH + Fe (7)
− −
(≡ Si − O − Si ≡) + OH ↔ (≡ Si − OH ) + ≡ Si − O (8)
− −
≡ Si − O + H2 O ↔ (≡ Si − OH ) + OH (9)

SiO2 + 2H2 O ↔ H4 SiO4 (10)

H O
MeCl x →
2
Me x+ + xCl − (11)

The chemical reaction (10) can also be written as a combination of the subsequent reactions (12)
and (13), meaning that initially H2 SiO3 is formed, which dissociates weakly and further reacts with
water to form silicic acid:
SiO2 + H2 O ↔ H2 SiO3 (12)

H2 SiO3 + H2 O ↔ H4 SiO4 (13)

The elements that are released during the degradation of R-glass are Na, K, Ca, Mg, Fe, Al, Si,
and Cl [12]. The cumulative mass loss is the summary mass of all of the ions released [12]. The total
mass loss is what causes the radius reduction [12]. As shown in the chemical reactions (1)–(11), various
competing reactions happen in parallel. During Phase I, these reactions happen at independent rates,
Fibers 2019, 7, 22 3 of 18

then one process later becomes limiting and dominates the behaviour (Phase II). Thus, the degradation
process should be divided into two aforementioned stages. Furthermore, the sub-processes may follow
individual kinetics and may be interdependent during Phase I [12]. During Phase I, the process is
rather complex. It involves competing leaching mechanisms in addition to the dissolution reactions,
i.e., ion exchange and gel formation. Therefore, it cannot be fully described with just one kinetic
equation [12,17,24]. The complexity and varying rates of the competing processes during Phase I turn
this stage of the glass degradation into a non-linear and non-steady-state process.
During Phase II, however, dissolution becomes dominant [17,24] and the process becomes
steady-state, governed by the dissolution kinetics [12,16]. For R-glass, the contribution from Si to the
total mass loss is the largest (56.1% by mass) and governs the long-term steady-state dissolution [12].
Once the steady-state is reached, the reactions proceed at constant independent rates [12]. It may be
speculated that, during Phase II, an equilibrium composition of glass (likely different from the bulk
composition) is obtained in the outer layers in contact with water. This allows elements to dissolve
at a limiting rate and proportionally to their content in the outer layers of the glass, being indicative
of the glass outer layer composition [12]. In other words, the long-term dissolution is dominated by
the slowest dissolving element and the other elements following proportionally, at some equilibrium
composition [12].
The non-linear non-steady-state stage of degradation can be approximated using the Dissolving
Cylinder Zero-Order Kinetic (DCZOK) model, when the exact shape of the non-linear short-term part
is not important, i.e., when the long-term degradation is of the most interest, such as in cases of the
long service life of structures implementing GFs [8,12].
Glass dissolution kinetics depend on the surface area of the glass in contact with water.
This surface area is proportional to the glass fibre radius. As the dissolution proceeds, the radius
decreases and results in the deceleration of the mass loss; the DCZOK model accounts for this
effect [12,16]. Glass dissolution kinetics are affected by the pH, temperature, glass composition
(i.e., E, ECR, R, S-glass), solution composition (distilled water or seawater), surface area of
the glass, the protective effect of the sizing, and the ionic strength of the solution, to a lesser
extent. The temperature and pH are considered to be key parameters for the long-term glass
dissolution [16,17,20,24,26,27]. Elevated temperatures accelerate the dissolution since the dissolution
is exothermic [17,20,26]. The activity of the water is dependent on the composition of the
aqueous environment, notably, on the, as well as ionic concentration of some species in the glass
composition [28]. Interestingly enough, for R-glass, the sizing was found to slow down the dissolution
kinetics of the GFs by almost an order of magnitude [16]. In addition, external loads may have
an influence on glass dissolution kinetics, due to stress corrosion [27,28].
Humid and water environments reduce the strength of silicate glasses due to the growth of
flaws, such as surface cracks, under the combination of stress and chemical attacks, known as stress
corrosion [1,15]. The glass corrosion (dissolution) rate is accelerated by this stress [1].
The classical theory of stress corrosion involves the chemical reaction of a water molecule with
silica, and was described elsewhere [27]. A model and the details of the chemical interaction between
the adsorbed water and the Si–O bond was described by Michalske and Freiman [29]. Their model
showed that water is efficient in the stress-activated corrosion of glass [27]. The basic mechanism of
the stress corrosion reaction was related to the stress-enhanced thermal activation of a dissociative
hydrolysis reaction, presented in Figure 1 [28,29].
Fibers 2019, 7, 22 4 of 18

Figure 1. The chemical mechanism of stress corrosion in glass, proposed by Michalske and Freiman [29].

Any molecule that is able to donate a proton (hydrogen ion and has a lone pair of electrons,
such as water, can react with the strained Si–O bond at the tip of a crack, provided that the molecular
diameter is smaller than 0.5 nm [30]. The first property is necessary for the harmonized reaction of
adsorption and the chemical scission of the siloxane bridges of the glass network. The small diameter
is necessary to avoid steric problems, so that the reactive water molecules can reach the strained bonds
at the crack tip [28]. The constrained Si–O bonds have an enhanced chemical reactivity with water
due to the intensity of the strain, as was shown via hydrolysis experiments for different siloxane
ring structures, after [31]. Michalske and Bunker [32] also supported the effect of stress on the stress
corrosion reaction rate. They proved this hypothesis by conducting molecular-orbital simulations of
the stress corrosion reaction on distorted siloxane bonds. Several other reactions are possible for other
glass compositions, but they all share the feature of being stress-enhanced and thermally activated,
therefore explaining the exponential (Arrhenius type) relationship [28]. In addition, it is believed
that the interaction of water with cations in the flaws leads to hydrolysis and a local increase in the
concentration of hydroxyl ions, leading to an increase in the pH locally [33].
Various approaches have been proposed on how to model the kinetics of glass degradation.
Grambow has developed a general rate equation for the reaction of glass in nuclear wastewater
solutions with different pH levels, surface areas, times, temperatures, and compositions of the
reagents [18]. Delage et al. described the dissolution of borosilicate glass in aqueous media using
a combination of surface reaction kinetics and diffusion [34]. Mišíková et al. have studied the
kinetics of the leaching of E-glass fibres in distilled water at various temperatures, linking the
glass fibre dissolution to the temperature using statistical methods [22]. Tournié et al. presented
an extensive experimental study of the glass corrosion mechanism in both acidic and basic solutions.
They reported that a strong acid attack was dependent on the composition of the glass [13]. Ma et al.
presented a model for the glass dissolution, based on the combination of the chemical affinity and
diffusion [35]. Bashir et al. studied the kinetics of the dissolution of E-glass fibres in hot alkaline
solutions using zero-order and shrinking cylinder models [23,36]. They reported that both an increased
temperature and increased alkalinity (higher pH) accelerated the dissolution reaction rates [23].
Bashir et al. concluded that the rate-limiting step was either the glass dissolution itself or the diffusion
of hydroxide ions [23]. Krauklis and Echtermeyer developed and presented a Dissolving Cylinder
Zero-Order Kinetic (DCZOK) model that explained the long-term dissolution of R-glass in water for
both fibres [12,16] and GFRPs [16]. The DCZOK model is able to predict both the mass loss and the
radius reduction kinetics due to dissolution [12].
Glasses formulated with significant amounts of various metal oxides, such as R-glass, have rarely
been studied, but they are becoming popular for structural applications [12,18]. Thus, this work focuses
on the dissolution kinetics of R-glass in water at various environmental conditions (pH, temperature,
stress corrosion). However, the model should be applicable to other types of glass as well, as SiO2 is
the major component in virtually all types of glass [23].
The aim of this work is to study the effects of the pH, temperature, and stress corrosion on the
R-glass fibre dissolution kinetics, in terms of the rate constants and activation energies, and to capture
these effects using the Dissolving Cylinder Zero-Order Kinetics (DCZOK) model.
Fibers 2019, 7, 22 5 of 18

2. Materials and Methods

2.1. Materials
A typical glass fibre used for marine, oil, and gas applications was selected for this study.
Boron-free and fluorine-free high strength, high modulus 3B HiPer-Tex™ W2020 R-glass fibre bundles
were used. These were classified as high-strength, high modulus R-glass (defined by an international
standard [37]). The average fibre diameter was 17 ± 2 μm [12,38]. The density of the glass (ρ f ) was 2.54
g/cm3 [12,38]. The authors estimated that a bundle had about 4098 fibres [12]. The specific surface area
of the glass fibres was determined to be 0.09 m2 /g from the geometrical considerations, as a product
of the number, circumference, and length of the fibres [12]. All of the fibres used during this work were
sized, whereas the dissolution of unsized R-glass fibres was previously studied in another study by
Krauklis and Echtermeyer [16]. A micrograph of R-glass fibre bundles is shown in Figure 2.

Figure 2. A micrograph of R-glass fibre bundles taken with a digital microscope, the Keyence VHX6000
(Osaka, Japan).

Distilled water (0.5–1.0 MΩ·cm) was used for the conditioning of the GF bundles, produced using
a water purification system, the Aquatron A4000 (Cole-Parmer, Vernon Hills, IL, USA). The pH of
the distilled water was 5.650 ± 0.010, being lower than neutral due to the dissolved CO2 from the
atmosphere in equilibrium. IUPAC standard buffer solutions (Radiometer analytical, Lyon, France)
were used for studying the effect of the pH on the kinetics of the GF dissolution. The solutions of pH
1.679 ± 0.010, 4.005 ± 0.010, 5.650 ± 0.010, 7.000 ± 0.010, and 10.012 ± 0.010 were used.

2.2. Methods

Glass Dissolution Experiments:

Aqueous dissolution experiments of GF samples were performed using a batch system.
The samples for the dissolution study were weighed using AG204 analytical scales (±0.1 mg; Mettler
Toledo, OH, USA) before the experiments. The samples were placed in inert closed vessels filled
with 50 mL of distilled water or pH buffer solutions. The tight sealing of samples was ensured.
The water-tight vessels, with the samples and water solutions in them, were placed into the water bath.
The water’s temperature (25, 40, 60, 80 ◦ C) in the bath was controlled via proportional–integral–
derivative (PID)-controlled heating, giving an accuracy of ±1 ◦ C. A two-stage heating system was
used in order to make sure that there is no contact of the sample water with potential contaminating
ion release sources, i.e., the heating element.
pH measurements were performed using a standard MeterLab PHM210 pH-meter (Radiometer
analytical, France) calibrated with IUPAC standard buffer solutions, providing an accuracy of
±0.010 pH.
Fibers 2019, 7, 22 6 of 18

The concentration of the dissolved ions in the water from the vessels was analysed over time
via high resolution inductively coupled plasma mass spectrometry (HR-ICP-MS), providing glass
dissolution kinetics. Experimentally, the total mass loss of glass material was measured as a sum
of all of the ions’ release, quantified with HR-ICP-MS. Analyses were performed using a double
focusing magnetic sector field ICP-MS Finnigan ELEMENT 2 (Thermo-Scientific, Waltham, MA, USA),
equipped with a PrepFAST sample introduction system (ESI/Elemental Scientific) and an UltraClave
pre-treatment/digestion machine (Milestone, Brøndby Kommune, Denmark). The acidification of
samples was performed using an ultra-pure grade HNO3 SubPur (Milestone) in order to avoid the
adsorption of ions into the wall of the sample vials that would otherwise give erroneous results.
The benefit of HR-ICP-MS versus gravimetric analysis is that it allows the measurement of the
dissolution kinetics of each separate ion, as well as the total mass loss [12]. The data obtained from the
HR-ICP-MS experiments are in the form of the non-cumulative mass concentration at each point in
time (g/L) and need to be converted to the cumulative mdissolved form by using Equation (14):
 t
mdissolved = Vwater cdt (14)
0

where Vwater is the volume of a water sample using the HR-ICP-MS measurement. The Vwater used
for the experiments was 50 mL. Equation (14) is valid for each individual ion release and for the total
mass loss.
For studying the influence of the external loads on the dissolution kinetics, a stress corrosion
rig was designed, built, and used during the experiments. The experimental rig is shown in Figure 3
and the design schematics of the stress corrosion setup are attached in Appendix A. The principle is
that glass fibres are inside an inert cylinder pushed by an inert rod, which transfers the stress from
the weights to the fibres. The water samples for the HR-ICP-MS analyses were taken from the main
cylinder, where the distilled water was in contact with the stressed glass fibres. The temperature was
PID-controlled by using a two-stage heating system, ensuring a temperature of 60 ± 1 ◦ C. Polymeric
spheres, seen in Figure 3, were used to reduce the evaporation rate of the heating water.

Figure 3. The glass fibre stress corrosion experimental setup.

Fibers 2019, 7, 22 7 of 18

3. Dissolving Cylinder Zero-Order Kinetic (DCZOK) Model

3.1. The Model and Its Assumptions

An approach used to describe the glass dissolution kinetics in the fibres was proposed and
explained in works by Krauklis and Echtermeyer [12,16]. This approach is used as a basis for this
study. In this work, the kinetic constants are obtained for the Dissolving Cylinder Zero-Order Kinetic
(DCZOK) model for various pH and temperature conditions, as well as for different stress levels.
The protective effect of sizing is accounted for in this model [16].
The rate of the dissolution is dependent on the glass dissolution rate constant (K0 ), the glass
surface area exposed to water (S), and the effect of sizing (ξ sizing ) [12]. For fibres, in infinite water
availability conditions, the dissolution, being a surface reaction, can be well-described with zero-order
kinetics [23], but the decrease in the fibre radius, and thus in the surface area over time, should be
accounted for [12]. For sizeless fibres, Equation (15) should be used [16]:

∂m
= K0 S ( t ) (15)
∂t
For sized fibres, the effect of sizing, ξ sizing , should also be accounted for, as shown in
Equation (16) [16]:
∂m
= K0 ξ sizing S(t) (16)
∂t
where m is the total cumulative mass dissolved after time t, K0 is a material/environment interaction
property, and ξ sizing is the protective effect of sizing. K0 is affected by environmental effects [12].
In this Equation (16), K0 and ξ sizing are the time-independent parameters, whereas S changes
with time. As the dissolution continues, the fibre’s radius is reduced and the total surface area
(S) is decreased. This leads to a decrease in the dissolution rate. The ion release rate slows down
linearly with a decrease in the fibre’s radius [12]. The glass fibre’s radius reduction is linear with
time. The proportionality is given by the dissolution rate constant (K0 (g/(m2 ·s))) and the density
of the glass (ρ glass (g/m3 )) [12]. A linear radius decrease of the glass fibres over time was previously
experimentally observed elsewhere [23] and is accounted for in the DCZOK model [12].
The effect of sizing on the glass dissolution (ξ sizing ) for the studied R-glass is 0.165, protecting the
fibres from water by almost an order of magnitude, in terms of dissolution rates. The effect of sizing
(ξ sizing ) is assumed to be time-independent [16]. ξ sizing goes into the model linearly, since fibres are
covered in “islands” or patches of sizing, meaning not the whole surface is coated [39]. These areas of
the bare unsized glass dominate the dissolution behaviour (the difference being roughly an order of
magnitude) [16].
Taking into account the geometry of the fibres, the mass loss kinetic DCZOK model equation in
its differential form is the following, as shown in Equation (17) [12]:
 
∂m K02 ξ sizing
2
= 2nπl r0 K0 ξ sizing − t (17)
∂t ρ glass

where n is the number of fibres (-), l is the length of the fibres (m), r0 is the initial fibre radius (m),
and ρ glass is the density of the glass (g/m3 ).
 The model showsthe rate  of mass loss due to dissolution. It can be used for the total mass loss
∂m ∂mSi
∂t , as well as for Si ∂t or another ion’s release.
As can be seen from the differential expression of the DCZOK model in Equation (17), the radius
reduction over time is accounted for in the model. The pH, temperature, and stress corrosion affect the
material-environment energy-activated interactions, thus affecting the dissolution rate constants (K0 ),
as represented by Equation (18).
K0 = f ( pH, T, σ) (18)
Fibers 2019, 7, 22 8 of 18

where pH is the acidity of the environment (-), T is the temperature (K), and σ is the stress (MPa).
The model involves the following assumptions: the model is deterministic and all of the glass
fibres are assumed to have the same initial fibre radius (r0 ); the cross-sectional surface area at the ends
of the fibres is assumed negligible in the surface area calculations; the length of the fibres (l ) is assumed
to be constant during the whole dissolution process. During the whole degradation process, the density
of the glass material is assumed to be constant (ρ glass ). The effect of sizing (ξ sizing ) is assumed to be
independent of the environmental conditions and time [12,16]. For free fibre bundles (not embedded
in the composite), the condition of the infinite availability of water is ensured by using large volumes
of water, thus making the rate of the reaction independent of the water concentration [12,16].

3.2. The Effect of the Environment on the Dissolution Rate Constant

It has been reported that the temperature dependence of the rate of the glass fibre radius reduction
follows an exponential trend and the diameter reduction is linearly related to the loss of material
in fibres due to dissolution [12,23]. The fact that the temperature dependence of a dissolution rate
constant is described by the Arrhenius equation is also confirmed by the evidence reported further in
this manuscript. The temperature dependence of a dissolution rate constant can be described using
the Arrhenius equation (Equation (19)), being an exponential function:
E A ( pH,σ)
K0 = Ae− RT (19)

where A is the pre-exponential factor (g/(m2 ·s)), R is the gas constant being 8.314 J/(mol·K), T is the
absolute temperature (K), and E A is the activation energy (J/mol). Both the pH and stress corrosion
are thought to affect the activation energy term in the Arrhenius equation [31,40].
The activation energy can be obtained through the graphing of the constant pH and σ,
using Equation (20) [18]:
E 1
lnK0 = − A + lnA (20)
R T
The rate constant can be calculated for each temperature from the general differential mass loss
DCZOK model equation. To obtain the E A , the lnK0 is plotted against T1 . The pre-exponential factor
( A) is assumed to be constant. The DCZOK model can be expanded (Equation (21)):
 2 
∂m (K ξ sizing )
∂t = 2nπl r0 K0 ξ sizing − 0ρglass t
⎛  2 ⎞
E ( pH,σ)
− A RT
⎜ E ( pH,σ)
Ae ξ sizing
⎟ (21)
= 2nπl ⎜
⎝r0 Ae
− A RT ξ sizing − ρ glass t⎟
⎠

3.3. Modelling of the Two Distinct Stages of Glass Degradation

Mass loss curves made during the aging of glass over time can be divided into two parts:
Short-term non-steady-state degradation and long-term steady-state dissolution [12]. When the
long-term dissolution (Phase II) is of the most interest, the shape of the non-linear short-term stage
(Phase I) is not important. In such case, an approximate solution can then be used to describe the Phase I
of degradation as well, using the DCZOK model. In the case when the short-term degradation (Phase i)
is of the most interest, the kinetic models for each of the individual sub-processes, i.e., ion exchange,
gel formation, and dissolution, should be implemented.
The model can also predict the fibre radius reduction kinetics during the dissolution, using the
same four parameters that can be determined experimentally [12]. The DCZOK model can be used
to describe the dissolution for each element separately, as well as for the total mass loss. The glass
mass loss (all elements included) is described by four parameters (r0 , tst , K0I , K0I I ). For each ion,
the individual K0I and K0I I values have to be obtained, while the tst and r0 for each individual ion
Fibers 2019, 7, 22 9 of 18

release are the same and are equivalent to that for the total mass loss [12]. K0I and K0I I are the dissolution
rate constants (g/(m2 ·s)) for the Phase I and Phase II regions, respectively.
For each material, the activation energy function ( E A ( pH, σ)) of the pH and stress has to be
established experimentally for both the short-term and long-term stages of glass degradation.
The total material loss and the release of Si at various environmental conditions were modelled
using the DCZOK model Equation (21) and the rate constants were obtained and reported.

4. Results
From the measured cumulative concentration data of the ions released during the dissolution of
glass, the dissolution rate constants were obtained using non-linear regression for Si (K0I Si and K0I SiI )
and for the total mass loss (K0I total and K0I total
I ). The Generalized Reduced Gradient (GRG) non-linear

regression method was used. It involved minimization of the sum of the squares of the differences
between the modelled values and the experimental values. The obtained parameters are systematized
in Tables 1–3 for the effect of the pH, temperature, and stress, respectively. The values obtained are
comparable with the dissolution rates of other glass fibres, reported in the literature [12,16,41].
In Figure 4, and further also for the glass dissolution kinetics at various stress levels and
temperatures, the points represent the experimental data, whereas the lines represent the DCZOK
modelled curves using the rate constants reported in Tables 1–3, respectively. In figures of the glass
dissolution kinetics at various environmental conditions, the glass mass loss is reported as normalized
per the initial surface area of the fibres (S0 ), as is common for surface reaction kinetics [12,16,42].
The effect of sizing on the glass dissolution (ξ sizing ) for the studied R-glass was 0.165 [16].

4.1. Effect of pH on the Glass Dissolution Kinetics

The dissolution rate constants for the glass dissolution from fibres in water at various pH levels at
60 ± 1 ◦ C are shown in Table 1. This steady-state was achieved after about a week.

Table 1. The glass dissolution rate constants obtained via the regression of the experimental data of the
R-glass fibre bundles using the Dissolving Cylinder Zero-Order Kinetic (DCZOK) model for Si and the
total mass loss at 60 ◦ C and various pH levels. KI0Si , KII I II
0Si , K0total and K0total are dissolution rate constants
for Phase I and II for Si and total glass dissolution, as indicated by indices.

pH KI0Si (g/(m2 ·s)) 0Si (g/(m ·s))

KII 2 KI0total (g/(m2 ·s)) 0total (g/(m ·s))
KII 2
Reference
1.679 ± 0.010 (3.10 ± 0.44)·10−7 (1.25 ± 0.09)·10−7 (1.70 ± 0.19)·10−6 (1.16 ± 0.08)·10−6 This work
4.005 ± 0.010 (2.59 ± 0.33)·10−8 (1.70 ± 0.11)·10−8 (8.48 ± 1.21)·10−8 (6.24 ± 0.36)·10−8 This work
5.650 ± 0.010 (6.67 ± 1.03)·10−9 (2.30 ± 0.16)·10−9 (1.82 ± 0.29)·10−8 (4.05 ± 0.29)·10−9 This work, [12,16]
7.000 ± 0.010 (3.64 ± 0.53)·10−8 (2.55 ± 0.19)·10−8 (5.46 ± 0.82)·10−8 (4.85 ± 0.38)·10−8 This work
10.012 ± 0.010 (8.97 ± 1.27)·10−8 (4.56 ± 0.32)·10−8 (1.39 ± 0.16)·10−7 (1.11 ± 0.07)·10−7 This work

The experimental glass dissolution data and the modelled DCZOK dissolution curves for various
pH levels are shown in Figure 4.
It is possible to obtain a pH influence on the activation energies of dissolution by rearranging
Equation (19) into the following form (Equation (22)):

E A = RT (lnA − lnK0 ) (22)

I I = f ( pH ) is shown in Figure 5. It is
A pH function of the activation energy of dissolution E A
fine to approximate the kinetic pH dependency as a polynomial function, being a parabolic function in
this case [43–46].
Fibers 2019, 7, 22 10 of 18

Figure 4. The glass dissolution kinetics at various pH levels: The experimental data and modelled
Dissolving Cylinder Zero-Order Kinetic (DCZOK) dissolution curves for the R-glass fibre bundles.
GF: Glass fibre.

Figure 5. The activation energy (yellow) and the rate constants (red) of the steady-state R-glass
dissolution as a function of the pH. Determination coefficients R2 are shown. 1.E-10 stands for
“1 × 10−10 ”.

Both the glass and the Si dissolution rate constant trends are similar, with an exception at low
pH levels, i.e., at a pH of 1, the rates of Si dissolution increase more dramatically than the total glass
dissolution rates, indicating a higher contribution of Si to the material loss during the dissolution in
strongly acidic environments. Thus, GFRPs should not be used in strongly acidic conditions. This is in
agreement with an observation in another study, stating that many GFRPs fail catastrophically after
a critical time when exposed to acids [1].

4.2. The Effect of Temperature on Glass Dissolution Kinetics

The dissolution rate constants for glass dissolution from the fibres in water at various temperatures
at a pH of 5.650 ± 0.010 are shown in Table 2. The steady-state was achieved after about a week.
Fibers 2019, 7, 22 11 of 18

Table 2. The glass dissolution rate constants obtained via the regression of the experimental data using
the Dissolving Cylinder Zero-Order Kinetic (DCZOK) model for Si and the total mass loss at a pH of
5.65 and at various temperatures. KI0Si , KII I II
0Si , K0total and K0total are dissolution rate constants for Phase I
and II for Si and total glass dissolution, as indicated by indices.

Temperature (◦ C) KI0Si (g/(m2 ·s)) 0Si (g/(m ·s))

KII 2 KI0total (g/(m2 ·s)) 0total (g/(m ·s))
KII 2
Reference
25 ± 1 (1.46 ± 0.23)·10−9 (2.60 ± 0.18)·10−10 (1.04 ± 0.12)·10−8 (1.42 ± 0.11)·10−9 This work
40 ± 1 (2.62 ± 0.37)·10−9 (1.08 ± 0.08)·10−9 (1.37 ± 0.19)·10−8 (2.72 ± 0.19)·10−9 This work
60 ± 1 (6.67 ± 1.03)·10−9 (2.30 ± 0.16)·10−9 (1.82 ± 0.29)·10−8 (4.05 ± 0.29)·10−9 This work, [12,16]
80 ± 1 (2.19 ± 0.31)·10−8 (8.91 ± 0.73)·10−9 (4.24 ± 0.59)·10−8 (1.47 ± 0.11)·10−8 This work

The Arrhenius approach was used to obtain the activation energy of the steady-state Si and glass
dissolution. The activation energy was obtained through graphing at a constant pH and σ, similar to
what was done in another study, using Equation (23) [23]:

EA 1
lnK0 = − + lnA (23)
R T

The graphing approach is shown in Figure 6. The obtained activation energy ( E A ) of Si

dissolution is 53.46 kJ/mol (using K0I SiI values). The pre-exponential factor ( A) for Si dissolution
is 6.82·10−1 g/(m2 ·s). The obtained activation energy ( E A ) of the total glass dissolution is 34.84 kJ/mol
(using K0I total
I values). The pre-exponential factor for the glass dissolution ( A) is 1.67·10−3 g/(m2 ·s).
The obtained values are consistent with the values reported in the literature, being slightly lower
than for E-glass (58–79 kJ/mol in alkaline solutions [23]). The activation energy for the Si dissolution
of various silica-based materials may be around 60 kJ/mol [47].

Figure 6. The graphing approach used to obtain the pre-exponential factors, the activation energies
of Si, and the total glass dissolution of the fibres at a pH of 5.65 and under no stress. GF: Glass fibre.
1/T stands for reciprocal temperature; R2 is the determination coefficient.

The experimental glass dissolution data and the modelled DCZOK dissolution curves for various
temperatures are shown in Figure 7.
Fibers 2019, 7, 22 12 of 18

Figure 7. The glass dissolution kinetics at various temperatures: The experimental data and the
modelled Dissolving Cylinder Zero-Order Kinetic (DCZOK) dissolution curves for R-glass fibre bundles.
GF: Glass fibre.

4.3. The Effect of Stress Corrosion on Glass Dissolution Kinetics

The dissolution rate constants for glass dissolution from fibres in water loaded with various
weights (0.03, 5, 8, and 10 kg) at a pH of 5.650 ± 0.010 and at 60 ± 1 ◦ C are shown in Table 3. The stress
was calculated as half of the weight divided by the cross-sectional area of the glass fibre bundles
(0.93 mm2 ). The division by two is due to the fact that the weight is carried by a bundle which goes
from each side of the rig, meaning that the total area is that of two bundles. In order to visualize the
setup and the fibre arrangement, the experimental rig in action is shown in Figure 3. The steady-state
was achieved after about 5 days (120 h), sooner than in all of the unstressed cases, indicating that stress
may slightly accelerate the transition towards the steady-state dissolution.

Table 3. The glass dissolution rate constants obtained via the regression of the experimental data using
the Dissolving Cylinder Zero-Order Kinetic (DCZOK) model for Si and the total mass loss at 60 ◦ C and
at a pH of 5.65 at various stress levels. σ stands for stress; KI0Si , KII I II
0Si , K0total and K0total are dissolution
rate constants for Phase I and II for Si and total glass dissolution, as indicated by indices.

Weight (kg) σ KI0Si (g/(m2 ·s)) 0Si (g/(m ·s))

KII 2 KI0total (g/(m2 ·s)) 0total (g/(m ·s))
KII 2
Reference
0 0.0 (6.67 ± 1.03)·10-9 (2.30 ± 0.16)·10−9 (1.82 ± 0.29)·10−8 (4.05 ± 0.29)·10−9 This work, [12]
0.03 0.2 (7.27 ± 1.15)·10−9 (2.38 ± 0.21)·10−9 (2.08 ± 0.42)·10−8 (4.12 ± 0.33)·10−9 This work
5 26.4 (7.45 ± 1.03)·10−9 (2.58 ± 0.19)·10−9 (2.04 ± 0.33)·10−8 (4.35 ± 0.35)·10−9 This work
8 42.2 (8.30 ± 1.33)·10−9 (3.47 ± 0.25)·10−9 (2.01 ± 0.41)·10−8 (5.56 ± 0.44)·10−9 This work
10 52.7 (9.21 ± 1.33)·10−9 (4.73 ± 0.33)·10−9 (1.65 ± 0.44)·10−8 (8.12 ± 0.67)·10−9 This work

The experimental glass dissolution data and the modelled DCZOK dissolution curves for various
stress levels are shown in Figure 8.
Fibers 2019, 7, 22 13 of 18

Figure 8. The glass dissolution kinetics at various stress levels: The experimental data and the
modelled Dissolving Cylinder Zero-Order Kinetic (DCZOK) dissolution curves for glass fibre bundles.
GF: Glass fibre.

I I = f ( σ ) for R-glass fibre bundles is

A stress function of the activation energy of dissolution E A
shown in Figure 9. An increase in the stress seems to accelerate glass dissolution, showing a similar
trend for both Si and the total mass loss kinetics. The increase in the stress reduces the activation
energy of dissolution linearly and accelerates the glass dissolution rates exponentially. Additional
studies involving higher stresses are recommended to establish this influence in more detail.

Figure 9. The activation energy of dissolution (yellow) and the dissolution rate constants (red),
functions of the stress (σ) for the glass fibre bundles at the steady-state. Determination coefficients R2
are shown. 4E-9 stands for “4 × 10−9 ”.

5. Discussion
R-glass fibres were studied in this work. However, there should be no limitations to applying
the DCZOK model to other types of glass fibres. Validation of the model with various types of
glass fibres, such as E, ECR, and S, is advised. The processes for all of these fibres are expected to
follow similar trends, whereas only the values of the terms in the model have to be assessed for each
material individually.
Fibers 2019, 7, 22 14 of 18

The industry is interested in the pH, temperature, and stress corrosion for environmental effects.
Temperatures are also interesting to the industry for accelerated testing purposes. As the industry is
concerned with lowering the testing time for the fibre-dominated property deterioration in GFRPs,
the model allows the prediction of the loss of glass material. While it takes time to obtain the parameters,
the most significant time saving comes from using the kinetic constants and the model.
The influence of the pH, temperature, and stress corrosion on the rate constants and dissolution
activation energies was obtained, however, the influence of each parameter was studied once at a time.
Thus, a suggestion for future work includes a cross-parametric study of the pH, temperature, and
stress corrosion, in order to deduce a general analytical solution for the environmental influence on
the activation energy of dissolution and to study whether there is a coupled effect. The DCZOK model
should be extendable to include the effect of the ionic strength and, finally, to be further used in linking
the reduction in the mechanical strength of the fibres with the dissolution kinetics.
The authors think that validating the DCZOK model for seawater conditions (about 1.84–12.62 mg
SiO2 /kg water; pH of seawater at 7.8) would be highly beneficial, especially for the marine and offshore
industries, since the real-life structures most often operate in the seawater environment. When GFs
are used in seawater, the dissolution of glass occurs at a slower rate, due to the presence of silica in
the seawater, which is in seawater from the contact with sand and minerals [48–51]. The approach in
distilled water is conservative with regards to seawater, meaning that structures implementing glass
fibres designed for distilled water conditions should not encounter penalties to their service time in
the seawater.
The model should be applicable to other types of glass as well, as SiO2 is the major component
in virtually all types of glass [23], but it would be beneficial to validate this model experimentally
with other types of glass fibres. Information on the exact composition of the studied R-glass was
not available from the manufacturer, supplier, or technical data sheets. However, it may be possible
to speculate that the composition of the dissolving ions during steady-state Phase II is somewhat
representative of the glass composition, since all ions follow the slowest dissolving element, as was
reported elsewhere for the same material [12]. The ion composition during the steady-state was the
following: 2.7 wt% Na, 1.4 wt% K, 14.3 wt% Ca, 15.1 wt% Mg, 0.1 wt% Fe, 6.9 wt% Al, 56.1 wt% Si,
and 3.4 wt% Cl [12].
Another aspect that should be investigated is the effect of the sizing on the glass fibre dissolution,
i.e., its chemical composition, amount, and distribution on the fibres. For the studied R-glass,
it is known that the sizing is epoxy-compatible and its loss on ignition (LOI) value is 0.64 wt%.
The protective effect of sizing for the studied R-glass fibres was reported in another study [16].
It was previously observed by Scheffler et al. [52] that strong alkaline solutions can significantly
damage glass fibres due to the chemical attack [52]. This observation is consistent with the results
in this work, indicating much higher dissolution rates at a pH of 10 (at least an order of magnitude
higher) compared to the neutral conditions.

6. Conclusions
The analytical model, termed the Dissolving Cylinder Zero-Order Kinetic model, was successfully
used to explain the long-term glass dissolution experiments of R-glass fibres with various
environmental conditions:
 2 
∂m (K0 ξ sizing )
∂t = 2nπl r 0 K 0 ξ sizing − ρ glass t
⎛  2 ⎞
E ( pH,σ)
− A RT
⎜ E ( pH,σ)
Ae ξ sizing
⎟ (21)
= 2nπl ⎜⎝r0 Ae
− A RT ξ sizing −
glass ρ t⎟
⎠
Fibers 2019, 7, 22 15 of 18

The model accounts for the influence of the pH, temperature, and stress, as well as the effects
of sizing protection (ξ sizing for the studied R-glass is 0.165). The glass dissolution rate constants
were obtained and reported for various pH levels and temperatures, as well as for various stress
corrosion conditions.
The temperature showed an Arrhenius-type influence on the kinetics, increasing the rate of
dissolution exponentially with an increasing temperature. The activation energy of the steady-state
glass dissolution was obtained and reported at a pH of 5.65 and with no stress (53.46 and 34.84 kJ/mol
for Si and the total glass dissolution, respectively). The activation energy of dissolution is affected by
the pH and stress. The activation energy decreases linearly as the stress increases. The influence of
the pH is more complicated and may be described by a parabolic polynomial function; the activation
energy peaks at a pH of 5.65 and decreases towards more basic, as well as more acidic, conditions.
In comparison with neutral conditions, basic and acidic aqueous environments showed an increase
in the dissolution rates, affecting the lifetime of the glass fibres negatively.
External loads also increased the glass dissolution rates due to stress corrosion. The stress seems
to accelerate the glass dissolution rates exponentially. The model was able to capture all of these effects.

Author Contributions: Conceptualization, A.E.K. and A.T.E.; methodology, A.E.K. and A.T.E.; software, A.E.K.;
validation, A.E.K. and I.K.; formal analysis, A.E.K. and I.K.; investigation, A.E.K.; resources, A.E.K., K.V., M.K.,
and A.T.E.; data curation, A.E.K.; writing—original draft preparation, A.E.K. and A.I.G.; writing—review and
editing, A.E.K., M.K., and A.T.E.; visualization, A.E.K.; supervision, A.T.E.; project administration, A.T.E.; funding
acquisition, A.T.E.
Funding: This research was funded by The Research Council of Norway (Project 245606/E30 in the Petromaks
2 programme).
Acknowledgments: This work is part of the DNV GL-led Joint Industry Project “Affordable Composites” with 19
industrial partners and the Norwegian University of Science and Technology (NTNU). The authors would like
to express their thanks for the financial support from The Research Council of Norway (Project 245606/E30 in
the Petromaks 2 programme). The authors are thankful to Børge Holen, Carl-Magnus Midtbø, Cristian Torres
Rodriguez, Melanie Shebel, Syverin Lierhagen, and Konstantı̄ns Viligurs. Andrey is especially grateful to Oksana
V. Golubova.
Conflicts of Interest: The authors declare no conflicts of interest.

Abbreviations
r Fibre radius (m)
r0 Initial fibre radius (m)
K0 Dissolution rate constant (g/(m2 ·s))
KI0Si Dissolution rate constant for Si only (non-steady-state) (g/(m2 ·s))
KII
0Si Dissolution rate constant for Si only (steady-state) (g/(m2 ·s))
KI0total Dissolution rate constant for glass (non-steady state) (g/(m2 ·s))
KII
0total Dissolution rate constant for glass (steady-state) (g/(m2 ·s))
tst Time when long-term steady state is reached (s)
n Number of fibres (-)
l Length of fibres (m)
S Glass fibre surface area (m2 )
S0 Initial glass fibre surface area (m2 )
pH Acidity of the environment (-)
A Pre-exponential factor (g/(m2 ·s))
R Universal gas constant (8.314 J/(mol·K))
T Absolute temperature (K)
EA Dissolution activation energy (J/mol)
ρglass Density of the glass (g/m3 )
t Time (s)
σ Stress (MPa)
m; mdissolved Total mass loss due to dissolution (g)
Fibers 2019, 7, 22 16 of 18

mSi Si mass loss due to dissolution (g)

ξ sizing Protective effect of sizing against glass dissolution (-)
c Non-cumulative ion mass concentration measured with HR-ICP-MS (g/L)
Vwater Volume of a water sample used for the HR-ICP-MS (L)
GF Glass Fibre
GFRP Glass Fibre-Reinforced Polymer; same as Glass Fibre-Reinforced Composite
DCZOK Dissolving Cylinder Zero-Order Kinetic (model)
E − glass “Electrical” glass
ECR − glass “Electrical/Chemical Resistance” glass
R − glass “Reinforcement” glass
S − glass “Strength” glass
FRP Fibre-Reinforced Polymer, same as fibre-reinforced composite
HR-ICP-MS High Resolution Inductively Coupled Plasma Mass Spectrometry

Appendix A. The Schematics of the Stress Corrosion Rig.

Figure A1. The design schematics of the stress corrosion rig for the glass fibre bundles, indicating the
placement of the fibres and the weights.

References
1. Agarwal, B.D.; Broutman, L.J. Analysis and Performance of Fibre Composites, 2nd ed.; John Wiley and Sons, Inc.:
Hoboken, NJ, USA, 1990; pp. 339–359, ISBN 978-0-471-51152-6.
2. Wallenberger, F.T. Commercial and Experimental Glass Fibres. In Fibreglass and Glass Technology;
Wallenberger, F.T., Bingham, P.A., Eds.; Springer: New York, NY, USA, 2010; ISBN 978-1-4419-0735-6.
3. Steinmann, W.; Saelhoff, A.-K. Essential Properties of Fibres for Composite Applications. In Fibrous and
Textile Materials for Composite Applications, Textile Science and Clothing Technology; Rana, S., Fangueiro, R., Eds.;
Springer: Singapore, 2016; ISBN 978-981-10-0232-8.
4. Krauklis, A.E.; Echtermeyer, A.T. Mechanism of Yellowing: Carbonyl Formation during Hygrothermal
Aging in a Common Amine Epoxy. Polymers 2018, 10, 1017. [CrossRef]
5. Xiao, G.Z.; Shanahan, M.E.R. Swelling of DGEBA/DDA epoxy resin during hygrothermal ageing. Polymer
1998, 39, 3253–3260. [CrossRef]
6. Toscano, A.; Pitarresi, G.; Scafidi, M.; Di Filippo, M.; Spadaro, G.; Alessi, S. Water diffusion and swelling
stresses in highly crosslinked epoxy matrices. Polym. Degrad. Stab. 2016, 133, 255–263. [CrossRef]
7. Grabovac, I.; Whittaker, D. Application of bonded composites in the repair of ships structures—A 15-year
service experience. Compos. Part A 2009, 40, 1381–1398. [CrossRef]
Fibers 2019, 7, 22 17 of 18

8. McGeorge, D.; Echtermeyer, A.T.; Leong, K.H.; Melve, B.; Robinson, M.; Fischer, K.P. Repair of floating
offshore units using bonded fibre composite materials. Compos. Part A 2009, 40, 1364–1380. [CrossRef]
9. Gustafson, C.-G.; Echtermeyer, A. Long-term properties of carbon fibre composite tethers. Int. J. Fatigue 2006,
28, 1353–1362. [CrossRef]
10. Salama, M.M.; Stjern, G.; Storhaug, T.; Spencer, B.; Echtermeyer, A. The First Offshore Field Installation for
a Composite Riser Joint; OTC-14018-MS; Offshore Technology Conference: Houston, TX, USA, 2002.
11. Echtermeyer, A.T.; Gagani, A.I.; Krauklis, A.E.; Mazan, T. Multiscale Modelling of Environmental
Degradation—First Steps. In Durability of Composites in a Marine Environment 2. Solid Mechanics and Its
Applications; Davies, P., Rajapakse, Y.D.S., Eds.; Springer: Cham, Switzerland, 2018; Volume 245, pp. 135–149,
ISBN 978-3-319-65145-3.
12. Krauklis, A.E.; Echtermeyer, A.T. Long-Term Dissolution of Glass Fibres in Water Described by Dissolving
Cylinder Zero-Order Kinetic Model: Mass Loss and Radius Reduction. Open Chem. 2018, 16, 1189–1199.
[CrossRef]
13. Tournié, A.; Ricciardi, P.; Colomban, Ph. Glass Corrosion Mechanisms: A Multiscale Analysis. Solid State
Ionics 2008, 179, 2142–2154. [CrossRef]
14. Brown, E.N.; Davis, A.K.; Jonnalagadda, K.D.; Sottos, N.R. Effect of surface treatment on the hydrolytic
stability of E-glass fibre bundle tensile strength. Compos. Sci. Technol. 2005, 65, 129–136. [CrossRef]
15. Bledzki, A.; Spaude, R.; Ehrenstein, G.W. Corrosion Phenomena in Glass Fibres and Glass Fibre Reinforced
Thermosetting Resins. Compos. Sci. Technol. 1985, 23, 263–285. [CrossRef]
16. Krauklis, A.E.; Echtermeyer, A.T. Dissolving Cylinder Zero-Order Kinetic Model for Predicting Hygrothermal
Aging of Glass Fibre Bundles and Fibre-Reinforced Composites. In 4th International Glass Fibre Symposium;
Gries, T., Pico, D., Lüking, A., Becker, T., Eds.; Mainz, G (Verlag): Aachen, Germany, 2018; pp. 66–72,
ISBN 978-3-95886-249-4.
17. Grambow, B.; Müller, R. First-order dissolution rate law and the role of surface layers in glass performance
assessment. J. Nucl. Mater. 2001, 298, 112–124. [CrossRef]
18. Grambow, B. A General Rate Equation for Nuclear Waste Glass Corrosion. MRS Symp. Proc. Libr. Arch. 1985,
44, 15–27. [CrossRef]
19. Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Near-Infrared Spectroscopic Method for Monitoring Water
Content in Epoxy Resins and Fibre-Reinforced Composites. Materials 2018, 11, 586. [CrossRef]
20. Stamenović, M.R.; Putić, S.S.; Rakin, M.B.; Medjo, B.; Čikara, D. Effect of alkaline and acidic solutions on the
tensile properties of glass–polyester pipes. Mater. Des. 2011, 32, 2456–2461. [CrossRef]
21. Amaro, A.M.; Reis, P.N.B.; Neto, M.A.; Louro, C. Effects of alkaline and acid solutions on glass/epoxy
composites. Polym. Degrad. Stab. 2013, 98, 853–862. [CrossRef]
22. Mišíková, L.; Liška, M.; Galusková, D. Corrosion of e-glass fibres in distilled water. Ceram. Silik. 2007, 51,
131–135.
23. Bashir, S.T.; Yang, L.; Liggat, J.J.; Thomason, J.L. Kinetics of dissolution of glass fibre in hot alkaline solution.
J. Mater. Sci. 2018, 53, 1710–1722. [CrossRef]
24. Hunter, F.M.I.; Hoch, A.R.; Heath, T.G.; Baston, G.M.N. Report RWM005105, AMEC/103498/02 Issue 2: Review
of Glass Dissolution Models and Application to UK Glasses; AMEC: Didcot, Oxfordshire, UK, 2015.
25. Li, H.; Gu, P.; Watson, J.; Meng, J. Acid corrosion resistance and mechanism of E-glass fibres: Boron factor.
J. Mater. Sci. 2013, 48, 3075–3087. [CrossRef]
26. Icenhower, J.; Steefel, C.I. Dissolution Rate of Borosilicate Glass SON68: A Method of Quantification Based
upon Interferometry and Implications for Experimental and Natural Weathering Rates of Glass. Geochim.
Cosmochim. Acta 2015, 157, 147–163. [CrossRef]
27. Gy, R. Stress corrosion of silicate glass: A review. J. Non-Cryst. Solids 2003, 316, 1–11. [CrossRef]
28. Ciccotti, M. Stress-corrosion mechanisms in silicate glasses. J. Phys. D Appl. Phys. 2009, 42, 214006–214039.
[CrossRef]
29. Michalske, T.A.; Freiman, S.W. A Molecular Mechanism for Stress Corrosion in Vitreous Silica. J. Am. Ceram.
Soc. 1983, 66, 284–288. [CrossRef]
30. Michalske, T.A.; Bunker, B.C. Steric Effects in Stress Corrosion Fracture of Glass. J. Am. Ceram. Soc. 1987, 70,
780–784. [CrossRef]
31. Michalske, T.A.; Bunker, B.C. A Chemical Kinetics Model for Glass Fracture. J. Am. Ceram. Soc. 1993, 76,
2613–2618. [CrossRef]
Fibers 2019, 7, 22 18 of 18

32. Michalske, T.A.; Bunker, B.C. Slow fracture model based on strained silicate structures. J. Appl. Phys. 1984,
56, 2686–2693. [CrossRef]
33. Schmitz, G.K.; Metcalfe, A.G. Stress Corrosion of E-Glass Fibres. Ind. Eng. Chem. Prod. Res. Dev. 1966, 5, 1–8.
[CrossRef]
34. Delage, F.; Ghaleb, D.; Dussossoy, J.L.; Chevallier, O.; Vernaz, E. A mechanistic model for understanding
nuclear waste glass dissolution. J. Nucl. Mater. 1992, 190, 191–197. [CrossRef]
35. Ma, T.; Jivkov, A.P.; Li, W.; Liang, W.; Wang, Y.; Xu, H.; Han, X. A mechanistic model for long-term nuclear
waste glass dissolution integrating chemical affinity and interfacial diffusion barrier. J. Nucl. Mater. 2017,
486, 70–85. [CrossRef]
36. Khawam, A.; Flanagan, D.R. Solid-State Kinetic Models: Basics and Mathematical Fundamentals. J. Phys.
Chem. B 2006, 110, 17315–17328. [CrossRef]
37. International Standard ISO 2078:1993 (revised in 2014), Textile Glass—Yarns—Designation. 2014. Available
online: https://www.iso.org/standard/6865.html (accessed on 11 February 2019).
38. 3B Fibreglass technical data sheet. HiPer-tex W2020 rovings, Belgium. 2012. Available online: https:
//www.3b-fibreglass.com/ (accessed on 24 July 2018).
39. Thomason, J.L. Glass Fiber Sizings: A Review of the Scientific Literature; J.L. Thomason: Middletown, DE, USA,
2012; p. 127, ISBN 978-0-9573814-1-4.
40. Shiue, Y.S.; Matthewson, M.J. Stress dependent activation entropy for dynamic fatigue of pristine silica
optical fibres. J. Appl. Phys. 2001, 89, 4787–4793. [CrossRef]
41. Eastes, W.; Potter, R.M.; Hadley, J.G. Estimating in-vitro glass fibre dissolution rate from composition.
Inhal. Toxicol. 2000, 12, 269–280. [CrossRef] [PubMed]
42. Papadimitriou, S.; Kennedy, H.; Kennedy, P.; Thomas, D.N. Kinetics of ikaite precipitation and dissolution in
seawater-derived brines at sub-zero temperatures to 265 K. Geochim. Cosmochim. Acta 2014, 140, 199–211.
[CrossRef]
43. Ozola, R.; Krauklis, A.E.; Leitietis, M.; Burlakovs, J.; Vircava, I.; Ansone-Bertina, L.; Bhatnagar, A.; Klavins, M.
FeOOH-modified clay sorbents for arsenic removal from aqueous solutions. Environ. Technol. Innov. 2016.
[CrossRef]
44. Krauklis, A.E.; Ozola, R.; Burlakovs, J.; Rugele, K.; Kirillov, K.; Trubaca-Boginska, A.; Rubenis, K.;
Stepanova, V.; Klavins, M. FeOOH and Mn8 O10 Cl3 Modified Zeolites for As(V) Removal in Aqueous
Medium. J. Chem. Technol. Biotechnol. 2017, 92, 1948–1960. [CrossRef]
45. Anxolabéhère-Mallart, E.; Costentin, C.; Policar, C.; Robert, M.; Savéant, J.M.; Teillout, A.L. Proton-coupled
electron transfers in biomimetic water bound metal complexes. The electrochemical approach. Faraday
Discuss. 2011, 148, 83–95. [CrossRef]
46. Secula, M.S.; Creţescu, I.; Petrescu, S. An experimental study of indigo carmine removal from aqueous
solution by electrocoagulation. Desalination 2011, 277, 227–235. [CrossRef]
47. Criscenti, L.J.; Kubicki, J.D.; Brantley, S.L. Silicate Glass and Mineral Dissolution: Calculated Reaction Paths
and Activation Energies for Hydrolysis of a Q3 Si by H3 O+ Using Ab Initio Methods. J. Phys. Chem. A 2006,
110, 198–206. [CrossRef]
48. Fournier, R.O.; Rowe, J.J. The solubility of amorphous silica in water at high temperatures and high pressures.
Am. Miner. 1977, 62, 1052–1056.
49. Crundwell, F.K. On the Mechanism of the Dissolution of Quartz and Silica in Aqueous Solutions. ACS Omega
2017, 2, 1116–1127. [CrossRef]
50. Von Damm, K.L.; Edmond, J.M.; Grant, B.; Measures, C.I.; Walden, B.; Weiss, R.F. Chemistry of submarine
hydrothermal solutions at 21◦ N, East Pacific Rise. Geochim. Cosmochim. Acta 1985, 49, 2197–2220. [CrossRef]
51. Holland, H.D. The Chemistry of the Atmosphere and Oceans; Wiley: New York, NY, USA, 1978; p. 351,
ISBN 978-0471035091.
52. Scheffler, C.; Förster, T.; Mäder, E.; Heinrich, G.; Hempel, S.; Mechtcherine, V. Aging of alkali-resistant glass
and basalt fibers in alkaline solutions: Evaluation of the failure stress by Weibull distribution function.
J. Non-Cryst. Solids 2009, 355, 2588–2595. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
APPENDIX H

PAPER VIII

KRAUKLIS A.E., GAGANI A.I., ECHTERMEYER A.T.

LONG-TERM HYDROLYTIC DEGRADATION OF THE SIZING-RICH COMPOSITE

INTERPHASE.

COATINGS (SWITZERLAND), 9(4), 2019, 263-286.

DOI:10.3390/COATINGS9040263

PAPER VIII

171
172
 coatings
Article
Long-Term Hydrolytic Degradation of the Sizing-Rich
Composite Interphase
Andrey E. Krauklis * , Abedin I. Gagani and Andreas T. Echtermeyer
Department of Mechanical and Industrial Engineering, Norwegian University of Science and Technology,
7491 Trondheim, Norway; [email protected] (A.I.G.); [email protected] (A.T.E.)
* Correspondence: [email protected] or [email protected]; Tel.: +371-268-10-288

Received: 1 April 2019; Accepted: 17 April 2019; Published: 19 April 2019 

Abstract: Glass fiber-reinforced composites are exposed to hydrolytic degradation in subsea and
offshore applications. Fiber-matrix interphase degradation was observed after the matrix was fully
saturated with water and typical water absorption tests according to ASTM D5229 were stopped.
Due to water-induced dissolution, fiber-matrix interphase flaws were formed, which then lead to
increased water uptake. Cutting sample plates from a larger laminate, where the fibers were running
parallel to the 1.5 mm long short edge, allowed the hydrolytic degradation process to be studied.
The analysis is based on a full mechanistic mass balance approach considering all the composite’s
constituents: water uptake and leaching of the matrix, dissolution of the glass fibers, and dissolution
of the composite interphase. These processes were modeled using a combination of Fickian diffusion
and zero-order kinetics. For the composite laminate studied here with a saturated epoxy matrix,
the fiber matrix interphase is predicted to be fully degraded after 22 to 30 years.

Keywords: composites; sizing; interphase; glass fibers; environmental degradation; aging; model;
kinetics; durability; hydrolysis

1. Introduction
Fiber-reinforced polymer (FRP) composites have experienced a rapid rise in use in the past 50 years
due to their high strength, stiffness, relatively light weight and good corrosion resistance, especially
when compared with more traditional structural materials such as steel and aluminum [1]. The reason
for such superior performance is the synergistic interaction between the constituent materials inside
the composite [1]. One such material is the sizing, which is a multi-component coating on the surface
of the fibers. During the manufacture of FRPs, this results in the formation of a sizing-rich composite
interphase between the reinforcing fibers and the matrix polymer [2]. This composite interphase
is of vital importance since the mechanical properties of composite materials are often determined
by whether the mechanical stresses can be efficiently transferred from the matrix to the reinforcing
fibers [3–5]. The quality of the interfacial interaction is strongly dependent on the adhesional contact
and the presence of flaws in the interphase [6]. It is generally agreed that the composite interphase is
often the mechanical weak link and a potential source for the initiation of defects in fiber-reinforced
composite structures [5].
Composite laminates are often exposed to aqueous and humid environments. Environmental aging
is especially interesting for marine, offshore and deep-water applications of composites, such as oil
risers and tethers [7–12]. It has been reported that water and humid environments negatively impact
the mechanical properties of FRPs partially because of a loss of the interfacial bonding [5,12–15].
Flaws in the interphase can be introduced due to the interaction of the interphase with water taken up
from the environment [6]. The removal of the sizing material can also lead to a microcrack initiation at

Coatings 2019, 9, 263; doi:10.3390/coatings9040263 www.mdpi.com/journal/coatings

Coatings 2019, 9, 263 2 of 24

the surface of glass fibers. Furthermore, various sizing components can be extracted by water, resulting
in the loss of the material [16–20]. Quantifying the water-induced aging is especially important for
glass fiber-reinforced composites since the glass fibers are highly hygroscopic [5]. The environmental
durability is one of the limiting factors in the structural applications [21], since the superior strength and
stiffness of such materials are often compromised by the uncertainty of the material’s interaction with
the environment [22]. Durability is a primary issue because environmental factors such as moisture,
temperature and the state of stress to which the material is exposed can degrade interfacial adhesion
as well as the properties of the constituent phases. Environmental aging is mainly important at high
temperatures, since the dissolution reactions are accelerated at higher temperatures. Therefore, it is
of great importance to understand the environmental aging and dissolution kinetics of a sizing-rich
composite interphase.

1.1. Sizing and its Composition

The sizing which forms the interphase, has typically a proprietary composition.
Available information about commercial glass fibers tends to contain only one or two sizing-related
details. The first is an indication of the chemical compatibility of the sizing with the matrix polymer,
e.g., epoxy, as in this case. The second is a value for the loss on ignition (LOI), which indicates the
amount of sizing [23]. The key functions of the sizing are: (1) to protect the glass fibers during handling
and production; (2) to ensure a high level of stress transfer capability across the fiber-matrix interphase;
and (3) to protect the composite matrix interphase against environmental degradation [12].
A typical sizing consists of about 20 chemicals. The most important chemical is an organofunctional
silane commonly referred to as a coupling agent [24–26], which is the main component that
promotes adhesion and stress-transfer between the polymer matrix and the fiber [12]. It also
provides improvements in the interphase strength and hygrothermal resistance of the composite
interphase [26–28]. The silane coupling agents have the general structure [X–Si(–O–R)3 ] where R is a
methyl or ethyl group and X is a reactive group towards the polymer, in this case, an amine group.
When applied to fibers, a silane coupling agent is first hydrolyzed to a silanol in presence of water.
It is unstable and further condenses onto the fibers by producing a siloxane/poly(siloxane) network,
which then partially becomes covalently bonded to the glass fiber surface. During the composite
manufacture, the X reactive groups of the silane may react with the thermosetting matrix polymer,
leading to a strong network bridging between the fiber and the matrix [12].
Although there are many different silane molecules available, the aminosilanes form the largest
proportion of silanes employed in the composites industry [12]. The most common coupling agent is
an aminosilane compound called γ-aminopropyltriethoxysilane (γ-APS), also known as APTES, which
is the coupling agent in the studied sizing [16]. Usually sizings contain about 10 wt % of the coupling
agent [29].
The composition of the sizing also consists of a number of multi-purpose components, such as a
film former which holds the filaments together in a strand and protects the filaments from damage
through fiber–fiber contact. Film formers are as closely compatible to the polymer matrix as possible.
Epoxies, such as in this case, are very common film formers [12]. Usually sizings contain about
70–80 wt % of the film former [12,29].
Much less is known about the other chemicals in the sizing [12]. The sizing may also contain
other compounds such as cationic or non-ionic lubricants, antistatic agents, emulsifiers, chopping aids,
wetting agents or surfactants, and antioxidants [2,12,30]. Poly(propylene oxide) (PPO) or its co-polymer
with poly(ethylene oxide) (PEO) is often used as a surfactant in sizings [2]. Polydimethylsiloxane
(PDMS) is a common adhesion promoter, wetting agent, or surface tension reducer [2].
The exact composition of the sizing used in this study was not known to the authors, but based
on technical details on the given R-glass fibers elsewhere [16], it is assumed that the sizing is based on
the general characteristics described above. The results obtained are compatible with this assumption.
Coatings 2019, 9, 263 3 of 24

1.2. The Structure of the Sizing-Rich Composite Interphase

The structure of the sizing-rich composite interphase is very complex [12], as the sizing itself is
heterogeneous and not uniform [12,31,32]. Furthermore, it has been observed by various researchers,
that sizing is coated on fibers in “islands”, “islets” or in patches, meaning that the fiber surface is only
partially covered by the sizing, also giving some roughness to the surface [12,33–38]. Thomason and
Dwight have concluded that epoxy-compatible sizings cover at least 90% of the glass fiber surface [39].
Mai et al. investigated APTES sizings using atomic force microscopy (AFM) and concluded that sized
fibers are rougher than unsized fibers [38]. Also, similar conclusions were drawn by a few other
researchers, including Turrión et al., who have shown that thickness of the sizing on the glass fibers
varies from some nanometers up to a few hundred nanometers due to roughness [6,31,37].
With regards to the molecular structure of the interphase, APTES forms chemical covalent and
physico-chemical hydrogen bonds and van der Waals interactions with the glass fibers and the amine
epoxy [12,40]. The majority of APTES molecules which react with the glass surface can only form
single Si–O–Si bonds with the glass due to steric limitations, while the vast majority of Si–O–Si
bond formation in the silane interphase is due to polymerization—formation of the poly(siloxane)
network [12]. A multilayer is formed on the glass fiber surfaces where the amino groups form
intramolecular ring structures [32,41].
The concept of a composite interphase can be represented by a matrix polymer/poly(siloxane)/glass
fiber model (shown in Figure 1) [5].
The siloxanes and poly(siloxanes) form covalent bonds with the glass fiber surface, resulting in a
two-dimensional interface, the thickness of which is governed by the length of the chemical bonds,
and is of an ångstrøm-scale (one tenth of a nanometer) [5].
The composite interphase is a gradient-type blend of the sizing compounds and the bulk
matrix polymer, usually being about a micrometer in thickness [5,12,29,42,43]. It was observed, that
an interfacial failure occurs at 0.5–4 nm from the glass surface in glass/γ-APS/epoxy interphase,
indicating that the interphase region, rather than the two-dimensional interface is the weak link [5].

Figure 1. The concept of a polymer–siloxane–glass interphase, after [5]. The dotted line indicates that
the sizing is rough [6,31,38].
Coatings 2019, 9, 263 4 of 24

1.3. The Aim of This Work

Composites take up water from their surroundings and may release some molecules into the
surrounding water. Water uptake curves for composites are not straightforward to interpret since
each constituent (matrix, fibers, sizing-rich interphase) interacts differently with the absorbed water.
The mass uptake curve presents the combined effect of all these individual interactions.
Testing of water absorption is usually stopped when the composite material’s water uptake has
reached a maximum. The typical test procedures follow ASTM D5229 [44], where testing is stopped
when two subsequent measurements do not differ by more than 0.5% [44]. However, when exposure
to water is extended for longer periods, experiments performed in this work showed that degradation
of the composite continues and additional mass gain and loss processes are involved. A similar
observation was made by Perreux et al. [45], who studied immersion in water of 2.7 mm thick glass
fiber epoxy composite plates for up to 10 years. They found that the weight gain of plates aged at
60 ◦ C increased strongly after saturation. After about five years in water at 60 ◦ C, a composite plate
started to continuously lose mass with time [45]. This study will show that these effects can be related
to the hydrolytic degradation of the fiber/matrix interphase.
The aim of this manuscript is to describe the degradation of the fiber/matrix interphase with
special emphasis on the reaction kinetics.

2. Materials and Methods

2.1. Materials
Composite laminates were made with an amine-cured epoxy. The epoxy was prepared by mixing
reagents Epikote Resin RIMR135TM (Hexion, Columbus, OH, USA) and amine based Epikure Curing
Agent RIMH137TM (Hexion), stoichiometrically, in a ratio of 100:30 by weight. The mixture was
degassed in a vacuum chamber for 30 min in order to remove bubbles. The density of the polymer
(ρm ) was 1.1 g/cm3 . Resin and hardener system consisted of the following compounds by composition:
0.63 wt % bisphenol A diglycidyl ether (DGEBA), 0.14 wt % 1,6-hexanediol diglycidyl ether (HDDGE),
0.14 wt % poly(oxypropylene)diamine (POPA) and 0.09 wt % isophorondiamine (IPDA) [46].
A typical glass fiber used for marine and oil and gas applications was selected: boron-free and
fluorine-free high-strength, high-modulus 3B HiPer-TexTM W2020 R-glass (3B-the fiberglass company,
Hoeilaart, Belgium). Stitch-bonded mats were used. The average fiber diameter was 17 ± 2 μm [47,48].
The density of glass (ρf ) was 2.54 g/cm3 [47,48].
Composite laminates 50 mm thick were prepared via vacuum-assisted resin transfer molding
(VARTM). Laminates were manufactured using the aforementioned fabrics and epoxy resin. The curing
was performed at room temperature for 24 h, continued by post-curing in an air oven (Lehmkuhls
Verksteder, Oslo, Norway) at 80 ◦ C for 16 h. Full cure was achieved [46,49]. The composite laminates
were cut into specimens with dimensions of 50 mm × 50 mm × 1.5 mm. The geometry of the samples
and cross section of the fibers is shown in Figure 2. Two configurations C1 and C3 were cut, as shown
in Figure 2. Configurations. C1 is representative of a typical composite where fibers are parallel to
one of the long sides. The surface area of cut fibers with exposed cross sections is 50 mm × 1.5 mm.
Configuration C3 was cut in a way that a maximum number of cut fibers were obtained having
exposed cross sections (50 mm × 50 mm). The length of the fibers was just 1.5 mm. This unusual
specimen was made to obtain maximum fiber exposure towards the water. The same specimens were
also used to measure anisotropic diffusivity in a separate study [49]. The specified dimensions were
achieved within 5% tolerance. The thickness was adjusted using a grinding and polishing machine
Jean Wirtz PHOENIX 2000 (Jean Wirtz, Dusseldorf, Germany) and SiC discs (Struers, Cleveland, OH,
USA; FEPA P500, grain size 30 μm).
Figure 2 also shows a micrograph of a surface with visible cross sections of cut fibers from a
specimen with C3 configuration. The micrograph was taken with a confocal microscope InfiniteFocus
G4 (Alicona, Graz, Austria).
Coatings 2019, 9, 263 5 of 24

(a) (b)

Figure 2. Glass fiber-reinforced epoxy composite plates: (a) sample configuration indicating alignment
of the fibers in the plate: C3 at the top; C1 at the bottom; (b) micrograph of the largest face of the
composite plate showing the cross section of the fibers at the surface.

Distilled water (resistivity 0.5–1.0 MΩ·cm) was used for conditioning of the composite
samples. It was produced using the water purification system Aquatron A4000 (Cole-Parmer,
Vernon Hills, IL, USA). The pH of the distilled water was 5.650 ± 0.010, being lower than neutral due
to dissolved CO2 from atmosphere in equilibrium.

2.2. Experimental Methods

2.2.1. Loss on Ignition

The loss on ignition (LOI) value of the fiber bundles was determined according to the standard
practice ASTM D4963 [23]. This technique allows measurement of the weight loss of a sized glass
sample. Since the weight loss is due to the burning off of the sizing, the method can be used to
determine the amount of sizing on the fiber [12]. According to the LOI measurements, the sizing was
0.64 wt % of the sized fibers. The temperature during the LOI measurement was about 565 ◦ C applied
for about 5–5.5 h.
The obtained LOI is consistent with literature. LOI of most glass fiber reinforcement products is
below 1.2 wt % [12]. For instance, Zinck and Gerard [50] also studied an APTES-based sizing which
had a similar LOI value of 0.77 wt %.

2.2.2. Constituent Volume and Mass Fractions of the Composite

The fiber volume fraction of the composite was 59.5% and was determined using the burn-off
test, after the ASTM Standard D3171 [51]. The void volume fraction of the composite was 0.44%
and was measured by image analysis of optical microscope images, as was described elsewhere by
Gagani et al. [49]. Fiber, matrix, interphase and voids volume fractions were 59.5%, 39.2%, 0.9% and
0.44%, respectively. The interphase volume fraction was obtained using the LOI value (0.64 wt %),
the mass of sized glass fibers (about 5.6 g), the density of the interphase (1.1 g/cm3 ) and the mass
of the composite (about 7.2 g). Fiber, matrix and interphase mass fractions were 77.2%, 22.3% and
0.5%, respectively. The fiber surface area of one plate was about 0.5 m2 on average. The composite
interphase mass fraction (mfi ) was calculated as:

LOI·mfibers
mfi = (1)
mcomp
Coatings 2019, 9, 263 6 of 24

where mfibers is the mass of the sized fibers; mcomp is the mass of the composite.

2.2.3. Conditioning of Composite Plates

Water uptake and hygrothermal aging of the composite laminates was conducted using a batch
system. A heated bath with distilled water (60 ± 1 ◦ C) was used for conditioning the samples.
Samples were weighed using analytical scales AG204 (±0.1 mg; Mettler Toledo, Columbus, OH, USA).
Samples were conditioned for a period of about a year. Three parallels were performed.

2.2.4. Specific Surface Area of the Fibers Obtained by N2 Sorption/Desorption and

Brunauer–Emmett–Teller (BET) Theory
The specific surface area of the sized and unsized glass fibers was obtained via N2 sorption
and desorption. The method uses physical adsorption and desorption of gas molecules based
on the Brunauer–Emmett–Teller (BET) theory [52]. The specific surface area was measured using
QUADRASORB SI (Quantachrome Instruments, Boynton Beach, FL, USA) equipment. BET tests
for specific surface area determination were performed according to the international standard ISO
9277:2010(E) [53]. The method is based on the determination of the amount of adsorptive gas molecules
covering the external surface of the solid [53].
Since the sizing’s surface is rough [12], the BET tests can provide the specific surface area. The BET
theory explains the physical adsorption of gas molecules on a solid surface of a material, and it is the
basis for the specific surface area determination.
Due to the roughness of the sizing on the fiber surface, the specific surface area of sized glass
fibers measured with BET was 0.180 m2 /g (see Figure 3), being higher than the specific surface area of
unsized glass fibers of 0.09 m2 /g (geometrical considerations as described in Section 2.1) or 0.084 m2 /g
using the BET method. For the unsized and sized glass fibers, the data with the BET model fit was
with a determination coefficient R2 of 0.968 and 0.994, respectively.

Figure 3. Brunauer–Emmett–Teller (BET) analysis of the specific surface area of the sized fibers.
Coatings 2019, 9, 263 7 of 24

3. Analytical Model

3.1. Mass Balance

When polymers take up water from the environment, their mass is affected by the water uptake
itself, leaching and aging mechanisms such as hydrolysis, chain scission or oxidation [46,54]. For the
studied epoxy, there is no significant mass loss due to chemical bond scission, since hydrolysis and
chain scission are not occurring [46,55].
The combination of the phenomenological perspective and mass balance approach provide a
useful tool for analyzing mass uptake/loss processes in composites during hygrothermal aging by
breaking down a complex process into constituent-related processes. The processes that affect weight
gain or loss of composites are summarized in Table 1.

Table 1. Summary of the processes during hygrothermal aging of composites that affect the mass balance.

Process Sign Reference

Water uptake of the polymer matrix + [49]
Water uptake by the composite interphase + [56]
Water uptake by the voids + [12,49,57]
Thermo-oxidation of the polymer matrix + [46]
Leaching from polymer matrix − [46]
Glass fiber dissolution − [28,47]
Sizing-rich interphase dissolution − This work

Gravimetric measurements determine the sample’s mass over time during conditioning in water.
The mass consists of the following terms:

mgravimetric (t) = mdry + mwater uptake (t) + moxidation (t) − mleaching (t)
(2)
−mglass dissolution (t) − minterphase dissolution (t)

The dissolution of the interphase is then simply given by:

minterphase dissolution (t) = mdry + mwater uptake (t) + moxidation (t) − mleaching (t)
(3)
− mglass dissolution (t) − mgravimetric (t)

The proposed model equation should be a phenomenologically full representation of the interaction
between the composite material and the water environment. More details will now be given for each
of the terms.

3.2. Water Uptake

The water uptake for composites includes three sub-processes: the uptake by the polymer matrix,
by the interphase, and by the voids [49]. The glass fibers themselves do not absorb any water.
The water taken up by the polymer matrix at any point of time is limited by the diffusivity
and the water saturation level [49,54]. The Fickian diffusion model can be used to model the water
uptake by the polymer and the interphase [49]. In addition, the effect of voids being filled with water
has to be considered [12,49]. The water content at saturation of the studied epoxy is 3.44 wt % if
no voids are present [56]. Saturation has been defined as the moment when the difference in two
consecutive water absorption measurements is lower than 0.5%, as defined by ASTM D5229 [44].
The composite’s saturation water content M∞ was determined to be 0.96 wt % [49]. It can be calculated
by Equation (4) [49]:
Mm (νm + νi )ρm + Mv∞ νv ρwater
M∞ = ∞ (4)
νf ρf + (νm + νi )ρm
Coatings 2019, 9, 263 8 of 24

where ρm is the matrix density, ρf is the fiber density, ρwater is the water density, νf is the fiber volume
fraction, νm is the matrix volume fraction, νi is the interphase volume fraction, νv is the void volume
fraction (νf + νm + νi + νv = 1), Mm v
∞ is the matrix saturation water content (3.44 wt %) and M∞ is the
void saturation water content (100 wt %). Fiber, matrix, interphase and voids volume fractions are
59.5%, 39.2%, 0.9% and 0.44%, respectively.
The sizing-rich interphase is assumed to have the same saturation water content as the epoxy
matrix, since it contains about 70–80 wt % epoxy film-former [12,29,56]. Since the volume of the sizing
is very small compared to the composite’s volume any deviation from this assumption would have a
minimal effect on the water uptake.
It is assumed here that the small voids will be completely filled with water Mv∞ = 1, as was
measured experimentally for the composite described here [49].
The water diffusivity of the studied epoxy polymer and the composites C1 and C3 in the thickness
direction (with the fibers running transverse and parallel to the thickness direction for C1 and C3,
respectively; see Figure 2) are systematized in Table 2 [49]. The higher diffusivity of the composite C3
is due to the fact that the diffusivity of the interphase in the direction parallel to the fibers is almost an
order of magnitude higher than that of the polymer, after [49].

Table 2. Diffusivities in the through-the-thickness direction, after [49].

Specimen D (mm2 /h)

Epoxy 0.0068
C1 0.0051
C3 0.0210

The following equation links the mass uptake to diffusivity from solving the 1-D Fickian diffusion
equation, as described by Crank [58]:
⎡ ⎤
⎢⎢  ∞ −(2i+1)( π )2 Dt ⎥
⎢ 8 e h ⎥⎥
⎢
M(t) = M∞ ⎢⎣1 − 2 ⎥⎥ (5)
π i=0 (2i + 1)2 ⎦

By fitting the exact solution of the diffusion equation to an exponential function, the ASTM
standard simplified equation is the following [44]:

0.75
−7.3( Dt2 )
M ( t ) = M∞ 1 − e h (6)

where M(t) is the water content, M∞ is the water saturation content, t is time, h is the thickness and D
is the diffusivity in the thickness direction of the plate.
More details and 3-D Fickian model calculations can be found elsewhere [49]. 1-D and 3-D Fickian
models gave the same result. Thus, for the sake of simplicity, the 1-D diffusion model for water uptake
is used in this work.
Experimental gravimetric measurements and modeled water uptake curves using Equation (6) are
shown in Figure 4 for a composite C3 with and without voids. It can be clearly seen that the absorption
of water in the voids needs to be modeled to get a good fit with the experimental data.
Coatings 2019, 9, 263 9 of 24

Figure 4. Experimental gravimetric measurements of composite C3 plates conditioned in water and

modeled water uptake curves using the Fickian 1-D model.

3.3. Oxidation of the Epoxy Matrix

Photo-oxidation is not present as the material is not exposed to high-energy irradiation [46].
The effect of thermo-oxidation on mass gain due to water uptake is negligible. Thermo-oxidation
for the studied epoxy polymer occurs via the carbonyl formation mechanism in the carbon-carbon
backbone via nucleophilic radical attack, as is described elsewhere [46].

3.4. Leaching of Molecules out of the Epoxy Matrix

Water molecules can migrate into the epoxy polymer while at the same time small molecules
may leach out of the matrix [59,60]. The leaching phenomenon may occur due to initially present
additives, impurities, unreacted hardener or degradation products diffusing out of the epoxy network
into the water environment, which is in contact with the polymer. Often leaching follows Fickian-type
diffusion [61]. The driving force of this process is due to the difference in concentration of these
chemicals inside the polymer, and in the surrounding aqueous environment.
Leaching was determined experimentally using HR-ICP-MS up to about 1100 h in another work
for the same epoxy material as used for making the composites in this study [46]. Krauklis and
Echtermeyer [46] found that for the studied epoxy polymer there was no leaching of hardener, whilst
the leaching occurred of epoxy compounds and impurities, such as epichlorohydrin and inorganic
compounds. Based on Fourier transform–near infrared (FT-NIR) spectra (reported in [46]) the leached
amount after about 1100 h of conditioning was estimated to be at 54.74 wt % of the initial leachable
compounds present in the material. This indicates that more than a half of the small molecules were
leached out after the relatively short time of 1100 h. The initial leachable compound content M0leaching
was found to be 0.092 wt % (about 1.5 mg) defined as the mass loss due to leaching divided by the
initial mass of the polymer (about 1.6 g).
Coatings 2019, 9, 263 10 of 24

The diffusivity of leached compounds through the epoxy polymer was determined according to
1-D Fickian diffusion [44,61]:
⎡ Dleaching t 0.75
⎤
⎢⎢ −7.3( ) ⎥⎥⎥
Mleaching (t) = M0leaching ⎢⎢⎣1 − e h2 ⎥⎦ (7)

The diffusivity was obtained by regression analysis of the data performing non-linear Generalized
Reduced Gradient (GRG) algorithm, while minimizing the residual sum of squares. The leaching
diffusivity Dleaching obtained in this study was 6.0 × 10−5 mm2 /h.
The leached-out compounds were experimentally measured with High-resolution inductively
coupled plasma mass spectrometry (HR-ICP-MS) (data from [46]). The modeled leaching behavior
from the matrix polymer is shown in Figure 5.

Figure 5. Polymer leaching determined experimentally with HR-ICP-MS, after [46], and modeled using
Fickian 1-D model, after [61].

3.5. Glass Dissolution

Glass fibers slowly degrade in water environments via dissolution reactions resulting in a
mass loss [47,62,63]. The degradation of glass fibers follows two distinct kinetic regions: short-term
non-steady-state (Phase I) and long-term steady-state degradation (Phase II), as described in the
dissolving cylinder zero-order kinetic (DCZOK) model for prediction of long-term dissolution of
glass from both fiber bundles [47]. During Phase I, the degradation is complex and involves such
processes as ion exchange, gel formation and dissolution. When Phase II is reached, the dissolution
becomes dominant and the degradation follows zero-order reaction kinetics. For the studied R-glass,
the transition from Phase I into Phase II occurs in about a week (166 h) at 60 ◦ C and pH 5.65 [28,47].
Elements that are released during degradation of R-glass are Na, K, Ca, Mg, Fe, Al, Si and Cl [47].
The glass mass loss is the cumulative mass loss of all these ions [47]. Si contribution to the total mass
loss of the studied R-glass is the largest (56.1 wt %) and seems to govern the dissolution process [47].
The rate of the dissolution depends on the apparent glass dissolution rate constant (K0∗ ) and the
glass surface area exposed to water (S) [28,47]. The glass surface area is proportional to the fiber
radius. As the dissolution continues, the radius decreases linearly with time resulting in the mass loss
deceleration; the DCZOK model accounts for this effect [47]. Rate constants at various environmental
conditions (pH, temperature and stress), as well as more details about the model can be found in other
works [28,47,63].
Coatings 2019, 9, 263 11 of 24

For a thin composite with fibers parallel to the short side through-thickness direction, such as in
this work, the dissolution of glass, compared to the free fiber bundles with sizing (not embedded in the
composite), is slowed down by 36.84% [28]. The differential mass loss equation for thin composites can
be written as [28]:
∂m
= K0∗ S(t) (8)
∂t
∗
The K0 includes the effects of diffusion and accumulation of the degradation products inside the
composite, the protective effect of the sizing and the availability of water [28,47,63]. The time-dependent
parameter is the fiber surface area S(t).
Considering the two distinct phases of the degradation, the full DCZOK model in the integral
form is the following, after [47]:
⎧  
⎪
⎪ K∗ I 2
⎪
⎪
⎨ t ≤ tst : mdissolved = nπl 2r0 K0∗ I t − 0ρ t2
⎪
⎪  f
 (9)
⎪ K∗ II 2
⎪
⎩ t > tst : mdissolved = mdissolvedtst + nπl 2rtst K0∗ II (t − tst ) − 0ρf (t − tst )
2

where n is the number of fibres (6450824); l is the length of fibres (1.5 mm); r0 is the initial fiber radius
(8.5 μm), and ρglass is the density of glass (2.54 g/cm3 ); K0∗ I and K0∗ II are the apparent dissolution rate
constants (g/m2 ·s) for the short-term non-steady-state (Phase I) and long-term steady-state (Phase II)
regions, respectively; rtst (m) and mdissolvedtst (g) are the fiber radius and lost mass after time tst (s),
when steady-state is reached (166 h [47,63]).
Using the composition of dissolving ions reported for the studied R-glass (Si contribution
56.1 wt %) [47], and the composite data after [28], K0∗ I and K0∗ II for the studied composite are 6.91 × 10−6
and 1.54 × 10−6 g/(m2 ·h), respectively. The dissolution rate constants are systematized in Table 3.
The glass mass loss was modeled using the DCZOK Equation (9) as shown in Figure 6. The glass mass
loss is normalized by the composite plate’s glass fiber surface area (about 0.5 m2 ).

Table 3. Apparent glass dissolution rate constants.

Phase K*0 (g/(m2 ·h))

Phase I 6.91 × 10−6
Phase II 1.54 × 10−6

These ions determined with HR-ICP-MS come from both glass material and the sizing-rich
interphase. HR-ICP-MS can capture ions from interface and interphase (ionic products of the
polysiloxane/siloxane hydrolysis), but ICP does not allow carbon detection due to CO2 in the plasma,
thus the organics from sizing-rich interphase are not captured. In other words, the predicted mass
loss due to dissolution includes ions coming from the interface and interphase, but does not include
organic compounds from the interphase. This is what makes the difference between the HR-ICP-MS
determined mass loss and the gravimetric mass loss of the composite.
Coatings 2019, 9, 263 12 of 24

Figure 6. Glass-fiber dissolution modeled using dissolving cylinder zero-order kinetics (DCZOK) for
the studied composite, after [28,47,63].

3.6. Interphase Dissolution

The aging of the sizing-rich composite interphase is the least understood constituent. The small
amount of the interphase sizing compared to the composite bulk material makes analysis difficult.
The proprietary nature of the sizing’s composition allows only general evaluations. For typical sizing
formulations, water interacting with the interphase may hydrate the Si–O–Si and Si–O–C bonds [5].
It was found that water molecules adsorbed in the epoxy matrix could migrate towards the sizing/glass
fiber interface through the sizing, resulting in the dissolution/decomposition of the polysiloxane [30].
The reaction with water breaks strained Si–O–Si bonds and generates Si–OH sites [12]. Principle silane
chemical bonding is reversible in the presence of water, thus the Si–O–Si bonds can be broken due to
hydrolysis, as shown in Chemical Reaction (10) [12]:

≡ Si–O–Si ≡ +H2 O ↔ 2 ≡ Si–OH (10)

In this work, the sizing-rich interphase loss is modeled assuming a simple zero-order kinetic model.

4. Results and Discussion

The increase of the composite’s mass with time within the first few hundred hours could be
fairly well described by a standard diffusion approach, as shown in Figure 4. It was important to
include the water uptake of the voids in the calculations. However, the diffusion approach would
predict a constant mass over time once saturation has been reached (0.96 wt %). The data of C3 show a
slight gradual drop in mass after saturation was reached, whereas the mass of C1 is clearly increasing,
as shown in Figure 7.
Coatings 2019, 9, 263 13 of 24

Figure 7. Long-term water uptake by composite laminates. Dashed line corresponds to a time when a
test following standard practice ASTM D5229 would be stopped [44].

If the water uptake experiments are stopped as suggested by ASTM, then the long-term behavior
is not captured. This observation is also consistent with the results of another study on long-term water
uptake by composite plates [45]. The diverging behavior of water uptake by C1 and C3 composites can
be observed starting only after about 20 h0.5 (about 2 weeks), only after the saturation M∞ (0.96 wt %)
has been already achieved. The discussion on how the diverging behavior of C1 and C3 can be captured
will follow.

4.1. Samples with Short Fibers C3

Firstly, the gravimetric behavior of C3 is addressed. As described above, a mass loss can be caused
by leaching material out of the epoxy and by the glass fibers losing ions. If these effects are added to
the mass vs. time curve a fairly good agreement with the experimental data is achieved, as shown in
Figure 8. It could be argued that the agreement is sufficient within the experimental scatter. However,
a closer look at the data can give some insight in the behavior of the sizing (interphase), although the
evaluation is at the limit of what can be analyzed considering the scatter of the results.
Looking at Figure 8, a slightly better fit of the data can be obtained with a curve that has a higher
mass loss with increasing time, i.e., is a bit steeper. This extra loss of material could be related to the
disintegration of the interphase. The simplest approach is to model the mass loss of the interphase
using the zero-order kinetics [64]:
∂mi
= Ki0 Si (t) (11)
∂t
where mi is the mass of the interphase, Ki0 is the kinetic coefficient of the interphase dissolution and
Si is the surface area of the interphase. The solution of this equation for cylindrical fibers is given in
Equation (9). For small mass changes and short times, the equation can be approximated by its first
linear term with the sizing having a constant surface area S0i to be:

mi (t) = m0i − Ki0 S0i t (12)

The initial mass of the sizing m0i (35.7 mg) was determined by the burn-off test to be 0.64 wt % of
the sized fibers. Fitting the data in Figure 8 allows finding Ki0 S0i , which basically describes the slightly
Coatings 2019, 9, 263 14 of 24

steeper slope compared to the previous analysis based only on matrix and glass fiber dissolution.
Using linear regression, as shown in Figure 9, The best fit for Ki0 S0i = 1.80 × 10−7 g/h.

Figure 8. Experimental composite C3 plate mass change during the conditioning in water, shown over
a square root of time. Water uptake and mass balance are modeled.

Figure 9. Linear regression of the difference between the experimental data and the all modelled terms
except the interphase. The regressed line provides insight about the rate of the composite interphase
dissolution in water.

Since dissolution is a surface reaction, a surface area of the sizing-rich interphase has to be obtained
in order to determine the kinetics of dissolution. Unfortunately, we do not know the exact surface
area of the sizing. Using the BET method, it was found that unsized fibers have a surface area of
Coatings 2019, 9, 263 15 of 24

0.084 m2 /g and sized fibers have a surface area of 0.180 m2 /g, roughly twice the value of the unsized
fibers. As discussed in the introduction, the sizing is rough which creates a larger surface [6,31,37],
but it also covers only parts of the fiber [12,32–37]. A typical sizing coverage of 90% of the glass fiber
surface is assumed, after [39]. Furthermore, the sizing is bonded to the glass fiber on one side and
the epoxy matrix on the other side, which does not create free surfaces at all. Based on the currently
available information, the only possibility is to calculate Ki0 for a number of plausible scenarios for the
surface area S0i .
The thickness of the interphase is obtained from the volume of the interphase Vi taking geometry
and known coverage (i.e., 0.9 or 1) into consideration. The volume of the interphase is known from
LOI (0.64 wt %; 35.8 mg) and interphase density (1.1 g/cm3 ), Vi = 0.0325 cm3 . The thickness of the
interphase is then obtained as follows:

Vi
δi = (13)
Coverage·Sglass

where Sglass is the total glass fiber surface area in a composite plate (about 0.5 m2 ). For 90% and 100%
coverage, a mean interphase thickness is 72 and 65 nm, respectively.
Scenario 1. The minimum surface area S0i would be just the cross-sectional area of the sizing
exposed on the surface of the composite specimen. The fiber fraction was 59.5% and the area of one
exposed surface of a C3 specimen was 50 mm × 50 mm. The surface area of fibers on both exposed
surfaces is then 2975 mm2 . The radius of an individual fiber was 8.5 μm. Based on the burn-off
method (LOI 0.64 wt %) and assuming extreme 100% coverage, the average sizing thickness was
65 nm. The ratio of exposed sizing cross sectional area to fiber cross sectional area is then 0.0149 and
the exposed sizing area is 44.3 mm2 . In this scenario Ki0 = 4.06 × 10−3 g/(m2 ·h). The sizing would be
dissolved along the axis of the fibers while the exposed cross section would remain constant until the
sizing is completely dissolved. Equation (12) would accurately describe dissolution in this scenario.
For these 1.5 mm-thick samples, the time to dissolve the sizing would be 22.7 years.
Scenario 2. The other extreme would be to argue that the epoxy is quickly saturated with water
(after about 100 and 81 h for C1 and C3, respectively), The water can then attack and dissolve the
sizing. In that case, the exposed area of the sizing would be much bigger. The BET method measured a
specific surface area of sized fibers to be 0.180 m2 /g. Then, the total surface area of sized fibers (5.6 g
fibers) in one plate is 1.01 m2 . Since the sizing covers only parts of the fiber, not all of this surface is
from the sizing. But to obtain an outer bound Ki0 can be calculated for this maximum surface area
(assuming coverage of 100%). In this case using Equation (12), Ki0 = 1.78 × 10−7 g/(m2 ·h). The Ki0
should be accurately determined by this equation for the relatively small area reduction during the
measurement. However, the proper cylindrical Equation (9) taking the surface area reduction with
time into account should be used to obtain the long-term dissolving of the sizing. The time to dissolve
the sizing would be 30.5 years.
Scenario 3. Considering the descriptions of the literature about sizing, a typical sizing covers
approximately 90% of the fiber [39]. In that case, the surface area of the sizing would be 0.91 m2 .
Using the same approach of a cylindrical sizing exposed to water in the epoxy as described for Scenario
2 above the Ki0 for this case would be 1.98 × 10−7 g/(m2 ·h) and the time to dissolve the sizing would be
30.5 years.
The parameters of the three scenarios are systematized in Table 4.

Table 4. Systematized scenarios of the interphase dissolution kinetics.

Sizing Coverage K0i Time to Total

Scenario K0i S0i (g/h) δi (nm) S0i (m2 )
(%) (g/(m2 ·h)) Dissolution (years)
Scenario 1 1.80 × 10−7 100 65 4.43 × 10−5 4.06 × 10−3 22.7
Scenario 2 1.80 × 10−7 100 65 1.01 1.78 × 10−7 30.5
Scenario 3 1.80 × 10−7 90, after [39] 72 0.91 1.98 × 10−7 30.5
Coatings 2019, 9, 263 16 of 24

The mass loss due to long-term gravimetric behavior of composite C3 could be successfully
modeled, because the C3 samples did not have a significant accumulation of the degradation products.
The C3 plates have a short fiber length (1.5 mm). Once the matrix is saturated with water, the water
can attack and degrade the interphase. Any reduction products can be quickly transported along the
interphase to the surface of the sample and will be absorbed by the surrounding water.

4.2. Samples with Long Fibers C1

The C1 samples showed a mass increase with time, see Figure 7, an additional 0.66 wt % of water
was taken up after 6673 h of conditioning. Since C1 and C3 samples were made from the same laminate,
just cut in a different direction, the change in behavior must be related to the sample’s geometry.
Compared to the C3 samples the C1 samples have much longer fibers and subsequently much longer
fiber matrix interphases (1.5 mm vs. 50 mm).
The matrix of both sample types absorbs water in roughly the same period (see Table 2). The water
will attack the interphase between fibers and matrix in the same way. But, it is believed that
degradation products (of fibers and interphase) cannot easily move along the interphase and escape
into the surrounding water at the composite’s surface. Instead, the weakening of the interphase causes
the formation of flaws. The degradation products and water can accumulate in these flaws. Thus,
the mass of the composite does not decrease with time as for samples C3, but the mass of C1 samples
increases with time. Figure 10 shows schematically what such a flaw could look like. Figure 11A shows
that such flaws are, indeed, observed in the samples.

Figure 10. Interphase flaw is formed and is filled with water.

Since the laminate absorbed another 0.66 wt % of water, it is possible to estimate the size of
flaws needed to accommodate this amount of water. The initial mass of the C1 plate was about
7.36 g. Water in the interphase flaws should thus weigh 48.6 mg, taking up volume of 4.86 × 10−8 m3 .
Assuming for the moment that all fibers have evenly distributed flaws, the following calculations can
be made. Dividing volume necessary to accommodate the extra water by the amount of fibers in a
composite C1 plate (193525) and the length of a fiber (50 mm), the cross-sectional area of a water-filled
interphase flaw around one fiber is found to be 5.02 × 10−12 m2 . The radius of the glass fiber is 8.5 μm,
thus the cross-sectional area of the fiber is 2.27 × 10−10 m2 . By combining cross-sectional areas of
the interphase flaw and the fiber, and deducting the radius of the fiber, an average thickness of a
water-filled interphase flaw of 93.5 nm is obtained.
In reality, not all interphase flaws are the same size and not all fiber/matrix interphases are
damaged equally, as shown in Figure 11. The weakest links will fail first. Once cracks are formed,
stresses are released and more complicated processes follow. However, it is interesting that the first
Coatings 2019, 9, 263 17 of 24

fiber matrix debondings, as shown in Figure 11A, have dimensions similar to the calculated value of
93.5 nm. Fiber/matrix debondings shown in Figure 11A range from about a 100 nm to a few microns,
as was observed experimentally using microscopy after 6673 h of conditioning. The thickness also
matches debonding dimensions observed elsewhere for the same composite [65,66].
Three damage mechanisms were observed in the micrographs:

• Fiber/matrix debondings, shown in Figure 11A.

• Matrix transverse cracks, shown in Figure 11B. These cracks seem to be inside the bundle.
This location may be also a result of the weakening of the fiber/matrix interphase, which was
covered in point 1.
• Splitting along the fibers, shown in Figure 11C.

Fiber/matrix debonding appears to be the first failure mechanism, caused by hydrolysis of the
interphase. This failure mechanism is described by the observations made for the C3 samples in
Section 4.1 When these failure mechanisms accumulate, creating a weakened local region, they can
easily combine into a longer “matrix crack” due to a release of curing, thermal and swelling stresses,
resulting in a crack formation. The reason for the observed splitting along fibers is less clear. It could be
related to the matrix cracks, but it could also be caused by the fibers used for stitching the reinforcing
mat. All these flaws (cracks) create volume that can be filled with water and increases the mass of
the composite.

Figure 11. Micrograph of a composite sample exposed to water for 6673 h at 60 ◦ C. The micrograph
indicates the (A) fiber/matrix debondings; (B) matrix transverse cracks; (C) splitting along the fibers.

Perreux, Choqueuse and Davies [45] investigated long-term water uptake by 2.7 mm-thick
composite plates. The plate was made with an anhydride-based curing agent while this study looked at
an epoxy laminate made with an amine-based curing agent. They observed that after ASTM saturation
was achieved, there was still a significant continuous mass gain up to about 5 years of conditioning in
water at 60 ◦ C. After this point, an abrupt and continuous mass loss occurred for the following 5 years
until the measurements were stopped. The data is schematically shown in Figure 12.
Coatings 2019, 9, 263 18 of 24

Figure 12. Schematic representation of the long-term water uptake at 20, 40 and 60 ◦ C observed by
Perreux, Choqueuse and Davies for the 2.7 mm thick composite plates [45].

The results seem to be a combination of what was found for samples C1 and C3 investigated here.
An explanation for the behavior observed by Perreux et al. [45] may be given by the findings of this
work. Initially flaws form in the composite interphase that is filled with water, resulting in a mass gain
as found in C1 samples. At one point, so many flaws have accumulated that an open interpenetrating
network with access to the surface of the laminate has formed. This network allows the degradation
products and previously absorbed water to diffuse out, creating a mass loss similar to samples C3
(since fiber lengths in C3 were so small the interpenetrating network was present from the beginning).
Since the observed mass reduction happened gradually, this means that the network of flaws and
cracks is gradually being connected to the sample’s surface. The mass drop was not observed for tests
carried at lower temperatures. In that case all processes are slower and the samples only increased
their mass, indicating the formation of flaws and cracks. But a network of the cracks reaching the
surface was not created yet. It should be noted that the matrix of anhydride-based epoxies studied by
Perreux et al. [45] is also prone to hydrolysis, so the hydrolysis in their samples may have affected the
matrix and the fiber–matrix interphase.

4.3. General Aspects

For the composite laminates studied here, about 3.5 wt % of the interphase was dissolved in a
year’s time. The expected total dissolution for the geometry of the C3 sample would occur between 22.7
and 30.5 years, according to the three scenarios at 60 ◦ C. At lower temperatures, the processes would
be significantly slower, because diffusivities and dissolution rates follow Arrhenius-type temperature
dependence [63,67]. Activation energies of these individual processes differ. Thus, it is likely not a
straightforward Arrhenius-type influence on the process rate as a whole (summary mass uptake or
loss).
The degradation time (22.7 to 30.5 years) should be independent of sample geometry and should
be applicable once the matrix has reached saturation. For thick composite laminates the fiber–matrix
interphase may only degrade in the surface region, because the matrix in the inside may remain dry.
Degradation may also be stopped or slowed down by an accumulation of reaction product, if the
degradation reaction is reversible, such as reaction (a). The mass uptake data obtained here showed a
slight slowdown of the reaction after 9 months, close to the point when experiments were stopped, see
Coatings 2019, 9, 263 19 of 24

Figure 7. But it is unclear whether the data really flatten out. The test results from Perreux et al. [45]
run over 10 years indicate that the degradation continues all the time.
Damage caused by the hydrolytic aging of the sizing-rich composite interphase very likely leads
to a decrease in interfacial strength. For instance, Gagani et al. [68] and Rocha et al. [42] have reported
the composite interphase-related deterioration of the mechanical properties due to aging in water. It is
likely that the formation of the interphase flaws described in this work is the mechanistic origin of the
interfacial strength deterioration of composites.
The authors think that studying the effect of seawater on the hydrolysis of the interphase would
be useful, since the composite marine structures are most often used in the seawater environments.
The dissolution in seawater conditions is expected to occur slower than in distilled water due to
the presence of silica (dissolved from sand and other minerals). The reason for an expected aging
rate slowdown in seawater is that the degradation products are already present in the surrounding
environment, thus decreasing the driving force—a concentration gradient.
The length of glass fibers should not affect the molecular structure or morphology of the interphase
per se. However, it should be added that what is affected by the fiber length is the path (or length) that
the hydrolytic degradation products have to travel in order to escape the composite material and diffuse
out into the surrounding water. It was shown in this work that water interaction with composites with
very short interphase leads to mass loss, whereas for a typical composite an interpenetrating flaw
network takes a relatively long time to form in order for degradation products to leave the composite.
This leads to another aspect that needs to be studied in more detail: a diffusion of degradation products
through the interphase. It is important to understand whether there is a diffusion-controlled aspect.
This paper covers hydrolysis of the composite interphase, but the same approach should be
applicable for all other environmental agents and solvents (in general, solvolysis).

5. Conclusions
Glass fiber composites absorb water with time and the mass of the composites increase subsequently.
When measuring diffusivity and saturation level of water according to ASTM D5229 [44] testing is
stopped when the mass increase with time stops, i.e., it is reaching a plateau, in this case at about
200 h. However, continuing the tests exposing the laminates to water for longer, the mass of the
composite increases again, measured up to 9 months. This additional water uptake was found to be
due to the hydrolytic degradation of the sizing-rich fiber matrix interphase. Due to water-induced
dissolution interphase flaws being formed which developed further into matrix cracks. The internal
volume created by the flaws and cracks can be filled with water leading to the observed mass increase.
The microscopically measured size of the flaws matches the order of magnitude of the volume required
for obtaining the measured additional mass increase.
The hydrolytic degradation of the fiber matrix interphase could be investigated directly
by cutting non-typical specimens from a thick composite laminate. The test specimens were
50 mm × 50 mm × 1.5 mm coupons where all the fibers were running parallel to the short edge.
This created specimens with a short fiber–matrix interface length and the interphases being connected
to the large sample’s surface. When these specimens were conditioned in water, their mass increased
during the first 200 h as the typical specimens described above. Continuing the test for longer
times leads, however, to a mass loss. For these specimens, the flaws created by the fiber matrix
interphase hydrolysis were open towards the surface of the test specimen, since the interphase length
(and fiber length) was so short, 1.5 mm. The reaction products of the hydrolysis could migrate into the
surrounding water bath leading to a mass drop. This mass loss allowed the product of the dissolution
rate constant and the surface area of the interphase Ki0 S0i to be determined. The small specimens tested
here would degrade the entire interphase within 22 to 30 years at 60 ◦ C. The calculation is based on a
full mechanistic mass balance approach considering all the composite’s constituents: water uptake
and leaching of the matrix, dissolution of the glass fibers, and dissolution of the composite interphase.
These processes were modeled using a combination of Fickian diffusion and zero-order kinetics.
Coatings 2019, 9, 263 20 of 24

Based on long-term test data from the literature tested for close to 10 years, it seems that
composites will initially absorb extra water in the flaws and cracks created by interphase hydrolysis.
Eventually these cracks will create a network that is connected to the surface of the composite laminate.
When this network is formed reaction products can leave the laminate and the mass will be reduced.
The possible strength degradation due to the flaws in the fiber matrix interface forming within 22
to 30 years (for the tested type of epoxy laminate) in saturated laminates should be taken into account
in designs for long lifetimes.

Author Contributions: Conceptualization, A.E.K. and A.T.E.; Methodology, A.E.K.; Formal Analysis, A.E.K.;
Investigation, A.E.K. and A.I.G.; Resources, A.E.K., A.I.G. and A.T.E.; Data Curation, A.E.K., A.I.G. and A.T.E.;
Writing—Original Draft Preparation, A.E.K.; Writing—Review and Editing, A.E.K. and A.T.E.; Validation, A.E.K.;
Visualization, A.E.K.; Supervision, A.T.E.; Project Administration, A.T.E.; Funding Acquisition, A.T.E.
Funding: This research was funded by The Research Council of Norway (Project 245606/E30 in the Petromaks
2 programme).
Acknowledgments: This work is part of the DNV GL led Joint Industry Project “Affordable Composites” with
19 industrial partners and the Norwegian University of Science and Technology (NTNU). The authors would like
to express their thanks for the financial support from The Research Council of Norway (Project 245606/E30 in the
Petromaks 2 programme). The authors are thankful to Erik Sæter, Valentina Stepanova, Susana Villa Gonzalez
and Julie Asmussen. Andrey is especially thankful to Oksana V. Golubova.
Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations
ρf Density of the glass fibers (g/m3 )
ρm Density of the matrix polymer (g/m3 )
ρi Density of the sizing-rich composite interphase (g/m3 )
ρwater Density of the water (g/m3 )
h Thickness of a material plate (m)
νf Volume fraction of the fibers (m3 /m3 )
νm Volume fraction of the matrix polymer (m3 /m3 )
νi Volume fraction of the composite interphase (m3 /m3 )
νv Volume fraction of the voids (m3 /m3 )
M(t) Time-dependent water content of the composite (wt %)
M∞ Saturation water content of the composite (wt %)
Mm (t) Time-dependent water content of the matrix polymer (wt %)
Mm ∞ Saturation water content of the matrix polymer (wt %)
Mv∞ Saturation water content of the voids (wt %)
D Through-thickness water diffusivity of the material (mm2 /h)
Mleaching (t) Time-dependent content of leached compounds from the polymer (wt %)
M0leaching Initial leachable compound content in the polymer (wt %)
Dleaching Through-thickness leachable compound diffusivity of the material (mm2 /h)
r(t) Time-dependent fiber radius (m)
r0 Initial fiber radius (m)
rtst Fiber radius when the steady-state dissolution is reached (m)
K0 Glass dissolution rate constant (g/(m2 ·s))
K0∗ Apparent glass dissolution rate constant (g/(m2 ·s))
K0∗ I Apparent glass dissolution rate constant (non-steady-state; Phase I) (g/(m2 ·s))
K0∗ II Apparent glass dissolution rate constant (steady-state; Phase II) (g/(m2 ·s))
tst Time when long-term steady-state is reached (s)
n Number of fibers (–)
l Length of fibers and the interphase (m)
Coatings 2019, 9, 263 21 of 24

S(t) Time-dependent glass fiber surface area (m2 )

S0 Initial glass fiber surface area (m2 )
t Time (s)
m; mdissolved Glass mass loss due to dissolution (g)
mdissolvedtst Dissolved glass mass when the steady-state is reached (g)
ξsizing Protective effect of sizing against glass dissolution (–)
norder Order of the water availability term (–)
Si (t) Time-dependent surface area of the composite interphase (m2 )
Si0 Initial surface area of the composite interphase (m2 )
specific
Si Specific surface area of the composite interphase (m2 )
Ki0 Zero-order rate constant of the composite interphase dissolution (g/(m2 ·s))
mi (t) Time-dependent mass of the composite interphase (g)
mi0 Initial mass of the composite interphase (g)
GF Glass fiber
GFRP Glass fiber-reinforced polymer; same as glass fiber-reinforced composite
DCZOK Dissolving cylinder zero-order kinetic (model)
DGEBA Bisphenol A diglycidyl ether
HDDGE 1,6-Hexanediol diglycidyl ether
POPA Poly(oxypropylene)diamine
IPDA Isophorondiamine
R-glass “Reinforcement” glass
FRP Fiber-reinforced polymer, same as fiber-reinforced composite
HR-ICP-MS High-resolution inductively coupled plasma mass spectrometry\
VARTM Vacuum-assisted resin transfer molding
BET Brunauer–Emmett–Teller theory
LOI Loss on ignition
γ-APS APTES γ-aminopropyltriethoxysilane
PPO Poly(propylene oxide)
PEO Poly(ethylene oxide)
PDMS Polydimethylsiloxane

References
1. Berg, J.; Jones, F.R. The role of sizing resins, coupling agents and their blends on the formation of the
interphase in glass fiber composites. Compos. Part A 1998, 29, 1261–1272. [CrossRef]
2. Feih, S.; Wei, J.; Kingshott, P.; Sørensen, B.F. The influence of fiber sizing on the strength and fracture
toughness of glass fiber composites. Compos. Part A 2005, 36, 245–255. [CrossRef]
3. Dai, Z.; Shi, F.; Zhang, B.; Li, M.; Zhang, Z. Effect of sizing on carbon fiber surface properties and fibers/epoxy
interfacial adhesion. Appl. Surf. Sci. 2011, 257, 6980–6985. [CrossRef]
4. Yuan, X.; Zhu, B.; Cai, X.; Liu, J.; Qiao, K.; Yu, J. Optimization of interfacial properties of carbon fiber/epoxy
composites via a modified polyacrylate emulsion sizing. Appl. Surf. Sci. 2017, 401, 414–423. [CrossRef]
5. DiBenedetto, A.T. Tailoring of interfaces in glass fiber reinforced polymer composites: A review. Mater. Sci.
Eng. A 2001, 302, 74–82. [CrossRef]
6. Plonka, R.; Mäder, E.; Gao, S.L.; Bellmann, C.; Dutschk, V.; Zhandarov, S. Adhesion of epoxy/glass fiber
composites influenced by aging effects on sizings. Compos. Part A 2004, 35, 1207–1216. [CrossRef]
7. Grabovac, I.; Whittaker, D. Application of bonded composites in the repair of ships structures—A 15-year
service experience. Compos. Part A 2009, 40, 1381–1398. [CrossRef]
8. McGeorge, D.; Echtermeyer, A.T.; Leong, K.H.; Melve, B.; Robinson, M.; Fischer, K.P. Repair of floating
offshore units using bonded fibre composite materials. Compos. Part A 2009, 40, 1364–1380. [CrossRef]
9. Gustafson, C.-G.; Echtermeyer, A. Long-term properties of carbon fibre composite tethers. Int. J. Fatigue
2006, 28, 1353–1362. [CrossRef]
10. Salama, M.M.; Stjern, G.; Storhaug, T.; Spencer, B.; Echtermeyer, A. The first offshore field installation
for a composite riser joint. OTC-14018-MS. In Proceedings of the Offshore Technology Conference,
Houston, TX, USA, 6–9 May 2002. [CrossRef]
Coatings 2019, 9, 263 22 of 24

11. Echtermeyer, A.T.; Gagani, A.I.; Krauklis, A.E.; Mazan, T. Multiscale modelling of environmental
degradation—First steps. In Durability of Composites in a Marine Environment 2. Solid Mechanics and Its
Applications; Davies, P., Rajapakse, Y.D.S., Eds.; Springer: Cham, Switzerland, 2018; Volume 245, pp. 135–149.
ISBN 978-3-319-65145-3.
12. Thomason, J.L. Glass Fiber Sizings: A Review of the Scientific Literature; James L Thomason: Middletown, DE,
USA, 2012; ISBN 978-0-9573814-1-4.
13. Weitsman, Y. Coupled damage and moisture-transport in fiber-reinforced, polymeric composites. Int. J.
Solids Struct. 1987, 23, 1003–1025. [CrossRef]
14. Weitsman, Y.J.; Elahi, M. Effects of fluids on the deformation, strength and durability of polymeric
composites—An overview. Mech. Time-Depend. Mater. 2000, 4, 107–126. [CrossRef]
15. Roy, S. Moisture-induced degradation. In Long-Term Durability of Polymeric Matrix Composites; Pochiraju, V.K.,
Tandon, P.G., Schoppner, A.G., Eds.; Springer: Boston, MA, USA, 2012; pp. 181–236. ISBN 978-1-4419-9307-6.
16. Peters, L. Influence of glass fibre sizing and storage conditions on composite properties. In Durability of
Composites in a Marine Environment 2. Solid Mechanics and Its Applications; Davies, P., Rajapakse, Y.D.S., Eds.;
Springer: Cham, Switzerland, 2018; Volume 245, pp. 19–31. ISBN 978-3-319-65145-3.
17. Culler, S.R.; Ishida, H.; Koenig, J.L. Hydrothermal Stability of γ-Aminopropyltriethoxysilane Coupling Agent
on Ground Silicon Powder and E-Glass Fibers; Technical Report; Department of Macromolecular Science:
Cleveland, OH, USA, 1983.
18. Wang, D.; Jones, F.R.; Denison, P. TOF SIMS and XPS study of the interaction of hydrolysed
γ-aminopropyltriethoxysilane with E-glass surfaces. J. Adhes. Sci. Technol. 1992, 6, 79–98. [CrossRef]
19. Wang, D.; Jones, F.R.; Denison, P. Surface analytical study of the interaction between γ-amino propyl
triethoxysilane and E-glass surface. Part I Time-of-flight secondary ion mass spectrometry. J. Mater. Sci.
1992, 27, 36–48. [CrossRef]
20. Wang, D.; Jones, F.R. Surface analytical study of the interaction between γ-amino propyl triethoxysilane and
E-glass surface. Part II X-ray photoelectron spectroscopy. J. Mater. Sci. 1993, 28, 2481–2488. [CrossRef]
21. Wang, M.; Xu, X.; Ji, J.; Yang, Y.; Shen, J.; Ye, M. The hygrothermal aging process and mechanism of the
novolac epoxy resin. Compos. Part B 2016, 107, 1–8. [CrossRef]
22. Halpin, J.C. Effects of Environmental Factors on Composite Materials; Technical Report AFML-TR-67–423;
Air Force Materials Laboratory: Dayton, OH, USA, 1969.
23. ASTM D4963/D4963M-2011 Standard Test Method for Ignition Loss of Glass Strands and Fabrics; ASTM:
West Conshohocken, PA, USA, 2011.
24. Loewenstein, K.L. Glass Science and Technology (Book 6), The Manufacturing Technology of Continuous Glass
Fibres; Elsevier: Amsterdam, The Netherlands, 1993; ISBN 978-0444893468.
25. Thomason, J.L.; Adzima, L.J. Sizing up the interphase: An insider’s guide to the science of sizing.
Compos. Part A 2001, 32, 313–321. [CrossRef]
26. Plueddemann, E.P. Silane Coupling Agents, 2nd ed.; Plenum Press: New York, NY, USA, 1991;
ISBN 978-0-306-43473-0.
27. Emadipour, H.; Chiang, P.; Koenig, J.L. Interfacial strength studies of fibre-reinforced composites. Res. Mech.
1982, 5, 165–176.
28. Krauklis, A.E.; Echtermeyer, A.T. Dissolving cylinder zero-order kinetic model for predicting hygrothermal
aging of glass fibre bundles and fibre-reinforced composites. In Proceedings of the 4th International Glass
Fibre Symposium, Aachen, Germany, 29–31 October 2018; pp. 66–72, ISBN 978-3-95886-249-4.
29. Joliff, Y.; Belec, L.; Chailan, J.-F. Impact of the interphases on the durability of a composite in humid
environment—A short review. In Proceedings of the 20th International Conference on Composite Structures
ICCS20, Paris, France, 4–7 September 2017.
30. Zhuang, R.-C.; Burghardt, T.; Mäder, E. Study on interfacial adhesion strength of single glass
fiber/polypropylene model composites by altering the nature of the surface of sized glass fibers.
Compos. Sci. Technol. 2010, 70, 1523–1529. [CrossRef]
31. Wolff, V.; Perwuelz, A.; El Achari, A.; Caze, C.; Carlier, E. Determination of surface heterogeneity by contact
angle measurements on glass fibres coated with different sizings. J. Mater. Sci. 1999, 34, 3821–3829. [CrossRef]
32. Ishida, H.; Koenig, J.L. An investigation of the coupling agent/matrix interface of fiberglass reinforced plastics
by fourier transform infrared spectroscopy. Polym. Phys. B 1979, 17, 615–626. [CrossRef]
Coatings 2019, 9, 263 23 of 24

33. Watson, H.; Mikkola, P.J.; Matisons, J.G.; Rosenholm, J.B. Deposition characteristics of ureido silane ethanol
solutions onto E-glass fibres. Colloids Surf. A 2000, 161, 183–192. [CrossRef]
34. Feresenbet, E.; Raghavan, D.; Holmes, G.A. Influence of silane coupling agent composition on the surface
characterization of fiber and on fiber-matrix interfacial shear strength. J. Adhes. 2003, 79, 643–665. [CrossRef]
35. Fagerholm, H.M.; Lindsjö, C.; Rosenholm, J.B.; Rökman, K. Physical characterization of E-glass fibres treated
with alkylphenylpoly(oxyethylene)alcohol. Colloids Surf. 1992, 69, 79–86. [CrossRef]
36. Thomason, J.L.; Dwight, D.W. The use of XPS for characterization of glass fibre coatings. Compos. Part A
1999, 30, 1401–1413. [CrossRef]
37. Turrión, S.G.; Olmos, D.; González-Benito, J. Complementary characterization by fluorescence and AFM of
polyaminosiloxane glass fibers coatings. Polym. Test. 2005, 24, 301–308. [CrossRef]
38. Mai, K.; Mäder, E.; Mühle, M. Interphase characterization in composites with new non-destructive methods.
Compos. Part A 1998, 29, 1111–1119. [CrossRef]
39. Thomason, J.L.; Dwight, D.W. XPS analysis of the coverage and composition of coatings on glass fibers.
J. Adhes. Sci. Technol. 2000, 14, 745–764. [CrossRef]
40. Wang, D.; Jones, F.R. TOF SIMS and XPS study of the interaction of silanized E-glass with epoxy resin.
J. Mater. Sci. 1993, 28, 1396–1408. [CrossRef]
41. Chiang, C.H.; Ishida, H.; Koenig, J.L. The structure of aminopropyltriethoxysilane on glass surfaces. J. Colloid
Interface Sci. 1980, 74, 396–404. [CrossRef]
42. Rocha, I.B.C.M.; Raijmaekers, S.; Nijssen, R.P.L.; van der Meer, F.P.; Sluys, L.J. Hygrothermal ageing behaviour
of a glass/epoxy composite used in wind turbine blades. J. Compos. Struct. 2017, 174, 110–122. [CrossRef]
43. Kim, J.K.; Sham, M.L.; Wu, J. Nanoscale characterization of interphase in silane treated glass fibre composites.
Compos. Part A 2001, 32, 607–618. [CrossRef]
44. ASTM D5229/D5229M-14 Standard Test Method for Moisture Absorption Properties and Equilibrium Conditioning
of Polymer Matrix Composite Materials; ASTM International: West Conshohocken, PA, USA, 2014.
45. Perreux, D.; Choqueuse, D.; Davies, P. Anomalies in moisture absorption of glass fibre reinforced epoxy
tubes. Compos. Part A 2002, 33, 147–154. [CrossRef]
46. Krauklis, A.E.; Echtermeyer, A.T. Mechanism of yellowing: carbonyl formation during hygrothermal aging
in a common amine epoxy. Polymers 2018, 10, 1017. [CrossRef] [PubMed]
47. Krauklis, A.E.; Echtermeyer, A.T. Long-term dissolution of glass fibers in water described by dissolving
cylinder zero-order kinetic model: Mass loss and radius reduction. Open Chem. 2018, 16, 1189–1199.
[CrossRef]
48. 3B Fibreglass Technical Data Sheet; HiPer-Tex W2020 Rovings: Belgium, Brussel, 2012.
49. Gagani, A.I.; Fan, Y.; Muliana, A.H.; Echtermeyer, A.T. Micromechanical modeling of anisotropic water
diffusion in glass fiber epoxy reinforced composites. J. Compos. Mater. 2017, 52, 2321–2335. [CrossRef]
50. Zinck, P.; Gerard, J.F. On the hybrid character of glass fibres surface networks. J. Mater. Sci. 2005, 40,
2759–2760. [CrossRef]
51. ASTM D3171/D3171-15 Standard Test Methods for Constituent Content of Composite Materials; ASTM International:
West Conshohocken, PA, USA, 2015.
52. Brunauer, S.; Emmett, P.H.; Teller, E. Adsorption of gases in multimolecular layers. J. Am. Chem. Soc. 1938,
60, 309–319. [CrossRef]
53. International Standard ISO 9277:2010(E) Determination of the Specific Surface Area of Solids by Gas Adsorption—BET
Method; ISO: Berlin, Germany, 2010.
54. Popineau, S.; Rondeau-Mouro, C.; Sulpice-Gaillet, C.; Shanahan, M.E.R. Free/bound water absorption in an
epoxy adhesive. Polymer 2005, 46, 10733–10740. [CrossRef]
55. Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Hygrothermal aging of amine epoxy: reversible static and
fatigue properties. Open Eng. 2018, 8, 447–454. [CrossRef]
56. Krauklis, A.E.; Gagani, A.I.; Echtermeyer, A.T. Near-Infrared Spectroscopic Method For Monitoring Water
Content In Epoxy Resins And Fiber-Reinforced Composites. Materials 2018, 11, 586. [CrossRef]
57. Thomason, J.L. The interface region in glass-fibre-reinforced epoxy resin composites: 2. Water absorption,
voids and the interface. Composites 1995, 26, 477–485. [CrossRef]
58. Crank, J. The Mathematics of Diffusion, 2nd ed.; Clarendon Press: Oxford, UK, 1975; ISBN 978-0-19-853411-6.
59. Maggana, C.; Pissis, P. Water sorption and diffusion studies in an epoxy resin system. J. Polym. Sci. Part B
1999, 37, 1165–1182. [CrossRef]
Coatings 2019, 9, 263 24 of 24

60. Toscano, A.; Pitarresi, G.; Scafidi, M.; Di Filippo, M.; Spadaro, G.; Alessi, S. Water diffusion and swelling
stresses in highly crosslinked epoxy matrices. Polym. Degrad. Stab. 2016, 133, 255–263. [CrossRef]
61. Bruchet, A.; Elyasmino, N.; Decottignies, V.; Noyon, N. Leaching of bisphenol A and F from new and old
epoxy coatings: Laboratory and field studies. Water Sci. Technol. 2014, 14, 383–389. [CrossRef]
62. Schutte, C.L. Environmental durability of glass-fiber composites. Mater. Sci. Eng. R Rep. 1994, 13, 265–323.
[CrossRef]
63. Krauklis, A.E.; Gagani, A.I.; Vegere, K.; Kalnina, I.; Klavins, M.; Echtermeyer, A.T. Dissolution kinetics of
R-glass fibres: Influence of water acidity, temperature and stress corrosion. Fibers 2019, 7, 22. [CrossRef]
64. Khawam, A.; Flanagan, D.R. Solid-state kinetic models: basics and mathematical fundamentals. J. Phys.
Chem. B 2006, 110, 17315–17328. [CrossRef] [PubMed]
65. Gagani, A.I.; Mialon, E.P.V.; Echtermeyer, A.T. Immersed interlaminar fatigue of glass fiber epoxy composites
using the I-beam method. Int. J. Fatigue 2019, 119, 302–310. [CrossRef]
66. Rocha, I.B.C.M.; van der Meer, F.P.; Raijmaekers, S.; Lahuerta, F.; Nijssen, R.P.L.; Mikkelsen, L.P.; Sluys, L.J.
A combined experimental/numerical investigation on hygrothermal aging of fiber-reinforced composites.
Eur. J. Mech. Sol. 2019, 73, 407–419. [CrossRef]
67. Bonniau, P.; Bunsell, A.R. Water absorption by glass fibre reinforced epoxy resin. In Composite Structures;
Marshall, I.H., Ed.; Springer: Dordrecht, The Netherlands, 1981; pp. 92–105. ISBN 978-94-009-8122-5.
68. Gagani, A.I.; Krauklis, A.E.; Sæter, E.; Vedvik, N.P.; Echtermeyer, A.T. A novel method for testing and
determining ILSS for marine and offshore composites. Comp. Struct. 2019, 220, 431–440. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).