PHD THESIS - u-bordeaux.frori-oai.u-bordeaux1.fr/pdf/2010/BUDZIK_MICHAL_2010.pdf · 2010. 12....
Transcript of PHD THESIS - u-bordeaux.frori-oai.u-bordeaux1.fr/pdf/2010/BUDZIK_MICHAL_2010.pdf · 2010. 12....
Numéro d'ordre de la thèse: 4041
Gdansk, 18.06.2010
PHD THESIS
BORDEAUX UNIVERSITY ECOLE DOCTORALE
DES SCIENCES PHYSIQUES ET DE L'INGENIEUR
GDANSK UNIVERSITY OF TECHNOLOGY
by Michał K. BUDZIK
TO OBTAIN THE DEGREE: PhD
IN: Mecanique et Ingenierie
SUPERVISORS:
Prof. Martin E.R. SHANAHAN (Bordeaux 1 University)
D.Sc. Krystyna IMIELIŃSKA (Gdansk University of Technology)
COORDINATOR:
PhD Julien JUMEL (Bordeaux 1 University)
FRACTURE IN ASYMMETRIC BONDED JOINTS
JURY:
Prof. A.S. WRONSKI (University of Bradford, United Kingdom) – Chairman, Reviewer
Prof. W. POSSART (University of Saarland, Germany) – Reviewer
DSc. M. WASILCZUK (Gdansk University of Technology, Poland) – Reviewer
DSc. M. SZKODO (Gdansk University of Technology, Poland) – Member
Prof. A. ZIELINSKI (Gdansk University of Technology, Poland) – Member
PhD. J. JUMEL (Bordeaux 1 University, France) – Member
Prof. M.E.R. SHANAHAN (Bordeaux 1 University, France) – Supervisor
DSc. K. IMIELINSKA (Gdansk University of Technology, Poland) - Supervisor
Within Joint-Thesis Agreement between Bordeaux University and Gdansk University of
Technology
FRACTURE IN ASYMMETRIC BONDED JOINTS
2
Acknowledgements
I wish to thank
My supervisors:
Prof. Martin E.R. Shanahan for taking me under his wings, much precious advice, inspiration
and coordination of my research as well as friendship
and
Dr. Krystyna Imielińska for coordinating my research programme in Poland, encouragement
and friendship, valuable consultations and assistance
I am grateful
Julien Jumel, my amazing coordinator in France for his guidance and inspiration, as well as
his friendliness and partnership cooperation that made my work and life in Bordeaux much
easier
Jean Marc Olive for his commitment in the French-Polish collaboration program and creating
the opportunity for my doctoral studies in France
Prof. Andrzej Zielinski for establishing and developing the scientific contacts with LMP,
which allowed my work there
Jeremy Guitard and Lech Targan for their everyday efforts to facilitate my experimental work
Bernard Hosten, Michel Castaings and Christine Biateau for their assistance in ultrasonic
studies
The French Government for financial support (PhD grant) that enabled my stay and work in
France
My family, wife – Jadzia for her patience and support, parents, sister for believing in me
Michał K. Budzik
3
Abstract
Adhesion was studied in asymmetric bonded joints using fracture mechanics tests. The
asymmetric bonded joints consist of two different type and/or thickness materials bonded by
an adhesive. Mentions of asymmetric bonded joint tests employed so far are rare in the
literature. They are imperfect and therefore are not standardized. Accordingly three new tests
were introduced in this work to study bonded joints. The new metrological routines and
models were built for the CRT (Constant Rate Test), CFT (Constant Force Test) and the CDT
(Constant Displacement Test). The routines were validated with the new Artificial Crack Tip
test developed in this thesis. Different bonded systems were examined: the adhesives - epoxy,
cyanoacrylate, acrylic mastic, PSA; the bonded adherends - aluminium alloys, CFRP
composites and polycarbonate. The results obtained in the new tests are very promising in
terms of the accuracy and continuous observation of crack kinetics. In addition it was found
that the novel strain gauge technique, introduced in the CDT, test allows precise joint
monitoring when the adhesive is in a metastable state and a precise estimation of process
zone. Another interesting achievement of this work was describing the phenomenon of double
crack curvature in the vicinity of the strong – weak adhesion transition zone. The next
achievement was derivation of the strain energy release rate from the elastic foundation
model, which gives better understanding of the mechanics and the measurements behind
adhesive bonding. Atomic Force Microscope (AFM) and Scanning Electron Microscope
(SEM) studies were made of the bonded substrates to estimate the surface treatment effects.
The stress state in asymmetric geometry was studied using finite element analysis (FEA),
which explained the formation of the curved crack front and the origin of river patterns. The
new tests developed in this study appear promising since they offer accurate and reliable
results for the materials tested.
FRACTURE IN ASYMMETRIC BONDED JOINTS
4
Streszczenie
W pracy przedstawiono wyniki badań asymetrycznych złączy adhezyjnych z wykorzystaniem
prób mechaniki pękania. Asymetryczne złącza klejone powstają przez połączenie dwóch
dowolnych materiałów, o innym wskaźniku na zginanie, za pomocą kleju. Zaproponowano
trzy nowe metody eksperymentalne dla których zaprojektowano i zbudowano stanowiska
badawcze. Poprawność modeli fizycznych i matematycznych wykorzystanych do interpretacji
badanych zjawisk potwierdzono wykorzystując opracowaną w pracy próbę ze sztucznym
frontem pęknięcia. Próby przeprowadzono na złączach klejonych wykorzystując kleje:
epoksydowe, cjanoakrylowy, typu Mastic i PSA, łączonymi materiałami były: stopy
aluminium, kompozyt węglowy oraz poliwęglan. Wyniki badań uzyskane przy wykorzystaniu
nowych metod wskazują na ich pewność i dokładność, dodatkowo umożliwiają ciągłą
obserwację propagującego pęknięcia. Nowa metoda pomiaru parametrów mechaniki pękania
oparta na pomiarach tensometrycznych została z powodzeniem zastosowana do złączy w
których klej znajduje się w stanie metastabilnym. W pracy zaobserwowano i przeanalizowano
zjawisko powstawania podwójnego frontu pęknięcia na granicy ośrodków o różnych siłach
adhezji, dotąd nie opisanego. Po raz pierwszy wyprowadzono równanie szybkości uwalniania
energii sprężystej wykorzystując model belki na sprężystym podłożu, co umożliwia lepsze
zrozumienie zachowania się złączy klejonych. Przeprowadzone badania z użyciem
mikroskopii sił atomowych (AFM) oraz skaningowej mikroskopii elektronowej (SEM)
umożliwiły ocenę efektów obróbki powierzchniowej stopów aluminium. W pracy dokonano
analizy naprężeń w asymetrycznych złączach adhezyjnych z wykorzystaniem metody
elementów skończonych (FEM). Analiza FEM pozwoliła na wyjaśnienie zjawiska
powstawania zakrzywionego frontu pęknięcia w badanych przełomach. Zaproponowane nowe
metody badań złączy asymetrycznych zapewniają dokładność i pewność pomiaru dla
materiałów wykorzystanych w badaniach.
Michał K. Budzik
5
Résumé
Des tests de fissuration ont été réalisés sur des joints collés asymétriques. Ces assemblages
asymétriques sont constitués de deux substrats de nature et/ou d’épaisseur différentes liés par
un adhésif. Cette géométrie d’éprouvette semble peu utilisée pour effectuer des essais de
fissuration. Elles présentent certains inconvénients et de ce fait n’ont fait l’objet d’aucune
normalisation. Dans le cadre cette géométrie d’éprouvette d’essai a été néanmoins été retenue
et utilisée dans trois configurations différentes pour la caractérisation mécanique des
assemblages collés. Des protocoles d’essai et les schéma d’analyse associés ont été définis
pour des essais de fissuration à vitesse de déplacement imposé (CRT : Constant Rate Test), à
force imposée (CFT : Constant Force Test) ainsi qu’à déplacement imposé (CDT : Constant
Displacement Test). Les analyses ont pu être étalonnées au moyen d’un étau matérialisant
artificiellement la position d’un fissure. De nombreuses configurations ont été étudiées à
l’occasion de ce travail tant du point de vue des adhésifs utilisés (epoxy, cyanoacrylate,
acrylic mastic, PSA) que du point de vue des substrats encollés (alliage d’aluminium,
composite à matrice organique, polycarbonate). Ces nouveaux tests s’avèrent prometteurs car
ils permettent de suivre de façon continu et précision la propagation de fissure au sein de la
liaison. En outre, l’analyse des déformations mesurées par extensométrie lors d’essai réalisés
à déplacement imposé (CDT) permet un suivi précis de la réponse de la couche adhésive dans
la zone contrainte et d’estimer la taille de cette dernière, y compris lorsque la colle est dans un
état métastable. Un autre résultat important de ce travail concerne l’analyse de la modification
du front de fissure celle-ci voit une diminution de l’adhésion entre les deux substrats. Enfin,
un calcul plus précis du taux de restitution d’énergie a été proposé pour prendre en compte le
caractère élastique de la couche adhésive, mais permettant aussi de mieux appréhender la
redistribution des contraintes dans la colle, et ainsi mieux interpréter les essais mécaniques
réalisés. Des analyses par Microscopie à Force Atomique (AFM) ainsi qu’au microscope
électronique à balayage (MEB) ont été effectués sur les substrats encollés pour évaluer
l’impact des traitements de surface. Des simulations numériques par la method des éléments
finis ont été réalisées pour determiner l’état de contrainte dans nos éprouvette d’essai et ainsi
expliquer la courbure du front de fissure et les faciès de rupture observés. Les nouvelles
configurations expérimentales décrites dans ce manuscript nous paraissent prometteuses car
elle offrent une amelioration très significative en terme de fiabilité et précision
comparativement aux tests présentés habituellement pour ce type de travaux.
FRACTURE IN ASYMMETRIC BONDED JOINTS
6
NOTATION ............................................................................................................................... 8
INTRODUCTION ...................................................................................................................... 9
CHAPTER 1. BIBLIOGRAPHY ............................................................................................. 11
1.1. Historical perspective of adhesive bonding ................................................................... 11
1.2. Adhesive bonding applications ...................................................................................... 14
1.3. Structural adhesives ....................................................................................................... 19
1.4. Theories of adhesion ...................................................................................................... 29
1.5. Surface treatment of adherends ...................................................................................... 38
1.6. Mechanical testing of adhesive bonding ........................................................................ 46
1.6.1. Characterisation by failure stress ............................................................................ 47
1.6.2. Characterisation by failure energy .......................................................................... 52
CHAPTER 2. ASYMMETRIC JOINTS .................................................................................. 60
2.1. Materials and specimens ................................................................................................ 60
2.1.1. Characterization of substrates ................................................................................. 64
2.1.2. Surface preparation of the substrates ...................................................................... 65
2.1.3. AFM and SEM control of the surface ..................................................................... 67
2.1.4. Fabrication of adhesive joints ................................................................................. 69
2.2. Microscopic studies ....................................................................................................... 70
2.3. Crack path observations ................................................................................................. 70
CHAPTER 3. ADHESIVE BONDING TESTS AND ANALYSIS ........................................ 71
3.1. Constant Rate Test (CRT) .............................................................................................. 72
3.1.1. Data reduction method ............................................................................................ 73
3.1.2. Calibration of the crack length - artificial crack tip test .......................................... 76
3.1.3. Fracture of aluminium joints bonded with nanoparticle adhesive .......................... 77
3.2. Constant Force Test (CFT) ............................................................................................ 83
3.2.1. Data reduction method ............................................................................................ 84
3.2.2. Artificial crack tip test ............................................................................................. 87
3.2.3. The (macro) fracture behaviour of different adhesives ........................................... 88
3.3. Constant Displacement (Asymmetric Wedge) Test ....................................................... 91
3.3.1. Data reduction method ............................................................................................ 92
3.3.2. Calibration of the CDT test using artificial crack tip test ....................................... 98
3.3.4. Adhesive joint with variable adhesion properties ................................................. 102
3.3.5. Effects of The Adhesive Compliance ................................................................... 115
3.3.7. Use of the CDT test for assessment of curing time ............................................... 128
3.3.8. Temperature effects on fracture using the CDT test ............................................. 133
CHAPTER 4. MODELLING OF STRESSES IN ASYMMETRIC ADHESIVE JOINTS ... 143
4.1. FEM model .................................................................................................................. 143
4.2. Finite Element Analysis ............................................................................................... 146
4.2.1. Stress state with the straight crack front ............................................................... 146
4.2.2. Stress mixity .......................................................................................................... 147
4.2.3. Anti-plane shearing effect on fracture structure .................................................... 153
4.2.4. Crack depth ........................................................................................................... 154
DISCUSSION OF THE TESTS ............................................................................................. 159
Michał K. Budzik
7
CONCLUSIONS .................................................................................................................... 165
PERSPECTIVES .................................................................................................................... 166
APPENDICES ........................................................................................................................ 167
Appendix 1. Mode II contribution in Constant Displacement Test ................................ 167
Appendix 2. Friction dissipation in Constant Displacement Test ................................... 170
Appendix 3. List of publications ..................................................................................... 173
Appendix 4. Gantt chart of thesis progress ..................................................................... 174
REFERENCES ....................................................................................................................... 175
LIST OF FIGURES ................................................................................................................ 183
LIST OF TABLES ................................................................................................................. 190
FRACTURE IN ASYMMETRIC BONDED JOINTS
8
NOTATION
NOTATION UNIT PARAMETER
E GPa Modulus of elasticity in tension
ν - Poisson’s ratio
I mm4 Moment of inertia
G J m-2
Strain energy release rate
Gc J m-2
Critical fracture energy
U J Elastic energy
a mm Estimated crack length
aD mm Measured crack length
F N Applied force
C mm/N Compliance
M Nm Bending moment
x,y,z - Cartesian coordinates
xi mm Strain gauge position
b mm Width
Δ mm Deflection
h mm Flexible plate thickness
H mm Rigid plate thickness
e mm Adhesive layer thickness
l, L mm Length
ε - Unit elongation, strain
R mm Curvature of bended beam
δa mm Crack depth
k GPa Elastic foundation rigidity
v mm h-1
Crack speed
f - Surface fraction
λ m-1
Wave number
A,B,C,D Calculus constants
ω - λa
η, - SBT/Winkler energy release rate ratio, II/I stress mixity ratio
θ - SBT/Winkler strain ratio
γ J m-2
Environmental dependent fracture energy
N N Normal force
ρ mm Plate curvature
V N Shear force
η MPa Shear stress
κ - III/I stress mixity ratio
ζ MPa Normal stress
Michał K. Budzik
9
INTRODUCTION
Adhesive bonding is one of the most innovative and fast developing processes of joining
structural elements. In applications such as primary aircraft structures or automobile elements
adhesive bonding competes with traditional bolting, riveting or welding. The advantages of
adhesive bonding include high strength/weight ratio, possibility to join any combination of
materials, high corrosion resistance. Some drawbacks exist such as great sensitivity to the
fabrication process, limitations on the upper continuous operating temperature (ca. 300oC),
degradation of properties during environmental (hot/wet) exposure. The success of adhesive
bonding depends on numerous factors which include the theoretical knowledge about the
adhesion (adhesion science), mechanics as well as experience in fabrication of the joints.
Adhesion science is inherently an interdisciplinary field requiring fundamental understanding
of mechanics, physics, chemistry, surfaces and materials. Adhesion as it is now interpreted
refers to a complex set of inter-connected phenomena that is far from being completely
understood. D. H. Kaelble [1] defines adhesion (lat. adhaesio-adhering) as the phenomena
causing two boundary layers of two different bodies to be held together by their attraction
forces. In general adhesion includes the set of mechanisms that allow two components
(substrates, adherends) to be held together by a third component, the adhesive or simply, glue.
Free Encyclopaedia defines the adhesive or glue as: a compound in a liquid or semi-liquid
state that adheres or bonds items together. In addition, the adhesive should assure the
required properties (strength, durability etc.) over the designed service life of the bonded
system. Modern structural adhesives have been intensely developed for over 80 years now.
Primarily they were developed for reasons of war as a fast way to produce and to bond
aircraft. Nowadays due to the need to bond modern metals, polymer and composite materials.
The development of contemporary technology and industry is closely related to the creation of
new polymeric materials, among which adhesives are playing an increasing role. Their
production is being increased at higher rates than that of other polymeric materials. Such
enhanced interest in adhesives can be attributed to several factors [2]:
1. Modern technology employs new types of materials that cannot be joined by means of
traditional mechanical methods such as welds, rivets, screws, and bolts. These materials
include different types of ceramics, glass ceramics, alloys, composites etc.
2. Newly developed adhesives characterized by high strength, heat resistance, and
noncombustibility can meet the requirements of advanced technologies.
3. Adhesion is frequently the most effective way of joining very different materials using
relatively simple equipment. The range of materials that can be cemented is practically
unlimited.
4. Application of adhesives results in the improved characteristics of the article produced,
such as improved strength.
FRACTURE IN ASYMMETRIC BONDED JOINTS
10
In order to design properly bonded joint, thus assure high strength and durability of the
bonded joint, it is necessary to use appropriate testing procedures. Simple static tests do not
give sufficient information about life prediction, long time resistance, durability and
endurance of the bonded structures such as in aircraft. Fracture mechanics tests, especially the
Mode I test, are more appropriate since cleavage stress is considered as the fracture initiator.
This thesis has focused on new, reliable, mode I fracture mechanics based testing methods for
adhesive bonded joints characterization.
Although fracture mechanics is a relatively young branch of mechanics, numerous mechanical
tests have been introduced to interpret fracture behaviour of adhesive joints [3]. Most of them
are based on symmetric joint geometry. Asymmetric joints were studied only a few times
mostly in terms of physical and mathematical modelling. These studies concerned mode I/II
[4,5,6]. Mode I/III mixity conditions were reported rarely [7]. None of the test methods were
normalized or standardized. Accordingly there was the space for developing new procedures.
Furthermore the results obtained using available methods are not reliable [8] and often do not
allow continuous measurements over long times [9]. The existing models usually ignore the
adhesive and treat bonded joints as two joined materials without an intermediate layer, which
allows errors especially when the test is performed for a relatively soft adhesive, at elevated
temperature, or in a changing environment. Finally, neither crack propagation speed nor
fracture energy can be precisely estimated using the existing methods.
The present thesis has been divided into four chapters:
Chapter 1 presents the background information about the interdisciplinary science of adhesion
and the state of art in this field.
Chapter 2 outlines the experimental techniques and materials employed.
Chapter 3 describes the novel tests with physical and mathematical analysis and reports the
experimental results as well as a discussion of these.
Chapter 4 considers modelling of the asymmetric joint using a Finite Element Method so as to
estimate stress state in the joint.
The closing chapters gives conclusions and suggests future work perspectives.
Michał K. Budzik
11
Chapter 1. BIBLIOGRAPHY
1.1. Historical perspective of adhesive bonding
Over all known joining methods, adhesive bonding seems to have the longest history. The
first discovered adhesive occurs in central Italy on spear stone flakes glued to wood with
birch-bark-tar around 200,000 B.C. The use of compound glues to haft stone spears into wood
dates back to round 70,000 B.C. Evidence for this has been found in Sibudu Cave, South
Africa and the compound glues used were made from plant gum and red ochre. The Tyrolean
Iceman had weapons fixed together with the aid of glue. 6000-year-old ceramics show
evidence of adhesives based upon animal glues made by rendering animal products such as
horse teeth [10]. Archaeologists have also uncovered statues from Babylonian temples that
have ivory eyeballs glued into eye sockets. This tar-like glue has been known for almost 6000
years. Studying burial sites of prehistoric tribes dating to about 4000 B.C. foodstuffs buried
with the deceased in broken pottery vessels that had been repaired with sticky resins from tree
sap have been found. Egyptian carvings dating back 3300 years depict the gluing of a thin
piece of veneer to what appears to be a plank of sycamore. Flour paste was used to bond
together papyrus fibres that were then used as fabrics. Beeswax, tree pitches and bitumen
were used as protective coatings and adhesives. Egg whites were used to bind manuscripts at
one time and wooden objects were bonded with glues made from fish, horn, and cheese. The
period of time between 1500-1000 B.C. gave further proof that glue had become a method of
assembly. Paintings and murals showed details of wood gluing operations. A casket removed
from the tomb of King Tutankhamen shows the use of glue in its construction. Another
notable name in history who may owe his notoriety, at least in part to adhesives, is Genghis
Khan. Around the year 1000 A.D., Genghis Khan overcame all attackers because of the
exceptional power and range of the weaponry his men carried. Bows were made from
laminated lemon wood and bullhorn bonded with an adhesive whose formulation has been
lost to antiquity. During the 18th century, the technology of animal and fish glues advanced.
In the 19th century, rubber and nitrocellulose based cements were introduced. By 1900, the
U.S. had a number of factories producing glue from the aforementioned bases. The Industrial
Revolution caused an explosion in technical breakthroughs which resulted in new materials
becoming available for use in formulating adhesives. The first plastic polymer to be
synthesized was cellulose nitrate, a thermoplastic material derived from the cellulose of wood.
Its first use was in the manufacture of billiard balls, which had been made of ivory. The era of
plastics began with the introduction of bakelite phenolic, a thermoset plastic, in 1910. Within
a year adhesives using phenolic resin were on the market. The 1920´s, 30´s and 40´s saw
many new plastics and rubbers produced synthetically, many out of urgent necessity
developed during World War II. Although adhesives have been known for a very long time,
most of the technology of adhesives has been developed in the last 100 years. Nowadays,
more than 2,300,000 tonnes of adhesives are produced and used in Europe each year and this
volume is on the increase. Adhesive manufacturers offer more than 250,000 different products
for the most diverse applications – and these products are customised for virtually every
purpose [11]. This is important, because each adhesive must satisfy different requirements
depending on the application. The size of the 2007 total world market for adhesives is 10.5
FRACTURE IN ASYMMETRIC BONDED JOINTS
12
million metric tons. Its value was estimated at about $US 36 billion after allowing for various
currency fluctuations. The global market leaders are North America and West Europe regions
(Fig.1.1).
Fig.1.1. Global adhesive industry by region, 2007 share demand[11].
Global adhesive demand is still dominated by paper industries (Fig.1.2). The fastest
increasing submarkets are those of sealants and particularly structural adhesives.
Fig.1.2. World Adhesive (including sealants) demand in 2007[2].
Structural adhesives market is nowadays worth ca. $US 3 billion and is developing with 5.4%
year rate [2] with modern urethane and epoxy based adhesives are playing dominant role (see
Fig.1.3).
Michał K. Budzik
13
Fig.1.3. Structural adhesive market in 2002[12].
Structural bonding may offer many advantages over more traditional joining methods. A short
summary of the features of the common joining methods can be found in the Table 1.1.
Table 1.1. General comparison of joining methods[13].
WELDING BRAZING AND
SOLDERING
MECHANICAL
FASTENING
ADHESIVE
BONDING
JOINT FEATURES
PERMANENCE Permanent joints Usually
permanent
Threaded
fasteners permit
disassembly
Permanent joints
STRESS
DISTRIBUTION
Local stress
points in structure
Fairly good stress
distribution
Points of high
stress at fasteners
Good uniform
stress distribution
over joint area
APPEARANCE Usually
acceptable. Some
dressing
necessary for
smooth surfaces
Good Surface
discontinuities
sometimes
unacceptable
Joint almost
invisible
MATERIALS
JOINED
Generally limited
to similar material
groups
Some capability
of joining
dissimilar metals
Most forms and
combinations of
materials can be
fastened
Ideal for joining
dissimilar
materials
TEMPERATURE
RESISTANCE
Very high Limited by filler
metal
High Poor to elevated
temperature
MECHANICAL
RESISTANCE
Special provision
often necessary to
enhance fatigue
resistance
Good resistance
to vibration
Special provision
for fatigue
Excellent fatigue
properties.
Electrical
resistance
reduces corrosion
FRACTURE IN ASYMMETRIC BONDED JOINTS
14
Since this thesis is mainly devoted to load carrying structures and the methodology of their
testing, only structural bonding applications will be here presented.
1.2. Adhesive bonding applications
The biggest advantage of adhesive bonding is the possibility to join many materials without
affecting their properties. This allows bonding to be used in almost any application. The
industrial sectors which employs structural adhesive bonding include aeronautical, aerospace,
automotive, marine and off shore, construction, medical and sports.
Aeronautical and aerospace applications
Adhesives and aircraft have a long and interesting joint history. Even though flying vehicles
have progressed from glorified kites to commercial jet transports, supersonic missiles and
space vehicles, adhesively bonded structure has been crucial to virtually every one. Both
primary structure, which carries primary flight loads and failure of which could result in loss
of vehicle, and secondary structure are bonded. The military has historically led the way in the
development and application of adhesive bonding on aircraft. The earliest structural adhesive
applications were made during the First World War for bonding the wooden frames aircraft
(of biplanes), were strength was adequate but, by today’s standards, moisture resistance was
poor. This practice continues today, primarily with bombers, fighter and attack aircraft where
weight is a critical consideration, but also with support craft such as reconnaissance aircraft
and freighters. Much more so than the commercial world, military aviation is concerned with
aircraft performance. Because of this continued emphasis on adhesive bonding technology
development over the years, the airframes of modem front-line aircraft such as the B-2
bomber, the F-117, F-22, F-35, or Swedish JAS Grippen (Fig.1.4) fighters are largely
structurally bonded advanced composites [14].
Fig.1.4. JAS Grippen bonded primary elements[15].
Bonding
Bonding
Bonding
Bonding
Bonding
Michał K. Budzik
15
In addition, the use of adhesives prevents corrosion processes when different materials have
to be combined. Finally, due to uniform, plane load transfer through the adhesive, layer notch
sensitivity is reduced. The use of bonding also provides high potential for variation in styling
due to the possibility of combining different materials. In modern military and civil aircrafts,
as much as 50% of the airframe may be carbon fibre reinforced polymers (CFRP), with
adhesives being used for primary structural bonding [14].
Moreover, the average age of military and civilian aircraft is growing older at a fast pace. As
aircraft become older and accumulate more flight hours, the tendency they have to develop
corrosion problems, fatigue cracking, overload cracking etc. increases. When today’s aircraft
reach the end of their service life, fatigue cracks are found to have developed along rivet holes
and other highly stressed regions of the aircraft [16]. In order to extend the life of these
aircraft, repairs have been made to arrest these cracks [17]. Composite doublers or repair
patches provide an innovative repair technique, which can enhance the way aircraft are
maintained.
The very high cost of boosting spacecraft structure into orbit makes it cost-effective to spend
significant resources to save weight, thus is driving force behind this interest in bonding.
Exotic materials and processes that are too expensive for use on commercial aircraft are
commonplace in the space vehicle industry. Typical requirements for adhesively bonded
structure for space applications vary widely and differ substantially from those for
atmospheric vehicles. Because of widespread use of cryogenic rocket fuels, adhesives near
tank structure must maintain adequate properties at very low temperatures. At the other
extreme, adhesives have been used to bond ablative or insulative heat shields to the bottom of
re-entry vehicles since the advent of manned space flight [19]. Elevated temperature cure
epoxy film and paste adhesives supplanted the urethanes are standard today in areas of low to
moderately high temperature exposure because of their ease of fabrication and high strength.
Other adhesives such as silicone elastomers, cyanoacrylates and room-temperature epoxies
are used to bond many non-structural joints.
Automotive applications
Adhesives have been employed in the automotive industry since its beginnings, with the use
of natural resins to bond wood and fabric bodies [20]. Recent developments in synthetic resin
technology have resulted in a very wide range of adhesive materials available to the design
engineer. The drivers for the automotive industry are lightweight structures, use of mixed
materials, long term performance, crash performance and also styling and design. Since the
adhesive can improve the stiffness and strength of a joint the weight can be reduced. Adhesive
bonding can furthermore allow the realisation of combining different structural materials such
as FRP, metals, glasses and ceramics. It is quite clear that many parts of different materials
have to be brought together through bonding, sometimes together with rivets. In many
situations this is preferred to welding. Similar to aerospace bonding, galvanic processes can
be delayed or prevented when adhesives are used, in particular when different materials are
joined together. Improvement of crash performance is possible by the use of substrates and
FRACTURE IN ASYMMETRIC BONDED JOINTS
16
adhesives with a high potential of energy absorption. Finally, diversity of styling and design
are possible due the possibility of combining different materials and components and joining
them together by bonding. Adhesives for automotive and industrial bonding are modified
acrylics/methacrylates which provide high strength and elongation properties and also bond to
thermoplastics. Other adhesives are polyurethanes, which are tough and have a high abrasion
resistance and good adhesion at low temperatures. Silicones are also used for bonding to
glass, plastics and other materials.
Marine and off-shore applications
Casein and then formaldehyde resin compositions have been used as adhesives and gap fillers
in wooden boat construction for many years [20]. Many GFRP-hulled boats, both naval and
civilian, now relay significantly on resins for laminating, stiffening, the fabrication of
sandwich panels, and for bonding attachments. Traditionally the shipbuilding industry uses
welding as the primary process for joining the different structural parts in a ship. It is well
known however that this process results in induced stresses during the fabrication stage which
in turn lead to distortion in the shape of structural components and indeed, the ship itself.
Considerable effort has thus to be expended to mitigate these weld-induced distortions leading
to increased production costs. One additional problem, in the case of aluminium, is the
significant reduction in the fatigue load capacity in welded structures [21]. As a consequence,
either the structural topology has to be designed to cope with increased stress levels or the
scantlings of the structure have to be enhanced. In either case, there is an increase in the
weight of the structure. Since structural weight needs to be minimised, especially in high
speed, high performance ships, there is a need to investigate alternative joining techniques for
aluminium structures; adhesive bonding is one of them [22,23]. Adhesive bonding offers the
opportunity to replace welding of steel structures, to reduce distortion, effectively eliminate
residual stress and to improve fatigue performance when compared to welded connections.
Avoidance of hot-work leads to safer construction practices in hazardous environments.
Adhesive bonding of composites provides well-distributed loading and maximises the
utilisation of the adherent materials. Typical adhesives in the marine industry are polyesters
which are less expensive than epoxy and are widely used in other industrial applications.
Polyester, however, is chemically weaker than epoxies and experiences a high degree of
shrinkage. Vinyl ester, which provides higher strength, modulus of elasticity and elongation
than polyesters is still less expensive than epoxy are often preferable to polyesters. For
bonding to metallic parts epoxies are typically used. In addition modern composites materials
are also very attractive for marine industries. Whole composite body boats are already in use,
like luxury CRN 128 Sima [24] boat full FRP body boat joined with adhesive bonding.
Railway applications
In rail vehicle manufacture, new components such as those made from glass-fibre reinforced
plastic (GFRP) have resulted in enormous weight reduction and have transformed the
production process. For example ADtranz (part of Bombardier Transportation company)
Michał K. Budzik
17
regional trains are manufactured using a frame structure consisting of tensile and compression
struts in a triangular arrangement. This is the lightest design for bearing structures. Due to the
frame structure, it is not necessary to weld on metal sheets to take up the shear stresses. There
are only tensile and compressive forces in the struts. In order to enclose the frame, large outer
skin elements are bonded to the supporting metal structure using moisture curing 1-C
polyurethane adhesives [25]. No complex straightening and filling work is required, as is the
case when welding is used to attach the metallic outer skin elements. The GFRP elements that
are used for the outer skin have a sandwich structure and hence provide good thermal
insulation and have very good acoustic properties. As no heat distortion occurs, contrary to
when welding, the outer skin gives the rail vehicle a ready painted/lacquered, level, smooth
and flush outer surface with no additional work having to be undertaken. Joining a metallic
framework structure with components made of fibre reinforced composite plastic can only be
economically and reliably achieved using bonding technology. In order to compensate the
different linear deformations of the structure and outer skin when they are exposed to heat, a
highly elastic, thick film of adhesive is used. The resulting bonds have high strengths, even
when subjected to repeated temperature fluctuations in the range between -40°C and +80°C
and under impact-like stress. They also have very good resistance to aging, even in moist
environments. The thickness of the bonded joint varies between 4 and 18 mm. This depends
on the size of the components, which can be up to 8.5 metres long and therefore subject the
bonded joint to very high shearing deformation. Another advantage of this type of structure is
the high damping and the positive effect on the dynamic operating characteristics and ride
comfort. Finally bonded trains are about 25% lighter than comparable conventional trains, and
they also have improved ride characteristics, lower production costs and reduced energy
requirements for operation.
Construction applications
The adhesives are used in construction industry in repair and strengthening of existing
structures and new built structures.
a) Repair and strengthening of existing structures using bonding techniques is mainly
associated with concrete structures. During the 1950’s and 1960’s an enormous amount of
new constructions were built and, as these structures age, many faults have become evident.
The repair substrate may be ordinary concrete or polymer concrete, with or without
reinforcement. If increased structural capacity is needed then external plate bonding can be an
alternative. During the 1970’s and up to the end of the 1990’s, steel plate bonding was not
unusual, however in the last decade, the use of advanced composites for external
strengthening has become quite common (see Fig.1.5) [26,27]. Also a great number of models
for debonding have been presented [28-32].
b) In new-built structures the adhesives are often used for filling voids or gaps and to fasten
secondary, non-load carrying elements. Adhesives are also used to fasten bridge bearing and
expansion joint nosing. On steel decks, skid resistant surfacing materials in the form of small
gravel-filled epoxy or polyester resins are often used. There are also examples where epoxy
FRACTURE IN ASYMMETRIC BONDED JOINTS
18
resin systems have been used to bond precast concrete slab units directly to a steel girder
surface for a steel-concrete composite bridge (German Bonded Bridge, 1979). However, with
the increased use of new-built FRP structures requiring bonding for building and civil
purposes the situation may change. The example of a FRP structures is presented in Fig.1.6.
The Zaragoza Bridge Pavilion have all structural joints made using epoxy adhesive.
Fig.1.5. Bridge repaired and strengthened with CFRP patch (left).
Fig.1.6. Zaragoza Bridge Pavilion. Bridge build totally from cement reinforced with GFRP
(right).
Medical applications
The use of the adhesives in medical applications was for a long time restricted to the
manufacture of self-adhesive bandages (plasters, self-adhesive strips of fabric, etc.).
Adhesives are today employed in diverse areas of medicine, replacing traditional methods
with friendlier processes. For example stitches can be avoided by applying special
cyanoacrylate adhesives to quickly close skin wounds. An advantage here is that the whole
wound can be covered, so largely suppressing secondary bleeding and the risk of infection.
The use of methacrylate based adhesives has been a great success in orthopaedics for
anchoring hip socket implants to the bone. Also hip and knee implants anchored using
adhesive are in 90% of cases functional for about 15 years [25] (Fig.1.7).
Fig.1.7. One of the examples of structural bonding applications in medicine[25].
Michał K. Budzik
19
In dentistry, fillings based on UV curing acrylates have largely replaced traditional filling
materials such as amalgam. The products have a long open time (the period during which they
can be used after mixing) and bond in just a minute or so when exposed to UV light. Also
based on ceramics adhesives are used to repair broken tooth etc. (Fig.1.8) [33].
Fig.1.8. Structural adhesive application in dentistry.
Other developing markets
Sport: bonded bicycle frames, kayaks, rowing boats, rows, pole vault poles, golf clubs,
tennis rockets etc.
Nuclear energy application: radioactive waste tanks and containers, strengthening and
radiation resistance improvement of cement tanks for radioactive water
Oil and gas industry: oil tanks reconstructions, corrosion resistant barriers, repair of
pipelines, pipeline strengthening
Bonding of silicon in photovoltaic panels
1.3. Structural adhesives
Structural adhesives are based upon resins composition that polymerize to give high-modulus,
high-strength adhesive so that a load-bearing joint is formed [34]. The first structural adhesive
successfully applied was in the aircraft industry for bonding aluminium to wood. It was
polyvinyl formal composition with a phenoloformaldehyde resin. There is rarely a situation in
which a single adhesive is unique for a task. This is because adhesive are versatile and
selection does not usually, depend upon one single property, but rather balance of several
properties which more than one adhesive may be able to meet. Usually the first decision
regarding adhesive bonding is to choose the basic chemical type of adhesive, nowadays for
structural applications we can choose from:
Acrylic adhesives
Epoxy adhesives
Bismaleimide adhesives
Polyimide adhesives
FRACTURE IN ASYMMETRIC BONDED JOINTS
20
Polyurethane adhesives
Phenolic adhesives
Silicone adhesives
Acrylic adhesives
Acrylic adhesives were developed in late 1960’s in Germany as an outgrowth of
poly(methylmethacrylate) chemistry [35]. Acrylic adhesives are nowadays large class of
specifically designed products made to meet the needs of industry in the assembly of a wide
variety of components. These adhesives are solvent-free reactive engineering adhesives and
include cyanoacrylic, anaerobic and modified acrylic adhesives [34]. The basic acrylic
monomers or oligomers contain unsaturated double bonds (vinyl groups), and consequently
cure by addition polymerization involving free-radical reaction. Free-radical producing
compounds such as peroxides, peracetic acids, and sulfones are added to acrylic resins to
initiate polymerization. Free-radical polymerization of acrylics may also be induced by UV or
visible light. These UV-curing adhesives, most of which are based on acrylic or modified
acrylic resins are of increasing commercial importance today for the rapid bonding of
electronic devices, fiber optics etc. Reactive methacrylate monomers include
methylmethacrylate, diethyleneglycol dimethacrylate and butyl methacrylate. The
polymerization reaction for the basic monomer is depicted in Fig.1.9.
Fig.1.9. Basic acrylic reaction[36].
Cyanoacrylates are one-part, highly polar thermoplastic polymers. The resin monomers
(Fig.1.10) cure in seconds when in contact with a weak base such as the moisture that is
present on most of the surfaces. Many cyanoacrylate-adhesive formulations are commercially
available, but not widely used in electronics assembly because of their poor resistance to
solvents and moisture at elevated temperatures (>70°C). Cyanoacrylates have relatively low
impact and peel strengths and may be brittle unless toughened by the addition of elastomeric
resins. Currently available liquid acrylate and methacrylate adhesives can be seen as logical
developments of the anaerobic machinery adhesives against corrosion as well as the
considerable economic advantages gained by not having to maintain very close tolerances on
the machined parts.
Michał K. Budzik
21
Fig.1.10. Cyanoacrylate monomers structure, with R is usually alkyl group[35].
Although these adhesives are still by a margin more expensive (10 to 100 times when
compared to older bonding technologies) they achieve enormous commercial success thanks
to specific benefits they offer. Major features of acrylic adhesives include [36]:
Adhesion to a variety of substrates
Water resistance
Durability
Flexibility when modified
Good low temperature properties
Excellent optical properties
Low toxicity.
Epoxy adhesives
Epoxy or epoxide resins are a group of reactive compounds that are characterized by the
presence of the epoxy group, which in the simplest epoxy compound have structure of
ethylene oxide, oxirane group (Fig.1.11):
Fig.1.11. Ethylene oxide (oxirane).
They are capable to react with suitable hardeners to form cross-linked matrices of great
strength and with excellent adhesion to a wide range of substrates. This make them ideally
suited to adhesive applications in which high strength under adverse condition is prerequisite.
Work on epoxy resins started in the mid 1920’s when in 1926 Eisleb noted the reaction of
epichlorohydrin with secondary amines and subsequent dehyrogenation with caustic [37]. The
first commercially useful epoxy resins appeared during World War II, when in 1939 Greenlee
explored the epichlorohydrin-bisphenol A synthesis route for the production of new resins
without caustic-sensitive ester linkages. Today this resins, in a range of molecular weights,
constitute the majority of all epoxy resins used. By contrast, hardeners come in a variety of
shapes and sizes. Epichlorohydrin is capable of reacting with hydroxyl (-OH) groups with the
FRACTURE IN ASYMMETRIC BONDED JOINTS
22
elimination of hydrochloric acid. The most widely used epoxy resins are the family of
products produced by the reaction between epichlorohydrine and bisphenol A, to give
diglycidylethere of bisphenol A (DGEBA), shown in Fig.1.12. Epoxy resins react with
hardeners in stoichiometric quantities. Knowledge of the number of reactive sites is needed in
order to calculate correct ratios. For the resin this is given by the epoxide equivalent weight
(EEW), which is the quantity of resin required to yield one epoxy group. For a DGEBA type
in which n=0, the molecular weight is 340. Since we have two epoxy group per molecule,
therefore, EEW=170. Typically, pure epoxy resins available for adhesive applications are in
the range 180<EEW<310 (while for paints EEW=2000 and more!). The viscosity of epoxy
resin is primarily dependent on molecular weight. Even at low molecular weight, viscosity is
typical in excess of 6000cP, while at 190 EEW is usually around 12000cP. Stoichiometric
ratios can be calculated similarly for hardeners, from which amines are of the major interest.
Example of amine curing mechanism of epoxy is illustrated in Fig.1.13.
Fig.1.12. Bisphenol A-epichlorohydrine reaction[38].
Fig.1.13. Amine – curing mechanism of epoxies: a) initial step, formation of a secondary
amine and more hydroxyl groups; b) formation of tertiary amine; c) continued crosslinking
through reaction of hydroxyl groups[38].
Michał K. Budzik
23
Epoxy adhesives offer[36]:
Good adhesion to many surfaces
A range of mechanical properties, depending on formulation and curing
A wide range of cure characteristics (from RT to 180oC)
Low shrinkage
No by-products evolved during curing
Range of available adhesive forms (film, paste one or two component).
The epoxy adhesive are the most extensively used structural adhesives, particularly in the
transport industry. In Table 1.2 fracture resistance of the epoxy and toughened epoxy resins is
compared with different material. It has to be noted that modified epoxies (e.g. with rubber)
are at least 10 time stronger than pure epoxy resins.
Table 1.2. Values of measured Critical Fracture Energies[39].
Material Critical Fracture Energy, GIc
(J m-2
)
Inorganic glasses <10
Unmodified epoxies 70 – 175
PMMA 350 – 800
Modified epoxies 1750 – 3500
Aluminium/nylon/epoxy adhesive 7000 – 8800
Metals >17000
Bis-maleimide adhesives
In the narrow sense, bis-maleimide resin means the thermosetting resin composed of the bis-
maleimide of methylene dianiline (BMI, bis(4-maleimidopheny1)-methane) and methylene
dianiline (MDA, bis(4-aminopheny1)methane) (Fig.1.14).
Fig.1.14. The chemical structure of BMI and MDA[40].
FRACTURE IN ASYMMETRIC BONDED JOINTS
24
Because of the addition mechanism, the resin is cured without elimination, which is a
characteristic of this resin. Bis-maleimide resin is used as a thermally stable matrix up to
204°C in which typical epoxy resins may not normally be used. However, in spite of having
an imide structure, bis-maleimides are classified as being moderately thermally stable resins.
The aliphatic structure of the resin is not stable for long periods above 232°C. If a highly
aromatic thermally stable thermosetting resin is necessary, acetylene end-capped aromatic
imide-based oligomers should be used. We can conclude that bis-maleimide adhesives offer
[36]:
High temperature capability
Excellent electrical properties
No volatiles evolved during cure, which simplifies processing and reduces porosity
Poor peel resistance, owing to their stiffness. Reinforcing required.
Polyimide adhesives
Polyimides are formed from polyamic acid or polyamic ester precursors by heating to
temperatures as high as 400°C. Polyimide precursors are synthesized by reacting equimolar
amounts of aromatic diamines with aromatic dianhydrides forming polyamic acids or
polyamic esters as in Fig.1.15a. The prepolymer from pyromellitic dianydride (PMDA) and
4.4’-oxydianiline (ODA) has been the most widely studied and used [41]. The precursors are
available as solvent solutions (generally, N-methyl-2-pyrrolidone) that can then be formulated
as adhesives, coatings, or performs by adding fillers and other additives. Curing is not the
traditional crosslinking or chain propagation using curing agents, but occurs by heating to
eliminate water and close the imide rings along the chain forming the very stable polyimide
structure (Fig.1.15b).
Fig.1.15. a) Preparation of polyamide precursor, b) curing of polyamides by imidization[43].
Michał K. Budzik
25
The process of ring closure is known as imidization and requires step curing, which, for
silver-filled paste adhesives, is typically 30 minutes at 150°C, followed by 30 minutes at
275°C [42]. In some cases, however, step cures up to 400°C are necessary to achieve
complete imidization, an example of which is: 15 minutes at 135°C, followed by 30 minutes
at 300°C, and finally 10 minutes at 400°C. The initial temperature exposure of 135°C–150°C
assures the removal of the N-methyl- 2-pyrrolidone (NMP), a solvent that is present in almost
all polyimide precursor formulations.
Main features of polyimide adhesives are include [36]:
High-temperature capability (up to 300oC)
Excellent electrical properties, e.g. radome applications
Evolution of volatile materials during cure so extraction or high processing are
needed.
Polyurethane adhesives
The development of polyurethane adhesives can be traced back to pioneering efforts of Otto
Bayer and co-workers. Bayer extended the chemistry of polyurethanes (initiated in 1937) into
the realm of adhesives by combining polyester polyols with di- and polyisocyanates (about
1940) [44]. He found that these products made excellent adhesive for bonding elastomers to
fibers and metals. Early commercial applications included life rafts, vests, airplanes, tires and
tanks. Being thermoplastic, polyurethanes are also easier to rework than epoxies.
Polyurethanes are distinguished from other polymer types in containing the repeating urethane
group throughout its structure (Fig.1.16).
Fig.1.16. Urethane group.
Typical polyurethane adhesive may contain, in addition to the urethane linkages, aliphatic and
aromatic hydrocarbons, esters, ethers, amides, urea, and allophanate groups. Polyurethanes
are formed by the addition reaction of diisocyanates or polyisocyanates with polyols through a
step-growth polymerization mechanism (Fig1.17). Accordingly, polyurethanes may be either
thermoplastic or thermoset, depending on the functionality of the monomers. Urethane
adhesives are classified as one-component or two-component adhesives. Each category
includes several different types of adhesives. Urethane structural adhesive are two-
component. These adhesives may be rigid plastics similar in modulus to standard epoxy
adhesives, with glass transition temperatures of the cured adhesive being approximately 60°C.
Two-component urethane adhesives are used to bond sheet molding compound (SMC) panels
for automotive OEM (original equipment manufacturers) and aftermarket applications. Two-
part urethanes are used as laminating adhesives in the RV (recreational vehicle) industry.
FRACTURE IN ASYMMETRIC BONDED JOINTS
26
Fig.1.17. Basic polyurethane polymerization reaction[45].
One of the polyurethane features is good adhesion to the numerous substrates, this is due to
the following reasons:
Effective wetting of surfaces [46]
They readily form hydrogen bonds to the substrate
Their small molecular size allows them to permeate porous substrates
They form covalent bond with surfaces that have active hydrogen (Fig.1.18).
Fig.1.18. Typical mechanism for a urethane adhesive bonding covalently to a polar
surface[45].
Main polyurethane adhesive features are:
Good adhesion to variety of substrates
Good chemical resistance to solvents
Tough, flexible bonds
Available in many forms: liquids, dispersions, films, powders etc.
PREPOLYMER
H H H
NCO NCO NCO
PREPOLYMER
NH NH NH
O=C O=C O=C
Michał K. Budzik
27
Phenolic adhesives
Phenolic resins were the first totally synthetic plastics invented. They history dates back to
1853 with Gerhardt’s observations of insoluble resin formation while dehydrating sodium
salicylate [47]. The reasons for the long-term, commercial robustness of phenolic technology
include low cost, versatility, heat and flame resistance, durability, strength and stiffness, low
toxicity, and ease of processing. Adolph Baeyer is credited with the first recognition of the
general nature of the reaction between phenols and aldehydes in 1872 [47-50]. Prior to 1890,
formaldehyde was not commercially available [47]. Thus the first phenol-formaldehyde resins
were made using formaldehyde equivalents such as methylene diacetate or methylal [1,20].
The first true phenol-formaldehyde resin was made by Kleeberg at the direction of Emil
Fisher in 1891 [47,51]. The condensation polymerization reaction is shown in Fig.1.19.
Fig.1.19. Polymerization of phenol with excess of formaldehyde[52].
Phenolic resins are adaptable to many applications. The list is very long, however, the major
uses are wood binders, glass insulation binders, moulding compounds, laminates, foundry
binders, coatings, friction linings, abrasives, and oil well propants [53-58]. They have found
their way into a number of new, high technology uses such as rocket motor wear parts,
military armour, sports equipment, photoresists for computer chip manufacture, epoxy
crosslinkers, circuit board binders and microchip module packaging. There has been a
renewal of interest in the use of phenolics for aircraft construction because of their excellent
flame resistance and low smoke generating properties [59]. The formation of a phenolic resin
is often formally separated into two steps, though it probably should be three. In the three-step
model, the first step is activation of the phenol or aldehyde. The second step is methylolation,
and the third is condensation or chain extension. In addition to the clarity provided by the
formalism, these steps are also generally separated in practice to provide maximum control of
exothermic behaviour, with the strategy being to separate the exotherm from each step from
that of the others as much as possible. Main features of phenolic adhesives are:
High temperature curing (often treated as advantage – no curing in ambient
conditions)
Very brittle, requires compounded with other materials (polychloropropylen,
polyvinyl butryl, epoxides etc.)
FRACTURE IN ASYMMETRIC BONDED JOINTS
28
High mechanic strength
Good solvent and water resistance
High thermal stability
Flame retardant (often used as a compound to epoxides, rubbers etc.).
Silicone adhesives
Silicones are a unique class of polymers due to their semi-organic molecular structure. Instead
of the normal carbon-to-carbon backbone structure of most polymers, silicones have a silicon-
to-oxygen structure that gives them advantages of very high thermal stabilities (up to 300°C,
in some cases), flexibility at subzero temperatures (-80°C), and excellent electrical properties
under both extreme conditions. The -Si-O-Si-O- backbone of silicones is referred to as
siloxane. The silicon atoms may be linked to a wide variety of aliphatic or aromatic groups, as
shown in Fig.1.20, where the radical R groups are commonly methyl (-CH3), phenyl (C6H5-),
allyl (-CH2-CH=CH2) or vinyl (-CH=CH2).
Fig.1.20. General structure for a linear silicon polymer[43].
A silicone adhesive is known to be among the most durable and flexible. There are not a lot of
heavy-duty adhesives that can serve in the same capacity as a silicone adhesive, and this is
because they come in a range of styles and are capable of joining materials like plastic, metal
and glass. Accordingly, silicone adhesives are water and high temperature resistant.
Interestingly, a silicone adhesive will also retain flexibility after curing, making it applicable
in ways other adhesives are not. This type of silicone adhesive appears quite frequently in the
building industry where it can be used to join materials that must face exposure to glaring
sunlight, freezing temperatures, water, wind and other issues that would cause traditional
adhesives to fail.
Main silicon adhesives features include:
Wide variety of possible viscosities
Good flexibility
Very good resistance to high and low temperature (-120oC – ~400
oC)
Resistance to UV and IR radiation
Resistance to oxidation.
Michał K. Budzik
29
Properties of common used structural are summarized in the Table 1.3. The variety of
properties, meting different requirements can be easily noted.
Table 1.3. Common properties of structural adhesives.
Adhesive
/Properties
CTE*
(μm/moC)
Tension
strength
(MPa)
Application
Temperature
(oC)
Young
modulus
(GPa)
Density
(kg/m3)
Acrylic 54-150 19-90 >115 0.9-4.5 0.98-1.22
Cyanoacrylate 2.5-50 >100 1.05-1.11
Epoxy 20-60 28-75 >300 2.5-5.5 1-2
Bis – Maleimiade 15-100 12 >500
Phenolic 30-45 35-63 166-205 5.5-11 <1.5
Silicone 54-300 1-30 >370 0.0006-
0.005
0.82-2.82
Urethane 1-510 0.9-29 >130 0.004-0.8 0.89-3.2
Polyamid <100 <200 <200 <10 <1.4
* Coefficient of Thermal Expansion
1.4. Theories of adhesion
An adhesive must do two things when applied to surfaces which are to be bonded. It must first
wet the surface, as manifested by spreading and making a contact angle approaching zero.
Secondly, it must harden to give a cohesively strong solid. When contact is intimated, van der
Waals forces are formed, but other intermolecular forces, such as chemical bonds, may be
formed at the time of contact or during the hardening process. If the adhesive can penetrate
the substrate before hardening then mechanical interlocking will contribute to the strength of
the joint. Molecular interdiffusion will occur when adhesive molecules would intertwine those
of substrate. Other approaches are also possible. To understand fully the basics of adhesive
bonding and to be able to conclude properly from any experiments or structure behaviour
when bonding joints appears, one must understand how adhesion is created. Of major concern
are:
Mechanical interlocking theory
Electronic theory
Weak Boundary Layer (WBL) theory
Physical adsorption theory
Diffusion theory
Chemical bonding theory.
Mechanical interlocking
The mechanical interlocking theory was first proposed by MacBain and Hopkins [60]. The
theory assumes mechanical keying, or interlocking of the adhesives into the cavities, pores
FRACTURE IN ASYMMETRIC BONDED JOINTS
30
and asperities of the solid surface to be the major factor in determining adhesive strength.
Borroff and Wake [61] presented one of the most consistent examples presenting validity of
this theory. They measured adhesion between rubber and textile fabrics, clearly proving that
penetration of the protruding fibres into rubber was determining parameter in this specific
joint. However the possibility of producing good adhesion between smooth, polished surfaces
leads to the remark that mechanical interlocking theory cannot be used as the universal. To
overcome this difficulty, primarily by Gent and Schultz [62], Wake [63] has proposed that the
effects of both mechanical and thermodynamic interfacial interactions could be taken into
account, as a multiplying factors to estimate proper adhesion strength, G:
(1.1)
where:
α – constant
MC – mechanical interlocking component
IC – interfacial interactions component.
Therefore, due to interlocking theory, high strength of adhesive joint is achieved by increasing
both surface morphology (MC) and physicochemical surface properties of substrate and
adhesive (see Table 1.4). Work done by Packham [64] emphasized the dominant role of the
mechanical interlocking mechanism on adhesion strength, although in another study Wake
deduced that IC component may become greater than MC [63].
Table 1.4. Surface topography influence on peel energy[64].
Surface topography of copper foil Peel energy
Description Diagrammatic representation kJ/m2
Flat
Flat + 0.3 μm dendrites
3 μm high angle pyramids
Nickel foil with club-headed
modular structures
0.66
0.7
1
2.3
Another important aspect of mechanical adhesion theory is surface roughness. When
roughness is increased the actual surface area is growing by developing. In the literature we
can find a study by Wenzel on this subject [65], which defined simple roughness factor, r in
the form:
Michał K. Budzik
31
(1.2)
where:
A - true surface area
A0 - real surface area.
Thus, for a perfect flat surface r=1, when consisting of hemispheres r=2 etc. This simple
factor can define work of adhesion, Wa’ for rough surface:
(1.3)
where γ is surface free energy in the boundary between s – solid, v – vapour and l – liquid
phases.
Electronic theory
The electronic theory was proposed by Deryaguin and co-workers [66]. Authors have
suggested that an electron transfer mechanism between the substrate and the adhesive, having
different electronic band structures, can occur to equalize Fermi levels. This can lead to
double electronic layer formation at the interface, which may contribute to the adhesive
strength. Therefore, the adhesive-substrate junction can be analyzed as a capacitor. Interfacial
separation of the system leads to increasing potential difference until discharge occurs. The
separation energy, Ge is thus related to the discharge potential, Ve:
(1.4)
h – discharge distance
εd – dielectric consant
One of the interesting aspects of these theory is that it allows adhesion variation with gas, and
pressure of the gas in which the measurement is performed. In fact, Deryaguin et al. have
measured, by means of peel test significant variation in separation energy when measured in
argon and air environment with different pressure, which stays in very good agreement with
theoretical predictions. However several other analysis [e.g. 63] have not confirmed these
results, leading to the conclusion that Deryaguin was rather causal.
Weak Boundary Layer (WBL) and interphase
It is now well known that the adhesive/adherend interphase can have different properties from
the bulk adhesive and adherend. The first approach to this problem is due to Bikerman [67],
who stated that the cohesive strength of a weak boundary layer (WBL) can always be
considered as the main factor in determining the level of adhesion, even when the failure
FRACTURE IN ASYMMETRIC BONDED JOINTS
32
appears to be interfacial. According to this assumption, adhesion energy, Ea, is always equal
to cohesive energy, Gc (WBL) of weaker interfacial layer. This theory is based on statistics,
proving that the fracture should never propagate only along adhesive/substrate interface for
pure statistical reasons and that cohesive failure near weaker interface is more favourable.
Formation of WBL is caused by many physical, physicochemical and chemical phenomena,
for example [68]:
The orientation of the chemical group or the overconcentration of chain ends to
minimize the free energy of the interface
Migration toward the interface of additives or low molecular weight fraction
The growth of transcrystalline structure, for example, when substrate acts as a
nucleating agent
Formation of pseudoglassy zone resulting from a reduction in chain mobility through
strong interactions with the substrate
Modification of thermodynamics and/or kinetics of the polymerization or cross-
linking reaction at the interface through preferential adsorption of species or catalytic
effects.
Physical adsorption theory
Physical adsorption theory [52] involves long range bonding with van der Waals forces but
also thermodynamic-spreading theory.
a) van der Waals forces
van der Waals forces, which occurs between all atoms and molecules when they are close to
each other, are the weakest of all intermolecular forces, but their strengths are adequate to
account for the strength of the adhesive joint. Physical adsorption involves van der Waals
forces across the interface. These involve attractions between permanent dipoles and induced
dipoles (Fig.1.21), and are of three types.
Epp is the potential energy, in a vacuum, of a pair of permanent dipoles separated by
distance r at their centres and is given by equation:
(1.5)
r – distance between dipole pair
μ1, μ2 – dipole moments
ε0 – permittivity of the vacuum
k – Boltzmann constant
T – absolute temperature.
Michał K. Budzik
33
Fig.1.21. van der Waals interactions[69].
If a non-polar molecule is close to a dipole, then the latter will induce a dipole (μi) in
the former. The induced-dipole moment is given by:
(1.6)
α – polarizability of the non-polar molecule
E – electric field
The potential energy of such interaction can be given by:
(1.7)
Instantaneous dipoles exist in non-polar molecules because of the fluctuating
distribution of electrons. These lead to attractive forces between molecules, without
which non-polar gases such as helium and argon would not be able to liquefy. The
potential energy of a pair of molecules, Eii is given by equation:
(1.8)
where:
α1, α2 – molecule polarizabilities
I1, I2 – ionization potentials.
FRACTURE IN ASYMMETRIC BONDED JOINTS
34
b) Thermodynamic (spreading) theory [68]
Thermodynamic theory in its major aspect is limited to solid-liquid systems. Adhesives must
effectively wet and completely contact the surfaces to assure a strong bond. The ability to wet
a surface, wettability, is related to the ease with which a liquid spreads on a solid surface and
is essential in maximizing coverage and minimizing voids in the bondline [70]. Wettability is
measured by the equilibrium contact angle, θ, which is defined by balancing surface-tension
forces in Young’s equation.
(1.9)
where:
γsv – surface energy at solid – vapour interface
γsl – surface energy at solid – liquid interface
γlv – surface energy at liquid - vapour interface.
The measurement of contact angle (Fig.1.22) is a means of investigating adhesion by physical
adsorption. These are the weakest forces contributing to adhesive bonds, but are quite
sufficient to make strong joints.
Fig.1.22. Wettability of adhesives and contact angle.
The thermodynamic work of adhesion, Wa, is by definition the free energy change per unit
area required to separate to infinity two surfaces initially in contact surfaces, which results in
creating new surfaces. Work of adhesion can be related to a solid and a liquid phase forming
an interface across which secondary forces are acting. Free energy is given by Dupré
equation. The reversible work of adhesion in an inert medium may be expressed by:
(1.10)
i.e.
(1.11)
We can easily combine the Young and Dupré equations, to yield Young-Dupré equation in the
form:
(1.12)
Michał K. Budzik
35
The spreading coefficient, called S [71,72] or sometimes (although badly) work of adhesion
(Wa) in equation above, shows that good wetting (θ<90o) is realized when Wa is greater than 0.
Good wetting requires also using liquid with high surface tension. As Wa decreases, θ
increases from 0o to 180
o and conditions go from partial wetting to non-wetting [73]. Based
on this equation, conditions of wetting versus non-wetting are as presented in the Table 1.5
[70], and illustrated schematically in Fig.1.23.
Fig.1.23. Schematic presentation of good and bad wetting.
Table 1.5. Wetting conditions.
Complete wetting, liquid spreads spontaneously, high surface attraction
Partial wetting
Non-wetting: liquid beads up
Diffusion theory
The diffusion theory of bonding is based on the assumption that the adhesion strength of
polymers themselves (autocohesion) or to each other is due to mutual diffusion
(interdiffusion) of macromolecules across the interface, thus creating an interphase (Fig.1.24).
Fig.1.24. Adhesive – substrate interdiffusion[75].
Such mechanism was studied in details by the Russian Voyutskii [74]. The author implied
that the macromolecular chains or chain segments are sufficiently mobile and mutually
soluble. This is of great importance for many adhesion problems, such as healing and welding
process. If interdiffusion occurs the joint strength should depend on different factors, such as
contact time, temperature, nature and molecular weight of polymers etc. The reason for
diffusion is thermodynamic potential difference of molecules of both substrates. Vasenin [76]
has developed, from Fick's first law, a quantitative model of diffusion model that correlates
the amount of material, w diffusing in given x direction across a plane of unit area to the
concentration gradient, ∂c/ ∂x and the time, t:
FRACTURE IN ASYMMETRIC BONDED JOINTS
36
(1.13)
with:
Df – diffusion coefficient.
To estimate the depth of penetration of the molecules that interdiffused into the junction
region during the time of contact, tc, Vasenin assumed that the variation of the diffusion
coefficient is constant and of form Ddtc-β
, therefore we can deduced the depth of penetration,
lp from:
(1.14)
where:
k - constant
Dd – constant characterizing the mobility of polymer chains
β – constant of order 0.5.
Finally, Vasenin assumed that the measured peel energy, G was proportional to both, the
depth penetration and the number of chains crossing the interface between the adhesive and
the substrate so thus:
(1.15)
where:
K – constant that depends on molecular characteristic of the polymer in contact.
N - Avogadro number
ρ - density
M – molecular weight of polymer.
Chemical bonding theory
The chemical bonding theory of adhesion involves the formation of covalent, ionic or
hydrogen bonds or Lewis acid-base interactions across the interface.
Covalent bonds: There is some evidence that covalent bonds are formed with silane
coupling agents [77], as well as adhesive containing isocyanate groups react with
active hydrogen atoms (such as hydroxyl group with wood or paper). In these two
examples Si-O and C-O bonds are formed. Another possibility is to react epoxide
adhesive with surface containing amine group, to create N-C bonds.
Michał K. Budzik
37
Ionic bonds: For ionic bonds the potential energy, E± of charge z1e and z2e separated
by distance r is given by equation:
(1.16)
εr – relative permittivity of the medium
ε0 – permittivity of vacuum.
Hydrogen bonds: The perfect example is bonding of water molecules contributed to
the attachment of stamps to envelopes where the adhesive (polyvinyl alcohol) and
paper (cellulose fibres) both contain –OH group. Hydrogen bonds are weak and they
are easily broken and readily formed. The bond consist of a hydrogen atom which is
bonded with to two other atoms, that is that two atoms are bridged by hydrogen. It
consist of one normal A-H bond and a longer H···B bond, forming A-H···B [78-80].
Lewis acid-base interactions: Conventional or Bronsted [81] acids are donors of
protons (hydrogen ions H+) and the base are protons acceptors. The concept dates back
to 1923. In 1938, G. N. Lewis proposed a broader definition in that an acid is an
electron acceptor and a base is electron donor [82]. Acid-base bonds are usually meet
in polymer-adhesive or polymer-paint systems [83-85]. The strength of Lewis acids
and bases (or donor-acceptor) in poorly solvating solvents (usually hexan, cyclohexan)
can be obtained from their heats of reaction (-ΔH), which are related to Ea and Ca, the
empirical parameters for acid, and Eb and Cb corresponding values for base by
equation:
(1.17)
Ea and Eb - susceptibilities of acid or base to undergo electrostatic interactions
Ca and Cb - susceptibilities of acid or base to form covalent bonds.
The chemical bond strengths are collected in Table 1.6.
Table 1.6. Bond energy of some of the common interactions.
Type of interaction Example Energy Range
kJ/mol nm
Covalent
C-C
C-O
Si-O
368
377
368
0.1 – 0.3
Ionic
Na+Cl
-
Al3+
O2-
Ti4+
O2-
503
4290
5340
FRACTURE IN ASYMMETRIC BONDED JOINTS
38
Hydrogen bond
-OH···O=C
-OH···OH
-OH···N
F···HF
30
32
35
163
< 0.2
Lewis acid-base
BF3+C2H5OC2H5
C6H5OH+ NH3
SO2+ C6H5
64
33
4.2
Van der Waals
forces
Dipole-dipole
Dipol-induced
dipole
Dispersion
≤ 2
0.05
≤ 2
100 (retarded)
1.5. Surface treatment of adherends
Since adhesives function by surface attachment, the nature and condition of the substrate
surface are critical to the success of any bonding or sealing operation. Four common criteria
are generally recognized for an ideal bonding surface: cleanliness, continuity, stability and
wetting of the surface by the adhesive.
Cleanliness does not necessarily mean the absence of all surface films, since some surface
films are very strongly attached to the bulk substrate and offer a suitable surface for adhesion.
However, cleanliness does require the removal of unwanted or weak boundary layers such as
oil, dirt, or corrosion. The purpose of cleaning the surface is to remove any weakly attached
materials and to provide a surface that is relatively consistent from part to part.
Discontinuities on the adherend surface, whether chemical or physical, may adversely affect
the apparent strength of the joint by creating localized regions of poor bonding and stress
concentration within the joint. Discontinuities may also make surface cleaning or treating
processes non-homogeneous. These discontinuities could be due to inconsistent
manufacturing processes or chemical inhomogeneity within the substrate.
Stability of the substrate surface is important before bonding as well as after bonding.
Unwanted boundary layers could form during the time between surface preparation and
application of the adhesive, depending on the shop environment and the reactivity of the
surface. Boundary layers could also form during the time period after the adhesive is applied
and before it sets, depending on the reactivity of the surface with the components in the
adhesive or sealant. Certain boundary layers can also form after the adhesive is cured,
depending on the nature of the bond and the type of aggressive environment to which the joint
is exposed. The boundary layers that form after the assembled joint is in service may be the
most perplexing because they are often unexpected and may lead to catastrophically early
bond failure.
Michał K. Budzik
39
Wetting of the adherend surface is a required and important process in establishing adhesion.
There will be various degrees of wetting dependent on the chemistry of the surface that comes
into contact with the adhesive. Along with the wettability of the surface, surface roughness
and topology also influence the strength of bonded joints [86]. The suitability of the bonding
surface will also depend on the type and degree of cleaning or surface treatment that was
performed before application of the adhesive [87].
Surfaces of metallic, polymeric and ceramic adherends
The term surface in adhesive science is usually defined as portion of the adherend with which
the adhesive interacts. The surface is defined by both area and depth of interaction. When a
supposedly smooth solid surface is examined closely under a microscope, it is found to
contain irregularities. It is not flat and smooth but contains many surface asperities, such as
peaks and valleys, with a certain degree of roughness. A rough surface provides more bonding
area than a smooth one of the same gross dimensions. The greater effective surface area offers
a larger area for the forces of adhesion to operate, thereby providing a stronger joint. In an
ideal bonded assembly, the substrate should be the weakest link (Fig.1.25). In most
assemblies that are properly bonded, the adhesive is the weak link because the forces of
adhesion are greater than the forces holding the adhesive material together. Usually, the
internal strength of the substrate and adhesive or sealant system is well understood and can be
controlled.
Fig.1.25. The adhesive joint model, a) metal – metal joint, b) chain presentation[88].
Metallic surfaces such as steel or aluminium alloys might consist of several regions having no
clearly defined boundaries between them as shown in Fig.1.26. Virtually all common metal
surfaces exist as hydrated oxides. Even materials such as stainless steels, nickel, and
chromium are coated with transparent metal oxides that tenaciously bind at least one layer of
water. Thus, the adhesive used for these materials must be compatible with the firmly bound
layer of water attached to the surface metal oxide layer [89]. When working with metal
FRACTURE IN ASYMMETRIC BONDED JOINTS
40
adherends, one must recognize that the nature of the surface can be significantly different for
the same type of metal.
Fig.1.26. Schematic presentation of metallic surface[89].
The situation with organic substrates, such as plastics or elastomers, is even more complex
than with metals. These materials have lower surface energies and lower tensile strength than
metals, and most importantly, polymeric surfaces are more dynamic and likely to change than
metals. There is a greater probability of variation in the surface. As shown in Fig.1.27,
polymeric surfaces have the potential for low molecular weight fragments, oxidation products,
plasticizers, processing aids, lubricants and slip aids, adsorbed water, and organic
contaminants along with various other surprises.
Fig.1.27. Presentation of polymer surface[89].
Components within the polymeric bulk material can also migrate to the surface. It is common
to find low molecular weight polymers or oligomers, plasticizers, pigments, mould release
agents, shrink control agents, and other processing aids as well as adsorbed contaminants in
the surface region. More so than with metals, the surface regions of plastics are dynamic
regions, continuously establishing new equilibrium internally with the bulk material and
externally with the surroundings. Polymers, having both polar and non-polar regions in their
molecular chain can present different chain segments at the surface depending on whether the
surroundings are polar or not. Wiping a surface with an ionic solution will cause the polar
groups to orient toward the surface. While the same treatment with a non-polar solvent, such
as hexane, can bring the non-polar components to the surface. Exposure to heat after surface
treatment could cause fresh, untreated molecular species to appear on the surface, thereby
losing the beneficial characteristics of the surface treatment. As a result of these dynamic
reactions, it is difficult to be confident about the surface of any polymeric material. The actual
Bulk plastic
Oriented layers
Migrated process aids
Mould release agent
Process oil or dust
Bulk metal
Segregation layer
Metal oxide
Hydroxide and water
Other adsorber contaminants
Process oil or dust
Michał K. Budzik
41
surface to which we are bonding is not always the surface that we anticipate. It is also
possible that the surface could change once the bond is made and the assembled joint is
placed into service. Thus, a weak boundary layer that is not present during the bonding
process may form during the joint’s operating life and contribute to a weakening of the
interface.
Ceramic materials have smooth surfaces, usually with very high surface energy. Since
ceramics have high surface energies, they are usually easy to bond under normal conditions.
However, many commercially important ceramics have glazed (glass-like) surfaces. This
glazed surface could provide another interface in the joint that must be addressed. The polar
nature of the bonds between atoms in a ceramic material means that there will likely be an
adsorbed layer of water and hydroxide ions. This layer is tightly held to the ceramic surface.
Adhesives used with ceramics, as those used with metals, must be compatible with the surface
moisture layer.
Surface preparation
The main purpose of surface preparation is to ensure that adhesion develops to the extent that
the weakest link in the joint is either in the adhesive or in the adherend. With optimum surface
treatment, failure should not occur at the interface because of a weak boundary layer or
insufficient wetting. Surface preparation can provide several principal functions:
Remove weak boundary layers that impede wetting of the substrate and create weak
links at the interface. Common weak boundary layers are greases, oils, scale, rust,
tarnish, and other oxides
Protect the substrate surface so that weak boundary layers do not develop during
processing of the joint or during aging in service
Influence the surface energy of the substrate so as to reduce the contact angle between
the adhesive and substrate
Prebond treatments are intended to provide cohesively strong and easily wettable surfaces.
Surface preparations enhance the quality of a bonded metal joint by performing one or more
of the following functions: remove contaminants, control adsorbed water, control oxide
formation, poison surface atoms which catalyze polymer breakdown, protect the adhesive
from the adherend and vice versa, match the adherend crystal structure to the adhesive
molecular structure and control surface roughness [91]. Generally surface treatments
techniques are divided into:
Passive surface preparation methods
Active surface treatment methods
Specific methods for given group of materials
FRACTURE IN ASYMMETRIC BONDED JOINTS
42
Passive surface preparation methods
Passive methods do not actively alter the chemical nature of the surface. Passive processes
only clean the substrate and remove weak boundary layers in the form of contamination.
a) Passive chemical treatments
Passive chemical surface treatments remove soil and organic contaminants from the surface.
They include such common processes as solvent wiping, vapour degreasing, and chemical
cleaning:
Solvent cleaning is the process of removing soil and organic contaminants from a
substrate surface with an organic solvent. Where loosely held dirt, grease, and oil are
the only contaminants, simple solvent wiping alone will provide surfaces for weak to
medium strength bonds (see Fig.1.28). Solvent cleaning is widely used and should
precede any chemical or abrasive surface preparation. Perchloroethylene and
trichloroethylene are the most commonly used of the degreasing solvents. Although
non-flammable, these solvents are still toxic in both their liquid and vapour forms
Chemical cleaning methods are popular on polymeric surfaces where solvent cleaning
may degrade the part or on parts where the contamination is more easily removed by
an aqueous cleaner (e.g. salt films, dirt). Chemical cleaning is generally used in
combination with other surface treatments. The alkaline cleaners are generally used
for cleaning metal surfaces prior to bonding. The most popular types of chemical
cleaners are [92]:
Caustics (sodium or potassium hydroxide)
Silicates (sodium meta silicate)
Amines (triethanolamine, monoethanolamine)
Phosphates (trisodium phosphate, tetra potassium
pyrophosphate) [93]
Acids (phosphoric, hydrofluoric, citric, etc.) [94]
Chelates (EDTA)
b) Mechanical passive methods
Mechanical methods for surface preparation include abrasive blasting (sand or grid blasting)
[95-97], wire brushing, and abrasion with sandpaper [98], emery cloth, or metal wool [99].
These methods are most effective for removing heavy, loose particles such as dirt, scale,
tarnish, and oxide layers. Cleaning is generally required both before and after mechanical
surface preparation. Abrasive blasting is generally the preferred method for removing
contamination from most metal surfaces. It is particularly appropriate for removal of rust,
oxide layers, old coatings, and other heavy contamination. Certain low energy surfaces, such
as polyolefins and fluorocarbons, should generally not be abraded at all prior to application of
an adhesive or sealant. Abrasion and the resulting roughness on a low energy surface will
Michał K. Budzik
43
only increase the probability of air pockets being trapped in the crevices and valleys at the
interface. Also vapour-honing, ultrasonic and electrolyzing cleaning are efficient treating
methods for small, delicate parts.
Active surface treatments
Active surface treatments are chemical or physical processes that not merely clean the surface
or remove weak boundary layers, but they also transform the inherent surface chemistry. They
either improve wetting or modify the boundary layer to be more receptive to bonding.
a) Active chemical surface treatments
Chemical treatments change the physical and chemical properties of the surface to produce
one that is highly receptive to adhesion. Specific chemical treatments have been developed for
various metallic and nonmetallic surfaces. The chemicals used are acidic or alkaline in nature.
- Chemical treatment of metallic surfaces
Metal surfaces are usually some combination of oxides, sulfides, chlorides, acid salts,
absorbed moisture, oil, and atmospheric gases. The pure, bare metal surface may be very
reactive, and unwanted oxide layers and corrosion products could quickly form. Thus, the
surface preparation must not only remove the original surface, but replace it with a surface
coating that will protect the interface during further processing and during the joint’s service
life. A number of techniques have been developed to convert corrosion prone, clean surfaces
to less reactive ones. Three common conversion processes are phosphating [100-102],
anodizing [103-105] and chromating. These processes remove the inconsistent, weak surface
on metal substrates and replace it with one that is strong, permanent, and reproducible. For
aluminum, anodizing provides the most water durable adhesive joints. It is used by many
automotive and aerospace suppliers. The corrosion protection is provided by anodizing the
clean deoxidized aluminium surface in either chromic or phosphoric acid electrolytic baths. In
the USA, phosphoric acid anodizing is often used because of its lower toxicity and easier
disposal. Anodizing creates an oxide under controlled voltage and temperature conditions,
thereby creating a more protective surface [106] (see Fig.1.28).
- Chemical treatment of polymeric surfaces
The chemical modification of low energy polymer surfaces may be carried out by treatment
with chromic acid, metallic sodium complex dispersions, bleach/detergents, potassium
iodate/sulfuric acid and other mixtures. Chemical treatment of polymeric surfaces is generally
more difficult than metallic surfaces and requires special considerations. Polymeric products
often contain pigments, antioxidants, slip agents, mold release agents, etc. that can migrate to
the surface and interfere or alter a surface treatment process. Slight changes in the polymer
formulation or its fabrication process may result in changes in the surface condition and the
effectiveness of treating operations.
FRACTURE IN ASYMMETRIC BONDED JOINTS
44
b) Active physical surface treatments for polymeric materials
Because of the main disadvantages of chemical treatments (hazardous nature and a slow,
batch type process), a number of other active surface treatments have been developed for
polymeric materials. These processes utilize the reactivity of the polymeric surface to gain
change that is favourable for adhesion. Rather than chemical solutions, these surface
modifications are usually made by physical means such as:
- Corona discharge. Corona discharge treatment is a popular method of dry surface
preparation of polymer films [107]. The purpose of the treatment is to make the polymer
surface more receptive to inks or coatings, however, it has also been used effectively as a
pretreatment for adhesives. In this method a spark or corona discharge is produced by ionizing
the air in the gap between the electrodes. The ionized particles in the air gap bombard and
penetrate into the molecular structure of the substrate. Free electrons and ions impact the
substrate with energies sufficient to break the molecular bonds on the surface of most
polymeric substrates. This creates free radicals that react rapidly with oxygen to form polar
chemical groups on the substrate surface and increase the surface energy to a point where
many adhesives, paints, and coatings can wet the substrate. Corona discharge technique was
also tried to aluminium substrates [108] without further outcome.
- Flame treatment. Flame treatment consists of exposing a surface to a gas flame for less than
several seconds. Flame treatment burns-off contaminants and oxidizes the surface of the
polymer similar to corona treatment [109]. Flame treatment is used as a surface treatment for
many low energy polymeric parts prior to bonding.
- Plasma treatment. A gas plasma treating process has been developed for surface treatment of
many polymeric materials. It is a dry process that is becoming a common method of treating
many different engineering plastics when maximum joint strength is required. Low energy
materials, such as polyolefins, polytetrafluoroethylene, polyethylene terephthalate, nylon,
silicone rubber, etc. are readily treated with gas plasma, but also epoxy based composites are
suitable to increase adhesion after plasma treatment [110]. Plasma treatment was also used to
enhance adhesion of aluminium [111]. Operationally, a plasma differs from corona and flame
treatment in that the process is completed at less than atmospheric pressure and with gases
other than air. With the plasma treatment technique, a low-pressure inert gas is activated by an
electrodeless radio-frequency discharge or microwave excitation to produce metastable
excited species that react with the polymeric surface. The plasma treatment produces changes
only to the depth of several molecular layers. Generally, only very short treating times
(seconds to minutes) are necessary. It is generally believed that the plasma treating process
provides surfaces with greater stability than chemical etch, corona, flame, or other common
polymeric treatment processes [112].
- Other methods
Michał K. Budzik
45
Ion beam etching has been used on stainless steel, graphite, and fluorocarbon surfaces
[113]
Excimer laser surface treatment has been used for preparing polyester sheet molding
compounds (SMC) for adhesive bonding in the automotive industry. The excimer laser
preparation of SMC surfaces occurs through the following stages: ablation of surface
contaminates, selective ablation of calcium carbonate filler from the SMC, and
removal of polyester resin from the SMC [114,115]
UV irradiation has also been applied as a prebond surface treatment to a variety of
plastics [116]
It must be pointed out that in structural bonding (especially of aluminium) chemical and
mechanical processes dominates. The effect of the different, common aluminium surface
preparation on the bonded joint durability is illustrated in Fig.1.28.
Fig.1.28. Effect of surface pretreatment on the performance of aluminium joints with a
toughened epoxy adhesive and subjected to ageing in water[90].
Priming to improve adhesion
One approach to improve adhesion characteristics is to intentionally introduce an interphase
region to improve initial adhesion, provide chemical stabilization against degradation in
aggressive environments, and perhaps provide for a broad transitional zone in mechanical
properties between the phases. An interphase engineered to accomplish these goals that is
applied as a separate manufacturing step is referred to as a primer.
Primers are multifunctional compounds designed to provide a means for chemically coupling
to both the inorganic surface and to the organic adhesive or coating. The most widely used
FRACTURE IN ASYMMETRIC BONDED JOINTS
46
coupling agents are based on organosilicon chemistry, although titanates and zirconates have
also enjoyed modest technical and commercial success. An excellent (though slightly dated)
review of coupling agent technology may be found [117]. Silane coupling agents are
generally synthesized through addition of silicon hydrides to unsaturated organic molecules:
(1.18)
where X is a hydrolyzable group such as Cl-OCH2CH3, or -OCH3.
As primers for adhesive bonding, organosilanes (Table 1.7) are typically hydrolyzed in
solution to the corresponding silanol prior to application. In most cases, these silanols begin to
homopolymerize in solution and therefore have a limited shelf life. One result of this
condensation is that the structure of the final film has its origins in the solution history prior to
film deposition.
Table 1.7. Commercial silane coupling agents[13].
Organofunctional group Chemical structure
Vinyl
Chloropropyl Epoxy
Methacrylate
Primary amine Diamine Mercapto
Cationic styryl
While primers are believed to improve the performance of adhesively bonded aluminium
structures primarily through improved corrosion resistance and improved wetting of the
microscopically rough adherend surface by the adhesive, modification of the mechanical
properties of the adhesive near the substrate can have a large effect on both the stress
distribution and total strain energy [118].
1.6. Mechanical testing of adhesive bonding
Adhesion due to its complexity requires specially designed tests. Different substrates,
adhesive types - from very fragile through tough up to very plastic, surface treatments,
adhesion enhancements techniques oblige many different phenomena to occur. Among
different, industrial branches and trades, a variety of test methods for the evaluation of
adhesively bonded joints have been developed and established, including International
Organization of Standardization (ISO), European Committee for Standardization (EN) and
American Society of Testing and Materials (ASTM) standards. During last years lot of tests
were developed inside special laboratories to specifically describe adhesive joints. The spectra
Michał K. Budzik
47
of mechanical tests of the adhesive joints can be subdivided into tens of different ways only
dependent on the chosen criteria. Because of the thesis aim all tests are divided into to groups
dependent on proposed failure criteria:
Failure stress tests – in which stress in principal load direction at failure is estimated
Failure energy tests – in which energy required to separate material or materials is
estimated
1.6.1. Characterisation by failure stress
Fundamental understanding of adhesively bonded joints requires understanding the rudiments
of mechanics, the stress and strain states within the bonded joint. Generally when load is
applied to the bonded joint, the entire joint and the adhesive layer itself is subjected to
different kind of loads.
Tensile tests
One of the basic structural elements encountered in bonded joints is the straight, axially
loaded bar. The stress within loaded bar is given by the well know expression (assuming de
Saint-Venant’s rule):
(1.19)
where:
F - applied axial load
A - cross section are of loaded bar.
Axial deformation caused by applied force, F:
(1.20)
where:
L - initial bar length
E - elastic modulus (Young modulus).
Stored elastic energy, U is:
(1.21)
where:
F - applied external force
FRACTURE IN ASYMMETRIC BONDED JOINTS
48
A - area of bar cross section
E - Young’s modulus of elasticity.
In addition to the axial deformation, which occurs due to the applied axial load, deformations
in the transverse directions are also observed, and are related to the axial strains through the
Poisson’s ratio, ν of the material. Specifically, for uniaxial loading, Poisson’s ratio is defined
as the negative of the ratio of the transverse strain, εt to the axial strain, εa:
(1.22)
with:
ν - Poisson ratio
εt - transverse strain
εa - axial strain.
Theoretically ranging from -1 to 0.5, Poisson’s ratio for most engineering materials ranges
only from around 0.2 to nearly 0.5. Poisson’s ratio can have significant effects on the stress
states present in bonded joints, giving rise to complex three dimensional stress states.
Adhesive bonding tension tests include:
ASTM D897-08 - Standard Test Method for Tensile Properties of Adhesive Bonds
[119]
One of the most commonly used adhesive tensile test method is described in ASTM D897
standard. The recommended sample geometry is shown in the Fig.1.29. Because of high
sensitivity of the test to geometrical aspects like thickness of bondline, uniformity of the
bondline, alignment of the substrates results are often cumber with large scatter [120].
Fig.1.29. Specimen configuration for adhesive tensile testing.
F
F
Michał K. Budzik
49
ASTM C297-04 - Standard Test Method for Flatwise Tensile Strength of Sandwich
Constructions [121]
The standard describes specimen used to determination of the flatwise tensile strength
sandwich constructions.
ASTM D2095-6(2008) - Standard Test Method for Tensile Strength of Adhesives by
Means of Bar and Rod Specimens [122]
The standard recommends bar or rod specimens that are easier to manufacture than those from
D897. The results from all of the tensile tests are reported as simple force, F to bonded area, A
average stress. This can be misleading since average stress differs markedly from the
maximum stress and can easily over exceeds the average value [123]. Designers usually avoid
using adhesives in a direct tensile loading mode. Overlapping, scarfing or fingering the two
pieces can increase bonded surface significantly but also change stress state, allowing
preferable shearing to occur.
Shear tests
Shear tests are very exploited and popular because of their provisional simplicity. Rudiments
of the shear are here recalled. In a simple shaft of length l and second moment of inertia, I
applied torque T will result with:
Shear stress given by:
(1.23)
T – torque
ρ – torque arm
I – second polar inertia moment of the shaft section.
Angel of twist written as:
(1.24)
L – length of the shaft
G – shear modulus of elasticity.
And stored elastic energy:
(1.25)
FRACTURE IN ASYMMETRIC BONDED JOINTS
50
Shear tests most frequently are used to yield shear strength and modulus of the adhesive. Lap
joint test configuration is sometimes used in creep test as well as in durability approaches.
Napkin ring test [124]
The conservative way of yielding shear properties of the adhesive bond is described in both,
ISO 11003-1 [125] and ASTM E229-97 Standard Test Method for Shear Strength and Shear
Modulus of Structural Adhesives [126], which relates to so-called Napkin ring test (see
Fig.1.30).
Fig.1.30. Napkin ring test.
This test produce uniform shear stress within an adhesive layer (ca. 3% variation across the
adhesive ring) and hence is widely use for producing shear strength and modulus data [127]
although care must be taken when producing bondline.
Single Lap Joint Test
Different lap specimens configuration are recommended by ASTM D1002-05 Standard Test
Method for Apparent Shear Strength of Single-Lap-Joint Adhesively Bonded Metal
Specimens by Tension Loading (Metal-to-Metal) [128] (Fig.1.31). Results from this test must
be carefully studied since both, shear and tensile load occurs. Recently finite element studies
[129-132] described difficulties of analyzing test results from single-lap test.
Fig.1.31. Single lap test configuration (ASTM D1002).
ρ
L
T
Michał K. Budzik
51
Double Lap Joint Test
A large amount of tensile stress induced by bending deformation had led to introducing the
double-lap joint specimens. The standardized specimen configurations from ASTM D3528-
96(2008) Standard Test Method for Strength Properties of Double Lap Shear Adhesive Joints
by Tension Loading [133] is shown in Fig.1.32.
Fig.1.32. Double-lap shear adhesive specimen configurations (ASTM D3528).
The lap joint test is the most commonly used adhesive test, because test specimens are simple
to construct and resemble the geometry of many practical joints. However it must be pointed
out that stress distribution in the adhesive is not homogenous. That was primarily stated by
Otto Volkersen. The shear lag model, first published in 1938 [134], is one of the most
fundamental concepts in the transfer of load between two members joined by either discrete
connections, such as mechanical fasteners, or by a continuous layer such as an adhesive.
Solution for the shear lag model is nonuniform shear stress distribution along the bondline as
it is illustrated in Fig.1.33.
Fig.1.33. Graphical presentation of shear lag model.
And followed by equation:
(1.26)
FRACTURE IN ASYMMETRIC BONDED JOINTS
52
where:
E1 and E2 - Young’s modulus of two bonded substrates
h1 and h2 – thicknesses of substrates
F – applied shearing force
and ω:
(1.27)
As can be noted, for the balanced adherend case where the Eihi products for the upper and
lower adherends are the same, the coefficient for the hyperbolic sine term becomes zero, and
the shear stress distribution is symmetric about the centre of the joint. Additionally followed
by Volkersens analysis and those incorporated by Emil Winkler in 1867 [135] in his elastic
foundation model, Goland and Reissner introduced a new model of lap shear geometries. The
Goland and Reissner model is an extended Volkersen model, which includes peel stresses
(according to Winkler model) (Fig.1.34).
Fig.1.34. Peel stress distribution in shear lap geometry.
This becomes particularly important, since adhesives have by far worse properties in tension
than in shearing, and peel, or opening stress, is found to be the initiator of the crack. Up to
nowadays design of adhesive bonds is focused on minimizing peel forces, which is
complicated on real structures [136] in which crack initiation and propagation is driven by the
cleavage stress at the end of the bonded region. Furthermore any static test is not very
convenient in long term, hostile environment testing. Therefore, tests based on energy
required to separate bonded substrates are numerous. Mode I fracture tests are among the
most promising, giving important data- the fracture energy in the most critical opening
(cleavage) mode.
1.6.2. Characterisation by failure energy
Historically failure energy criteria in their essence try to accommodate fracture and
delamination phenomena within the framework of continuum theory. Nowadays experimental
Michał K. Budzik
53
but also mathematical fracture mechanics is based on works of famous engineers like Griffith,
Kies, Irwin, Orowan etc. The application of classical elasticity concepts leads to infinite
stresses at the crack tip. Cracks and sharp comers give rise to singular stress and strain fields
in the material surrounding them. A singular stress has the form ζ~rλ-1
, where r is the distance
from the crack tip or corner and λ is known as the order of the singularity. If 0<λ<1, the stress
ζ ∞ as r 0. Griffith’s principal contribution is an analysis of crack stability based on
energy required to drive the sharp crack [137], i.e. supply (at least) the surface energy for the
two newly formed surfaces.
Consider the situation shown in Fig.1.35.
Fig.1.35. Joint in mode I fracture (F – applied force, Δ – vertical displacement, a – crack
length, δa – crack increment).
If the crack, of length a, is in equilibrium, the decrease of strain energy U, must be equal to
the increase of surface energy S due to crack extension, that is:
(1.28)
where:
- the crack extension force, G, which is defined as strain energy release rate
- energy spent in increasing the crack area
γ - surface free energy.
Griffith’s work dealt with very brittle glass rods. When the material exhibits ductility, as in
most metallic materials, consideration of the surface energy alone is inadequate and then the
vast majority of the released strain energy is absorbed generating plastic flow in the material
near the crack tip. Thus, catastrophic fracture occurs when the strain energy is released at a
rate sufficient to satisfy the needs of all the energy requirements, parameter Gc. Fracture
mechanics analyses relate to sharp cracks, i.e. when the local stress for fracture is reached. For
metallic materials this is usually achieved by fatigue precracking from a notch. Application of
Fracture Mechanics to polymers presents additional problems, including viscoelasticity and
the generation of sharp precracks. For adhesive bonding the latter problem is frequently
F
Δ
a δa
FRACTURE IN ASYMMETRIC BONDED JOINTS
54
tackled by monitoring a moving, and therefore sharp, crack. Not only in ductile, but also in
near-brittle [138] materials a plastic zone develops at the tip of the crack. As the applied load
increases, the plastic zone increases in size until the crack grows and the material behind the
crack tip unloads. Hence, a dissipative term has to be added to the energy balance relation
devised by Griffith for brittle materials. In physical terms, additional energy is needed for
crack growth in ductile materials when compared to brittle materials. This was first stated by
George Rankin Irwin. Irwin's strategy was to partition the energy into two parts [139]:
The stored elastic strain energy which is released as a crack grows. This is the
thermodynamic driving force for fracture
The dissipated energy which includes plastic dissipation and the surface energy (and
any other dissipative forces that may be at work). The dissipated energy provides the
thermodynamic resistance to fracture.
Therefore, the total energy is:
(1.29)
with:
Gp - the plastic deformation dissipation (and dissipation from other sources) per unit area of
crack growth.
For brittle materials such as glass, the surface energy term dominates and G≈2γ. For ductile
materials such as steel, the plastic dissipation term dominates and G≈Gp. For polymers and
therefore adhesives, we can obtain full set of different properties.
Standard Mode I tests
The ASTM has standardized several cleavage tests for adhesives, from which all of them are
based on symmetric sample geometry.
ASTM D1062-08 Standard Test Method for Cleavage Strength of Metal-to-Metal
Adhesive Bonds test [154]
The ASTM D1062 specimen is shown in Fig.1.36. It is also referred as a compact tension
(CT) test of the adhesive joints. The test specimen is fabricated from two identical pieces of
metal that are adhesively bonded to form bondline. The load is applied off-axis, subjecting the
specimen to combined tension-bending load. The standard requires breaking force to be
registered per specimen thickness. Additionally, percentages of surface experiencing cohesive
and adhesive fracture are to be reported.
Michał K. Budzik
55
ASTM D3807-98 (2004) Standard Test Method for Strength Properties of Adhesives
in Cleavage Peel by Tension Loading (Engineering Plastics-to-Engineering Plastics)
test [155]
The beam cleavage test configuration, shown in the Fig.1.37 is referred in ASTM to D3807.
The standard recommends beam like cleavage specimens. The adherend strips are required to
be semirigid, such that they can bend through an appreciable angle without failing. The results
are reported as a force per width required to propagate the failure.
Fig.1.36. ASTM D1062 cleavage test of adhesive joints (left).
Fig.1.37. ASTM D3807 cleavage test specimen (right).
ASTM D3433-99 (2005) Standard Test Method for Fracture Strength in Cleavage of
Adhesives in Bonded Metal Joints [156]
This test was initially conceived as true adhesion fracture test. The specimen geometry is
illustrated in Fig.1.38. The specimen is composed from two strips (symmetrical) which are
separated using tensile machine. D3433 test is one of the most common and widespread
fracture test of the adhesive joints and is abbreviated to DCB (Double Cantilever Beam) test.
The fracture energy can be found as:
(1.33)
with: U – elastic energy and A being surface area created by the propagating crack (∂A=∂ab
where a – crack length, b – width of the substrate – constant).
Finally, the adhesive fracture toughness in cleavage test can be yield as:
(1.34)
FRACTURE IN ASYMMETRIC BONDED JOINTS
56
For the beam with the inertia moment define as I:
(1.35)
Fracture toughness expression becomes:
(1.36)
It is worthwhile noting that in this configuration the shear stress component is usually very
small, since a>>h. However, it should be pointed out that two values need to be estimated:
the crack length and force, both strongly affect the final toughness expression. Force can be
measured directly by the force sensor mounted to the tensile machine. The crack length is
very hard to measure directly and finally must be estimated in another way. Since the test is
designed to be made on a tension machine we must notice that due to increasing crack length,
the momentum required to separate plates is decreasing, which plays an important role in
compliance measurements. As is proven by experimental and analytical studies, also other
effects like anticlastic curvature, root rotation or residual stresses can imply even more
significant contributions to final toughness values [157,158]. Some of these problems were
solved by introducing tapered geometry, shown in Fig.1.39. The test is referred as TDCB
(Tapered DCB) and is recommended by the same ASTM standard – D3433.
Fig.1.38. ASTM D3433 test specimen (left).
Fig.1.39. ASTM D3433 TDCB test specimen (right).
Since during the test cross section of the beams or substrate varies with the crack length
increment additional factor, m occurs in the final energy release formula:
(1.37)
Michał K. Budzik
57
with factor m being:
(1.38)
Taper is chosen to vary the beam compliance, C (C=ΔF-1
), such that for given force F at
failure, fracture energy Gc is independent of the crack length, a. As can be found, TDCB is
free of crack length interpretation problem, but compliance arises the others. Since all tensile
machines have their own compliance, comparison between the results from different
laboratories give huge scatter of results [159,160]. In addition, adhesive elasticity and
viscoelastic properties allows root rotation to occur which significantly increase errors to
almost 30% in terms of fracture energy [161]. Tapered specimen geometry is not very
convenient from technological and practical points of view. Finally in situ, environmental
tests are difficult for both DCB and TDCB tests. Nowadays, hostile environment effects are
studied using aged samples are used. We must remember that ageing on unloaded samples
cannot be compared with combined environment-mechanical loading cycle e.g. because
ageing conditions require higher loads, which cannot be transferred when the structure is
mechanically loaded. Finally tapered geometry cannot be obtained for many materials e.g.
some composites or ceramics, because of the machining necessity.
ASTM D3762-03 Standard Test Method for Adhesive-Bonded Surface Durability of
Aluminium - The Wedge Test [162]
Limitations of the DCB test making use of tensile machines led to the development, first at
the Boeing company in 1980’s, of the adhesive bonding durability test referred to as the
Boeing Wedge Test, and now known as standard ASTM D3762 test. The test specimen is
shown in Fig.1.40.
Fig.1.40. D3762 wedge test configuration for durability testing.
FRACTURE IN ASYMMETRIC BONDED JOINTS
58
In this test a double cantilever beam of the geometry identical to the DCB test is loaded by
forcing a wedge of thickness Δ between the separated beams. The fracture energy is given as:
(1.39)
with:
Δ – wedge thickness.
Contrary to the DCB and TDCB tests in wedge test the crack propagation rate is decreasing
with time leading to stable propagation. A major drawback of the wedge test is high
dependence of the fracture energy on crack length, scaled to -4th
power.
Summary of the mode I fracture tests
Of the various adhesion tests available for evaluating the fracture strength of structural
adhesive joints, the double cantilever beam (DCB) and its close relative, the (so-called
Boeing) wedge test, are amongst the most versatile, and generally yield the most reliable
information about fracture energy [163-170]. It has to be pointed out that in standards only
symmetric geometries are listed, and thus asymmetric geometry is rarely studied in the
literature. With a judicious choice of test geometry, these systems lead to relatively small
adherend strains near the crack front [157,171]. As a result, local plastic strain, which leads to
supplementary energy dissipation, is relatively limited. If the length of the opening crack
(either within the adhesive or at the interface adherend/adhesive, depending on type of failure)
is represented by a, it may be shown that the energy release rate, equivalent to fracture
energy, Gc, follows a scaling rule of the form Gc~a2. Beam analysis based on the opening
displacement and the force applied allows a, and therefore Gc, to be evaluated. However,
since the bending moment leading to failure increases linearly with a, crack growth may
accelerate and become unstable in certain cases. This problem has been countered by the
development of the more refined, tapered double cantilever beam (TDCB) test, in which
stability is restored by using profiled adherends with thickness increasing away from the
region of force application [e.g. 172]. In principle, crack length need not be measured directly,
but such obtained results were found not precise [159,160,161]. In addition it is not always
convenient, or even possible, to use profiled adherends (for instance, when testing the
adhesion properties of automotive body assembly materials). Away from tensile machine
based techniques, various techniques have been used to study crack lengths in this tests. The
most basic techniques rely on direct, or microscopic, observations of the position of the crack
tip, sometimes with the addition of paint, or other marking fluid, to the joint edges to facilitate
observation [157,173-175]. Use has been made of optical correlation [173], laser moiré [176]
and speckle interferometry [177]. Electrical techniques have also been tried, such as
measurement of crack growth through changes in electrical resistance of carbon paint applied
to the edges of non-conducting substrates [178], or by employing piezoelectric techniques
[179]. The use of a single strain gauge technique has also been reported [180]. Displacement
Michał K. Budzik
59
sensors have been employed for continuously monitoring cracks [157]. Measurement of crack
length nevertheless remains a delicate process in many practical cases. Moreover
environmental testing seems to be crucial. In reality even while wedge test is performed the
crack measurements are difficult. After the sample is loaded with wedge insertion and placed
in the environmental chamber the crack length, a should be observed and recorded. Since very
sensitive crack measurements are required, and these were not possible under hostile
environmental conditions, such approach was left. Normal, and nowadays common, method is
to remove the sample from the chamber and measure crack length manually using e.g. simple
calliper gauge or a different method described above not assuring required accuracy and
reliability. In addition, immersion-drying cycle certainly must change the adhesive behaviour
e.g. due to possible fatigue. Another problem arises due to possible loading mode mixity. In
practice, most materials to be bonded are different in geometrical, physical or chemical
meaning which implies different mechanical behaviour. For example it was found, that when
the system becomes increasingly asymmetric, a greater degree of in-plane shearing
incorporating mode II [141]. The locus of failure is also changing, shifting from cohesive in
the adhesive to interfacial along the adherend surface [142-148]. The anti-plane shearing
effect – mode I/III mixity was studied very rarely [e.g. 152,153] particularly for the
development of the mode mixity tests. Mode mixity is also suspected to change final value of
the fracture energy, since the adhesives are much stronger in shearing [149] even small mode
mixity contribution may markedly change final fracture energy. The strain energy release rate
in mode mixity conditions is given by equation:
(1.31)
Where index I, II, III corresponds to opening, in-plane shearing and anti-plane shearing
(tearing) fracture modes respectively. The effects of the mode I/II mixity were studied by
Chen et al. [150,151]. The authors have introduced energy release rate mixity factor:
(1.32)
They pointed out that when η<3% the crack front will oscillate between the adherend
interfaces. When global mode mixity increase, η<14% the crack front has increased
directional stability and propagates along one adherend only. The effects of mode III/I mixity
are ignored by the authors.
FRACTURE IN ASYMMETRIC BONDED JOINTS
60
Chapter 2. ASYMMETRIC JOINTS
According to the literature and international standards fracture mechanics tests in the crack
opening mode I conditions are made using symmetric joint geometry, made from, in principle,
the same materials with the same dimensions.
The focus of this study was asymmetric joints (see Fig.2.1) made from two bonded plates one
being assumed rigid, of thickness H, the second elastically deformable, or flexible, of
thickness h. The asymmetric bonded joint configuration allows any materials combination to
be used. Schematic representation of the sample used, with all of the geometrical and material
features, is shown in Fig.2.1.
Fig.2.1. Asymmetric test sample.
The rigid-flexible assumption is kept by having appropriate flexural rigidity ratio of the two
bonded plates (beams):
(2.1)
where E and I are elastic modulus and second moment of inertia of the rigid (index H) and
flexible (index h) plates respectively.
2.1. Materials and specimens
To study the methodology of testing of the adhesive joints and validity of the developed
procedures different materials systems were studied. Material selected for rigid plates (in all
experiments) was aluminium-magnesium alloy AA5754. Flexible plates were made of:
aluminium-copper alloy AA2024, carbon fiber reinforced polymer - CFRP, polycarbonate -
PC. The adhesives used to bond rigid and flexible plates were: pure and nanoparticles
reinforced epoxy, acrylic mastic adhesive, cyanoacrylate adhesive and PSA type adhesive.
Aluminium-magnesium alloy AA5754-H111
The aluminium-magnesium alloy AA5754 was supplied by Alcoa (Pittsburgh, PA, USA) in
H111 condition which corresponds to strain hardening and annealing to give small increase of
Flexible adherend
The adhesive
Rigid adherend
h, Ih, Eh, l, b
H, IH, EH, L, b
e, b, Eadh, ladh Force, F
Deflection, Δ
Michał K. Budzik
61
strength. The chemical composition of AA5754 alloy to the supplier standard is given in
Table 2.1.
Table 2.1. AA5754 chemical composition.
Chemical Composition Limits Others
Weight % Si Fe Cu Mn Mg Cr Zn Ti Each Total
Minimum - - - - 2.6 - - - - -
Maximum 0.40 0.40 0.10 0.50 3.6 0.30 0.20 0.15 0.05 0.15
The magnesium concentration is usually maintained less than 4% (weight) in order to avoid
intermetallic phase Mg5Al8. Work hardened aluminium alloys tend to soften with time
because the microstructure is not stable even at ambient temperature. Therefore annealing is
(in 400oC) to the required strength (Table 2.2) and stability required operation after work
hardening. The 5754 alloy has good corrosion resistance and medium mechanical properties
and is mainly used in the shipbuilding industry, automotive bodies and as rivets. Plates of
thickness H=6 mm, width, b=25 mm, and length of L=180 mm were cut from 50 x500x6 mm
plate using milling machine. To avoid any changes in the 5754 alloy microstructure all
operations were taken in the stream of cooling liquid.
Table 2.2. Physical and mechanical properties of AA5754-H111 alloy.
Properties: Denisty
(gcm-3
)
Modulus of
Elasticity
(GPa)
Yield stress
(MPa)
Tensile
Strength
(MPa)
Coeff. of Thermal
expansion
(oC
-1)
Required: 2.66 68 100 215 24 x 10-6
Aluminium-copper alloy AA2024
The aluminium-copper alloy AA2024 was supplied by Alcoa (Pittsburgh, PA, USA). The
AA2024 series is most extensively used in aeronautic industry for plane fuselages. The
chemical composition of the alloy is shown in the Table 2.3.
Table 2.3. AA2024 chemical composition (supplier data).
Chemical Composition Limits Other
Weight % Si Fe Cu Mn Mg Cr Zn Ti Each Total
Minimum
-
-
3.8
0.30
1.2
-
-
-
-
-
Maximum 0.50 0.50 4.9 0.9 1.8 0.10 0.25 0.15 0.05 0.15
The usual heat treatment includes solution heat treatment and ageing, which gives final
structure of copper solid solution with metastable CuAl2 precipitates which gives rise to the
strength properties and hardness. Although, the corrosion resistance of AA2024 alloy is not
very good, since this alloy is particularly subjected to stress corrosion cracking due to
FRACTURE IN ASYMMETRIC BONDED JOINTS
62
possible copper precipitates, it has good fracture toughness and good weight/strength ratio.
Also it is the cheapest of the high performance aluminium alloys. The plate was delivered in
T3 condition – solution heat treated (ca. 500oC), naturally aged and finally cold rolled. The
mechanical properties are listed in the Table 2.4.
Table 2.4. Properties of AA2024-T3 (supplier data).
Properties: Denisty
(gcm-3
)
Modulus of
Elasticity
(GPa)
Yield stress
(MPa)
Tensile
Strength
(MPa)
Coeff. Of Thermal
Expansion
(oC
-1)
Required: 2.77 73.1 260 440 22.9 x 10-6
The significant advantage of the 2024 alloy, concerning the adhesive bonding, is the fact that
is very well known and studied and, in addition, was found especially susceptible to the
porous alumina layer formation on its surface [181]. The plates of thickness h=1.6 mm, width
b=25 mm and length 120 mm, were cut from 2000x2000x1.6 clad AA2024-T3 plate with
press cutter. Then all edges were finished with milling machine in the presence of cooling
liquid and finally polished to required dimensions.
Carbon Fibre Reinforced Composite
One of the biggest advantages of adhesive joints is the possibility to join fibrous composite
materials without weakening them e.g. by holes, thermal processes etc. Two epoxy resin
matrix, orthotropic carbon fibre woven composites were used. One was delivered by the
supplier, second was fabricated by the author in Merignac Institute de Maintenace
Aeronatique (IMA). Both epoxy composites were made of 6 layers of orthogonal pre-preg
HexPly M10/42%/193P/CHS-3k/1000mm, supplied by Hexcel (Stamford, CT, USA), giving
a thickness, h of 1.22 mm. Composite made at IMA was cured in 120°C under 0.3 bar
pressure for 1 hour under the vacuum bag. The precise supplier fabrication procedure is not
known. Mechanical properties of the fabricated composite were obtained using ultrasonic
Through Transmission Method (TTM) and verified in bending tests. The detailed procedure
of the TTM method can be found in authors master thesis [182]. Young’s modulus, obtained
from 3-point bending, was evaluated at 50 5 GPa, the same like in TTM method. Poisson’s
ratio, νC was 0.05. Properties of the composites employed are summarized in Table 2.5.
Finally plates (250x200) were cut with hydraulic press, edges were polished with the emery
paper to the 120x25 mm (length x width) plates.
Table 2.5. Properties of the CFRP composites.
Properties: Denisty
(gcm-3
)
Modulus of Elasticity
(GPa)
Thickness
(mm)
Tensile Strength
(MPa)
Poisson
ratio
Supplier
IMA
1.38
1.29
77
50
1.22
1.22
850
-
0.05
0.05
Michał K. Budzik
63
Makrolon® (C16H14O3)
Makrolon® General Purpose Sheet is a polycarbonate (PC) product. Sheet was supplied by
Bayer (Sheffield Plastics, Bayer MaterialScience LLC, Sheffield, MA, USA) is a clear,
colourless polymer used extensively for engineering and optical applications. It is available
commercially in both pellet and sheet form. In the studies 5 mm thick polycarbonate plates
were used, as a flexible adherend, to observe directly, using video microscope camera,
formation and shape of the crack front. The 120x25x5 mm plates were cut from 200x200x5
mm sheets using a band saw. After cutting, edges were polished with 1200 grade emery paper
to the desired dimensions.Properties of the PC plates are listed in the table 2.6.
Table 2.6. Used polycarbonate properties (supplier data).
Properties: Denisty
(gcm-3
)
Modulus of Elasticity
(GPa)
Tensile Strength
(MPa)
Average: 1.20 2.5 65
Bostik Araldite Cristal
One of the adhesive used was a commercial DGEBA (DiGlycidyl Ether of Bisphenol A:
epoxy resin synthesized by reacting bisphenol-A with epichlorohydrin in the presence of a
basic catalyst) epoxy resin (Bostik, La Défense, Paris, France) of average molecular weight
n<700 cured with N(3dimethylaminopropyl)-1.3propylenediamine. Crosslinking of the
adhesive was effected at ambient temperature (ca. 20°C) for 48 hours under 0.3 bar pressure
and at ca. 55% RH.
Epidian 6
Epoxy resin - Epidian®6 (E6) supplied by Chemical Works (Organika-Sarzyna S.A., Nowa
Sarzyna, Poland) is second resin produced from DGEBA used in the studies. The epoxy
equivalent weight, EEW, was 188 g/equiv. and the average molecular weight - n≤700. The
hardener used was 1-butyloimidazole, used in proportion 1phr (per hundred portions of resin).
To study the nanoparticles adhesive bonded joints efficiency basic E6 resin was reinforced
with 5% (by weight) of 1-D montmorillonite (MMT) nanoparticles. Nanoparticles were
dispersed in the resin using ultrasonic sound with frequency, f=35 kHz for ca. 10 min. Curing
of both of the E6 systems was effected in 150oC for 1 hour under constant pressure of 0.3 bar.
Colle Mastic Tout Fixer
As a example of highly elastic, low elasticity modulus adhesive (Eadh ca. 200 MPa) acrylic
mastic adhesive was used, Colle Mastic Tout Fixer (Leroy Merlin, 59712 Lille, France) was
FRACTURE IN ASYMMETRIC BONDED JOINTS
64
used. Curing was effected at ambient temperature (ca. 20°C) for 48 hours under 0.3 bar
pressure and at ca. 55% RH.
UHU® Super Glue
Cyanoacrylates posses very good mechanical properties, but are very expensive and
aggressive to the environment and human. UHU® Super Glue (UHU GmbH & Co. KG, D-
77813 Buhl, Germany) is a commercial fast curing cyanoacrylate adhesive (ethyl 2-
cyanoacrylate). Curing of the adhesive was effected at ambient temperature (ca. 20°C) for 48
hours under 0.3 bar pressure and at ca. 55% RH.
PSA adhesive
Pressure Sensitive Adhesive - PSA type adhesives are common in everyday applications.
Commercial, PSA type (caustic base) adhesive, double face scotch - Scotch Double Coated
Tape was supplied by 3M (St. Paul, MN, USA). The joint after bonding with the tape was left
for 48 hours under 0.3 bar pressure.
2.1.1. Characterization of substrates
Three point bending test and cantilever beam test were performed to evaluate bending
properties of the flexible substrates. Three different span/arm lengths (60, 80, 100 mm), three
different forces (5, 10, 20 N) and three different plates from the same materials (CFRP,
AA2024-T3) were used. Results of the test are summarized in the Table 2.7.
Table 2.7. Bending test results.
Properties: Modulus of
Elasticity
(GPa)
Yield stress
(MPa)
CFRP
AA2024-T3
50 ± 5
62 ± 5
-
200 ± 7
In addition ultrasonic TTM (Through Transmission Method) [183] was used to evaluate
mechanical properties of the flexible substrates, using device shown in Fig.2.2. Results from
TTM test are also given in the Table 2.8.
Table 2.8. Materials properties from TTM measurement.
Properties: Poisson
ratio in xy
Modulus of Elasticity in x*
(GPa)
Modulus of Elasticity in y*
(GPa)
CFRP
AA2024-T3
0.03
0.32
53.76
68.19
49.54
67.23 *
x and y being are longitudinal and transverse ordinates respectively.
Michał K. Budzik
65
Fig.2.2. Through Transmission Method device.
2.1.2. Surface preparation of the substrates
Before bonding the substrate plates were subjected to variable surface treatment affecting the
surface topography (sandblasting, abrading, polishing, electrochemical PAA process) and/or
chemical structure (PAA process). Composites plates where abraded using 400 grit emery
paper, wiped in the ethanol alcohol and dried in the stream of warm air. Procedures of
aluminium surface treatment are detailed in Table 2.9.
Within the thesis the PAA surface treatment was used for different aluminium alloys. It must
be pointed out that the influence of the alloying elements on the PAA process of aluminium is
not well known yet [184] concerning the growth of the alumina layer. When the aluminium is
electrochemically anodized, an oxide grows at the anode according to the reaction [185]:
(2.2)
And hydrogen evolves:
(2.3)
The PAA is recommended process because is less toxic than common chromic acid
anodization, in addition, the phosphoric acid electrolyte was found to give the best adhesion
properties of the porous alumina layer [186]. The pore formation mechanism is displayed
schematically in Fig.2.3 following four stages [187]. At the beginning of the anodization, the
barrier film, which consists of non-conductive oxide (R=1010
~1012
Ωcm) [188], covers the
entire surface of the aluminium (stage 1). The electric field is focused locally on fluctuations
of the surface (stage 2). This leads to field-enhanced or/and temperature enhanced dissolution
in the formed oxide and thus to the growth of pores (stage 3). Since some pores begin to stop
growing due to competition among the pores, the current decreases again. Finally, the pore
formation current maintains an stabilizes state. In this stage, pores grow in a stable manner
(stage 4). However, it is very often observed that during the stable pore growth, the current
Emitter
Sample
Receiver
Medium
- water
FRACTURE IN ASYMMETRIC BONDED JOINTS
66
density continues to decrease slightly. This is due to diffusion limits in the long pore channels
[189,190].
Table 2.9. Aluminium surface treatment procedures.
Polishing Sandblasting Anodization
Degreasing:
96 % ethanol
Abrasion/Polishing:
Emery paper
180 400 800 1200 2400/4800
Surface cleaning:
DI water stream
Ultrasound cleaning: Drying:
Methanol bath
T=30oC
t=15 min
f=35 kHz
Ethanol rinse
Hot air stream
Water break test
AFM Bonding Sandblasting: Anodization:
Perpendicular
40 mm distance
9 μm (150 μm)
SiO2 (Al2O3)
P=3 At
10% wt. H3PO4
10 V DC
23oC
20 min
20 mm distance / parallel
Ti grid cathod
Ultrasound cleaning:
Methanol bath
T=30oC
t=15 min
f=35 kHz
Drying:
Acetone rinse
Hot air stream
Bonding Bonding AFM
Michał K. Budzik
67
Fig.2.3. Schematic diagram of the pore formation at the beginning of the anodization. Stage1:
formation of barrier oxide on the entire area, stage 2: local field distributions caused by
surface fluctuations, stage 3: creation of pores by field-enhanced or/and temperature-enhanced
dissolution, stage 4: stable pore growth.
2.1.3. AFM and SEM control of the surface
Atomic Force Microscope (AFM) (Veeco Instruments Inc., Veeco Metrology Group,
Nanoscopes, New York, USA) and Scanning Electron Microscope (SEM) (Philips-FEI, XL30
ESEM-FEG/EDAX, Amsterdam, The Netherlands) were used to study the surface topography
and structure after the PAA surface treatment.
Alumina layer topography obtained for the AA2024 alloy is shown in Fig.2.4. Profile and the
section of Al2O3 layer has been illustrated in Fig.2.5. The darker, spherical areas within this
layer indicate porosity.
The alumina topography observed for AA5754 alloy using AFM is illustrated in Fig.2.6. To
be noted is that this topography is different than for the 2024 aluminium being less peaky.
Thus the 2024 alloy offers better surface development, enhancing adhesion.
ANODE:
Aluminium
Electrolyte:
10% wt. H3PO4
+ H2O
CATHODE:
Titanium grid
FRACTURE IN ASYMMETRIC BONDED JOINTS
68
Fig.2.4. Topography of AA2024-T3 before (left) and after anodization (right).
Fig.2.5. Alumina layer obtained on Al-Cu alloy.
Fig.2.6. Topography of AA5754-H111, before (left) and after (right) anodization.
Al2O3
2024 alloy
Michał K. Budzik
69
2.1.4. Fabrication of adhesive joints
Following the appropriate surface treatment (section 2.1.3) of the substrates the plates were
bonded using one of the adhesives described. Based on standard symmetric DCB sample
geometry a new asymmetric specimen was introduced in this work (see Fig.2.7).
Fig.2.7. Symmetric and asymmetric adhesive joints.
Routine preparation of the joint is shown by the block scheme Fig.2.8.
Fig.2.8. Block scheme o sample fabrication.
Bondline thickness, e was maintained by inserting PTFE (Teflon®) spacers of a given
thickness (depending on the test) at the two joint extremities (see Fig.2.9). The constancy of
the bondline was checked by optical microscopy.
Fig.2.9. The constancy of the bondline maintained by PTFE spacers.
e
Rigid
substrate The
adhesive
Flexible
substrate
PTFE spacers
Cutting of
substrates
Substrates
surface
preparation
Bonding
with
appropriate
adhesive
Adhesive
crosslinking
Edge
polishing
2Δ
SYMMETRIC JOINT
GEOMETRY
ASYMMETRIC JOINT
GEOMETRY Δ
Flexible
substrate
Rigid
substrate
FRACTURE IN ASYMMETRIC BONDED JOINTS
70
After crosslinking of the adhesive sample edges were polished using abrasive emery paper of
the grit 600-2400 to permit in situ crack path observation from the sample side (see Fig.2.10).
Fig.2.10. Flight-view of the side-camera observations made during the tests.
2.2. Microscopic studies
Atomic Force Microscopy (see section 2.1.3) was used to study the effects of PAA surface
treatment on surface topography. Optical Microscopy (OM), Scanning Electron Microscopy
(see section 2.1.3) were used to observe the quality of the adhesive joints, crack paths inside
the adhesive and interlayers but also for fractography. The optical microscope was Leica
Reichert MEF4M (Leica Camera AG, Wetzlar, Germany) coupled with Canon Power Shot
G5 photo camera (Canon Inc., Tokyo, Japan). For SEM and OM studies edge of the samples
were abraded using 200- 2400 grit paper and polished using alumina polishing powder.
Fracture surfaces were left untouched after fracture. Samples used for SEM study were
sputtered with gold – platinum conductive layer in order to allow electron flow.
2.3. Crack path observations
In situ crack path observations (see Fig.2.10) were made using one of the three camera
systems:
a) Canon D40 Photo Shot Camera (Canon Inc., Tokyo, Japan) - used in the test with
polycarbonate substrate and in CRT and CFT
b) Digital Micro Camera (Dino-Lite Pro-IS Production S.A., St. Genis Pouilly, France) –
used in the elevated temperature CDT test
c) CCD (Charge – Coupled Device) camera Ikegami ICD 47E (Ikegami Tsushinki Co.,
Ltd., Tokyo, Japan) – used in variable adhesion tests with CFRP substrate.
Polished
side Tracking
Camera
JOINT
Michał K. Budzik
71
Chapter 3. ADHESIVE BONDING TESTS AND ANALYSIS
In order to study the adhesive joint strength and the efficiency of adhesive bonding for given
materials systems the following tests were introduced and performed: Constant Rate Test
(CRT), Constant Force Test (CFT) and Constant Displacement Test (CDT). The CRT and
CDT tests were previously run for symmetric joints (DCB test and Wedge Test respectively),
accordingly introducing new asymmetric specimen required the development of entirely new
tests (features and metrology model in both cases and completely new metrology for CDT
test). The CFT test is entirely new test developed in this thesis. The Catia v5 (Dassault
Systems, Vélizy-Villacoublay, France) environment was used to design the test features.
Finite Element Method (FEM) using Caste3M (French Atomic Energy Commissariat, Gif-
sur-Yvette, France) was employed to analyze stress state within the joint introduced by the
proposed geometry of the joint. Additionally analytical model were proposed to study
background effects of the proposed tests. Proposed test are based on energy release rate
(Fig.3.1) consideration which are used here to evaluate the fracture energy.
Fig.3.1. Energy Release Rate principle (Δ-separation distance, F-applied force, a-crack length,
δa-crack length increment, U-stored elastic energy).
The strain energy release rate, for the cases when either displacement, Δ or applied force, F
are resulting in e.g. plate separation can be calculated as:
(1.30)
Crack propagation criteria are:
G=Gc – threshold
G<Gc – crack stabilization
G>Gc – unstable crack growth.
At threshold, G=Gc – is named critical energy release rate, fracture energy, fracture toughness
or like in the adhesion fracture tests – the work of adhesion forces.
F
Δ
FRACTURE IN ASYMMETRIC BONDED JOINTS
72
3.1. Constant Rate Test (CRT)
Constant Rate Test is closely related to the standard DCB test. The modification are
asymmetric sample geometry and data reduction method and test interpretation. Flexible
substrate deflection, Δ and applied force, F are recorded and used directly to calculate joint
fracture properties. The setup was built using universal tensile machine adopted to fix
specimen and the force as well as displacement sensors (see Fig.3.2).
Fig.3.2. Constant Rate Test sample (a) and principle of the test (b).
The rigid adherend (1) is fixed to tensile machine frame with screws. The flexible adherend
(2) is connected by the hinge (4) to the force sensor (5) and tensile machine crosshead with
the steel chain (Ф 4mm). In addition, displacement sensor (6) is mounted under the flexible
beam to permit direct plate deflection measurements. Test begins with crosshead move up to
the moment when strain energy stored in the flexible beam exceeds critical fracture energy,
this initiate crack propagation. Since tensile machine crosshead speed is kept constant during
the test the deflection Δ is increasing linearly with time, thus only force F is unknown. The
5
To tensile machine
crosshead
6
Tensile Machine frame
1 4
2
3
dΔ/dt = const
a)
b)
Michał K. Budzik
73
Δ = f(t)
F = f(t)
a = f(t)
z
x
force measurement was carried out using strain gauge force sensor Instron with 500N load
carrying capacity (Instron, Norwood, USA). Plate deflection measurements were made using
LVDT (Linear Variable Differential Transformer) sensor Peltron PSzl 20 (Peltron Ltd.,
Warsaw, Poland). This test was found previously (in symmetric specimens) to be very rate
dependent which is particularly pronounced when the adhesive is highly viscoelastic.
Therefore brittle adhesives were chosen for CRT. The CRT routine is illustrated in Fig.3.3.
Fig.3.3. CRT test routine.
3.1.1. Data reduction method
The CRT test was analysed using Euler-Bernoulli simple beam theory (SBT) [191]. The test
physical model is based on that of cantilever beam and is shown in Fig.3.4.
Fig.3.4. Constant Rate Test physical interpretation.
Using general fracture energy balance proposed by Griffith:
(3.1)
where:
G – fracture energy
U – strain energy release rate
a – crack length
b – constant (width of the beam).
Irwin – Kies fracture mechanics compliance criteria can be used to evaluate fracture energy:
(3.2)
where:
PC
F = f(Δ)
A/D
converter Displacement
sensor, Δ
Force sensor, F (V)
(V)
FRACTURE IN ASYMMETRIC BONDED JOINTS
74
F – applied force
C – compliance:
(3.3)
Δ – beam deflection at the point of applied force.
Bending moment, M distribution along the longitudinal x beam direction due to the applied
force, F:
(3.4)
Governing simple beam theory equation is given by:
(3.5)
where:
Eh – Young modulus of flexible adherend
Ih – second moment of inertia of the beam cross section.
Equation (3.5) is double integrated with appropriate boundary conditions to yield the solution
in terms of plate deflection. The boundary conditions are:
(3.6)
(3.7)
Following the coordinates from Fig.3.4, for simple beam model we can write down the
deflection of the beam equation:
(3.8)
From which compliance is evaluated:
(3.9)
Now:
(3.10)
Michał K. Budzik
75
Substituting (3.10) to (3.2) we obtain:
(3.11)
with Ih:
(3.12)
From (3.8) and (3.12) we may find apparent crack length:
(3.13)
It must be noticed, that the crack length equation is correct only when force and displacement
are measured in the same line, as was made in the studies. Finally, (3.11) and (3.13) gives
fracture energy:
(3.14)
Errors of crack length and energy release rate estimations were calculated using error
propagation method [192]. Following the rule:
(3.15)
Crack length estimation error is found:
(3.16)
Error of energy release rate estimation, G can be written as:
(3.17)
Finally, the energy release rate error in Constant Rate Test is found:
FRACTURE IN ASYMMETRIC BONDED JOINTS
76
(3.18)
3.1.2. Calibration of the crack length - artificial crack tip test
In order to compare the estimated crack length, a calculated using (3.13) with the real –
established and directly measured crack length, aD artificial crack tip test was developed, in
which a moving clamp was introduced to simulate the moving crack front (Fig.3.5). The same
aluminium adherends as described above, of thicknesses 6 and 1.6 mm were employed.
Instead of bonding them together with an adhesive, a simple screw-based, collar-like
clamping system was devised, which could be slid over the joint section, i.e. both adherends
were placed together, as though bonded, and secured at aD equivalent to crack-length,
measured directly with calliper gauge (see Fig.3.5). This technique permitted both, the
fabrication of artificial joints, described here, by the clamping of unbonded aluminium plates
at a desired value of aD, before the test, and also the reconstitution of bonded samples, either
partially or totally separated during prior tests, both to corroborate crack-length evaluation
and check that plastic adherend deformation had not occurred during a test. However, due to
the imposed, straight crack-front parallel to the y axis in this technique, any effects due to
anticlastic bending, or other phenomena leading to non-rectilinear fracture fronts are
necessarily neglected [177]. Similarly, any possible influence of a deformable elastic
foundation ahead of the crack front, or root rotation, is neglected [167,173,193,194].
Fig.3.5. Scheme of the artificial crack tip test.
The artificial crack tip test was performed for the range of the beam lengths, aD from 35 to
105 with 10 mm step, three times for each length for three different AA2024 plates. The
results of this test were used to find out crack length calibration curve (Fig.3.6) and are listed
in Table 3.1.
Force
sensor
LVDT
displacement
sensor
Machine
frame
aD
Moving
clamp
Michał K. Budzik
77
30 40 50 60 70 80 90 100 110
30
40
50
60
70
80
90
100
110
a (
mm
)
aD (mm)
a = 1.0089 aD
R2 = 99.97%
Fig.3.6. CRT calibration curve.
Table 3.1. Results of artificial crack tip test.
Direct crack length
aD (mm)
Estimated crack length
a (mm)
Standard deviation
σd(a)(mm)
35 35.9 0.8
45 46.6 0.5
55 54.4 0.6
65 63.3 0.4
75 75.3 0.2
85 84.9 0.5
95 98.4 1.0
105 108.0 0.4
3.1.3. Fracture of aluminium joints bonded with nanoparticle adhesive
The aim of this study was to compare the micro and macro scale behaviour of the
aluminium/aluminium adhesive joints bonded with two epoxy adhesives: pure and reinforced
with clay nanoparticles. Moreover, the focus was on the novel use of the constant (deflection)
rate test to study adhesion/adhesives efficiency. Nanoparticle reinforcement is a new trend in
the manufacturing of advanced adhesives with improved fracture resistance [195-197]. The
reason to use clay reinforcement is that the particles are natural, thus ecological, moreover
they are cheapest available nanoparticles. In addition density of MMT is much less than those
of metallic or ceramic particles, and are likely to increase strength of the joints [198,199].
Materials and specimen preparation
Aluminium alloys used were: AA2024 (Al-Cu) (upper, flexible plate, h=1.6 mm) and
AA5754 (Al-Mg) (lower, rigid plate, H=6 mm). The plates were bonded along ladh=70 mm
FRACTURE IN ASYMMETRIC BONDED JOINTS
78
with two adhesive systems. First was epoxy resin - Epidian®6, the second system was the
same resin reinforced with 5% of 1-D montmorillonite nanoparticles. Prior to bonding plates
were degreased, abraded, and electrochemically treated with PAA. Bondline thickness for all
samples was 600±40 μm (measured with optical microscope). Before the test samples were
precracked by insertion of the wedge between two bonded plates. The force sensor was
Instron with 200 N capacity and LVDT displacement sensor Peltron PSzl. The temperature
during tests was 20±2oC, and ca.55±5% RH. Two displacement/deflection rates were used:
vD=0.485 mmh-1
- fast, and vD=0.248 mmh-1
- slow with the corresponding numbers I and II
respectively for both adhesive systems (Fig.3.7). In addition, macro (Digital Camera Canon
D40), microoptical (Leica MEF 4M with Canon Power Shot G5) and SEM views of fractured
surfaces were made.
0 1 2 3 4 5 6 7 8 9 10 11 12
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Dis
pla
cem
en
t,
(m
m)
Time, t (h)
Epoxy I
Epoxy II
Epoxy + Nano I
Epoxy + Nano II
vD = 0.485 (mm h
-1)
vD = 0.248 (mm h
-1)
Fig.3.7. Specimen deflection vs. time for two adhesive systems and two displacement rates.
Macromechanics of fracture
Crack growth rate in the adhesive joints was evidently dependant on the adhesive system used
to bond the plates. Crack propagation plots are shown in Figs.3.8a and 3.8b. To be noted are:
nonlinear graph profile (Fig.3.8b) followed by quasi-constant crack growth (vc=const.) and
distinct difference in the behaviour of the two material systems at lower displacement rate
(constant crack growth rate regime). Crack growth rate of the Epoxy+Nano II (lower
displacement rate) specimens is markedly lower than one of the pure resin samples. This was
also found during macrographic studies of fracture surfaces. Three distinct crack speed
sections were found (see Fig.3.9). The advantage of the epoxy system strengthened with
nanoparticles over pure epoxy adhesive is illustrated in Figs.3.10. and 3.11. showing the
average fracture energy and crack growth rate vs. strain rate for the two adhesive joints
studied in this work. The difference is ca. 30% in terms of fracture energy in favour for MMT
reinforced adhesive. In addition, crack rate is in this case smaller, thus using of the MMT
reinforcement results in better joint performance.
Michał K. Budzik
79
Fig.3.8. Crack increment vs. time plots. Nonlinear graph profile (in square) followed by quasi-
constant crack growth (vc~const) (left). Nonlinear part of the graph at the onset of crack
growth (t=0-1.5h) (right).
Fig.3.9. Macrographic view of fracture surfaces: River patterns regions (inside white boards)
and three distinct crack speed sections indicated with white arrows: I – fast crack speed
(probably during wedge insertion), II – medium – decreasing crack speed (probably nonlinear
part of Fig.3.8), III – stable crack growth (linear part of Fig.3.8).
Fig.3.10. Average minimum fracture energy for aluminium bonded with reinforced and pure
epoxy adhesive (left).
Fig.3.11. Crack growth rate vs. deflection rate for the two systems: with and without
nanoparticles (right).
0
20
40
60
80
100
Fra
ctu
re e
nerg
y, G
Ic m
in (J
m-2
)
Epoxy + Nano
Epoxy
0.25 0.30 0.35 0.40 0.45 0.50
2.6
2.8
3.0
3.2
3.4
3.6
3.8
4.0
4.2
4.4
4.6
4.8
5.0
vc (
mm
h-1
)
vd (mm h
-1)
Epoxy
Epoxy + Nano
I I
I III
CRACK DIRECTION
25 m
m
70 mm
Riv
er p
att
erns
regio
ns
0 1 2 3 4 5 6 7 8 9 10 11 12
0
5
10
15
20
25
30
35
40 Epoxy I
Epoxy II
Epoxy + Nano I
Epoxy + Nano II
Cra
ck in
cre
me
nt,
a (
mm
)
Time, t (h)
0 1
0
5
10
15
Epoxy I
Epoxy II
Epoxy + Nano I
Epoxy + Nano II
Cra
ck in
cre
me
nt,
a (
mm
)
Tme, t (h)
FRACTURE IN ASYMMETRIC BONDED JOINTS
80
Micromechanics of fracture
The illustration of fracture surfaces are shown in Figs.3.12 and 3.13 for the two epoxy
systems at lower and higher strain rates. The centre of the specimens was selected for optical
examination. No difference in the fracture surface profile of the two adhesive systems was
noted for the same strain rate. However, there is a marked difference in the appearance of
fracture surfaces depending on the crack growth rate for both materials. The higher the
deflection rate the coarser the fracture surface which implies that microstructural effects are
more present at lower deflection rate (fine fracture surface). Accordingly, very slow
deflection rate is recommended in further study of microstructural effects. The explanation is
time dependant nature of the mechanical properties of epoxy-based materials. Excessively
high deflection rate does not allow micromechanical effects to take place at least from optical
microscope observations. Therefore SEM studies using 200x-4000x magnification were
carried out, where the differences are more pronounced. Results of these observations are
shown in Figs.3.14 and 3.15 Brittle fracture was found in all of the observations. Noteworthy
is the difference in fracture surfaces between pure epoxy and MMT modified resins where the
former gives bigger active fracture surface area due to visible 1-D nanoparticles and thus
additional surface development. It has to be pointed out that fracture surface depends on crack
speed, and thus at higher speed giving smooth, almost glassy surface while rough when the
crack speed is low.
Fig.3.12. Optical micrographs of fracture surface of E6 resin, after lower (left) and higher
crack growth rate (right).
Fig.3.13. Optical micrographs of fracture surface of E6 with nanoparticles, after lower (left)
and higher crack growth rate (right).
Michał K. Budzik
81
I – FAST
II – TRANSITION
III - STABLE
Fig.3.14. SEM fractographs of pure DGEBA epoxy for different crack speeds regimes (in
columns) and for different magnifications (in rows: 200-500-1000x).
FRACTURE IN ASYMMETRIC BONDED JOINTS
82
I – FAST
II – TRANSITION
III - STABLE
Fig.3.15. SEM fractographs of epoxy reinforced with 5% of MMT 1-D nanoparticles for
different crack speeds (columns) and magnifications (rows). Arrows indicates nanoparticles.
In addition, microscopic studies disclosed some interesting phenomena, difference between
fracture surfaces at the sample side and in the middle of the sample. Difference was
particularly marked in the fast crack propagation regions (see Fig.3.8). The fractographs
Michał K. Budzik
83
indicates possible high anti-plane shearing stress and thus mode mixity regions near the
sample side edges which results in river patterns (see Figs.3.15 and 3.16).
Fig.3.16. SEM fractographs made at the middle of the sample (left) and near the side edge
(right). Arrows indicates features of the river patterns phenomena.
Fig.3.17. Side view of Fig.3.16 made using optical microscope.
3.2. Constant Force Test (CFT)
Constant Force Test (CFT) is a new test introduced within the thesis. Differences between
CFT and the standard DCB test are in sample geometry and loading conditions. In this test
only flexible substrate deflection, Δ, is recorded during the test, the applied force, F, is kept
constant. The test setup was designed in Catia v5 (Dassault Systems, Vélizy-Villacoublay,
France) environment using Mechanical Design tool. The schematic illustration of the test and
its principle are shown in Fig.3.18. The rigid adherend (1) is fixed to the frame using screws.
The flexible adherend (2) is connected by a hinge, placed at the one of the flexible plate
extremities (see Fig.3.2), with the mass (4) by a steel chain (Ф4 mm). A displacement sensor
(3) is mounted to the frame so that the sensor tip is touching the flexible adherend exactly
under applied force. This permits direct plate deflection measurements in the point of the
highest deflection, in addition simplifying data reduction.
CRACK DIRECTION
100 μm
FRACTURE IN ASYMMETRIC BONDED JOINTS
84
a = f(t)
x
z
F = const
Δ = f(t)
Fig.3.18. Schematic representation of the CFT test.
The test begins by hanging an appropriate mass, thus applying constant force, F. When the
strain energy stored in the flexible substrate achieves the critical value of the fracture energy,
the crack begins to propagate, in CFT, initially at a slow rate. The crack accelerates with time
as the crack length, a, increase and thus bending moment grows. The crack position is
manifested by the change of deflection of the flexible substrate, Δ which is the only variable
to be measured. Deflection was measured with 10 mm range RDP DCTH 200 AG LVDT
(Linear Variable Differential Transformer) sensor (RDP Electronics Ltd., Wolverhampton,
UK). The block scheme of the built metrological circuit is illustrated in Fig.3.19. In addition
the CFT test, in principle, allows full characterization in one test of the crack speed, v vs.
fracture energy, GC.
Fig.3.19. Block scheme of metrological circuit.
3.2.1. Data reduction method
In the CFT test crack propagation is manifested by the change of the flexible plate deflection,
Δ. The physical model of the test, shown in Fig.3.20, is based on the cantilever beam. The
Euler-Bernoulli simple beam theory (SBT) mathematics was used for data reduction.
Fig.3.20. Constant Force Test physical model.
PC
Δ=f(t)
Displacement
sensor, Δ
(V) A/D
converter
3
Clamps
4
1
2
Michał K. Budzik
85
The distribution of the bending moment, M, along x due to the applied force, F is:
(3.19)
The governing equation of the simple beam theory SBT is:
(3.20)
The plate deflection is found from double integration of (3.20) with appropriate boundary
conditions. The boundary conditions used are:
(3.21)
(3.22)
Finally, the plate deflection along the x:
(3.23)
Because the LVDT sensor is measuring the deflection exactly under the applied force, F, it
can be written:
(3.24)
With the second geometrical moment of inertia of the flexible plate section:
(3.25)
We can write the estimate of crack length:
(3.26)
In order to find the fracture energy, the classical Irwin-Kies relation can be used:
(3.27)
FRACTURE IN ASYMMETRIC BONDED JOINTS
86
where C is compliance (Δ/F).
Now using (3.24):
(3.28)
(3.29)
Thus:
(3.30)
Finally, the energy release rate is given by:
(3.31)
It should be noted that the formula is exactly the same as for the Constant Rate Test. Thus the
change is in that F is now constant, which gives the metrological advantage of measuring only
one variable. In addition, the CFT test is compliance (e.g. machine, chains etc.) effect free.
Estimation of the CFT method error was achieved using propagation of errors. Crack length,
a, error:
(3.32)
(3.33)
Error of energy release rate estimation, G:
(3.34)
Michał K. Budzik
87
(3.35)
3.2.2. Artificial crack tip test
To verify assumption about the mathematical and physical model the artificial crack tip test,
described previously (see section 3.1.2) was performed. Schematic presentation of the made
test is illustrated in Fig.3.21.
Fig.3.21. Artificial crack test in Constant Force Test configuration.
Three aluminium AA2024 plates were used. Four different force values were applied. The test
was performed for the range of crack lengths, aD, from 35 to 105 mm with 10 mm step
(measured with calliper gauge). Since plates were used instead of beam elements, equation
3.24 was verified. Results are illustrated in the left of Fig.3.22. The calibration curve is
illustrated on the right of Fig.3.22. All data from the artificial crack tip test are summarized in
the Table 3.2.
Fig.3.22. Linear force – deflection relation (left) and crack length calibration curve (right).
30 40 50 60 70 80 90 100 110
30
40
50
60
70
80
90
100
110
120
a (
mm
)
aD(mm)
a = 1.069 aD
R2 = 99.92 %
0 2 4 6 8 10
0
5
10
15
20
25
30
35
40
F = 3.0356
R2 = 99.86 %F =
4.5
271
R2 =
99.9
8 %
F =
7.0
236
R2 =
99.
82 %
F =
11.6
75
R2 =
99.9
5 %
F =
20.0
7
R2 =
99.7
6 %
areal
= 40 mm
areal
= 50 mm
areal
= 60 mm
areal
= 70 mm
areal
= 80 mm
Fo
rce,
F (
N)
Deflection, (mm)
LVDT
displacement
sensor
aD
Artificial crack tip
Δ
Mass, F
FRACTURE IN ASYMMETRIC BONDED JOINTS
88
Table 3.2. Results of artificial crack tip test for CFT configuration.
Direct crack estimate, aD
(mm)
Estimated crack length, a
(mm)
Standard deviation, σd(a)
(mm)
35 37.3 0.4
45 51.4 0.4
55 58.0 0.7
65 66.6 0.3
75 79.9 0.2
85 90.8 0.9
95 105.3 0.5
105 113.7 0.7
3.2.3. The (macro) fracture behaviour of different adhesives
In order to find out the features of the Constant Force Test when applied to real joints, the test
was used to evaluate fracture properties of different popular adhesive systems. Three different
adhesives were used: cyanoacrylate adhesive, acrylic mastic and PSA double face Scotch®.
The reasons to choose those adhesives adhesive are their different behaviour in loaded
conditions as well as their common commercial use. In addition, they can be competitive in
many fields to epoxy adhesives. Particularly PSA adhesives are nowadays extensively
developed for practical reasons (easy, fast application) and their possible applications in
structural bonding. In fact, PSA adhesives promise such easiness of application that all
engineers wish to have, still their strength and durability must be improved. The aim of this
comparative study of several different adhesives was to consider macro scale behaviour of the
aluminium/aluminium adhesive joints. The focus was also on the novel Constant Force Test
to study adhesion/adhesives efficiency. It must be pointed out that the CFT test and
characteristics obtained are presented for the first time within this work (to own knowledge).
Materials and specimen preparation
Aluminium alloys used were: AA2024 and AA5754. The plates were bonded along ladh=70
mm. Prior to bonding plates were degreased, abraded, and electrochemically treated with
PAA. The bondline thickness, e, for all samples was 200±10 μm. Before the test
cyanoacrylate samples were precracked by insertion of the wedge between two bonded plates.
Mastic and PSA adhesives did not require this operation. The conditions of the test were:
20±2oC, and ca. 55±5% RH.
Crack behaviour
The crack increment, δa vs. time, t characteristics are shown in Fig.3.23. To be noted is the
significant difference in the crack increments between elastic adhesives (mastic and PSA) and
brittle (cyanoacrylate). The crack growth rate in the adhesive joints tested was evidently
Michał K. Budzik
89
dependent on the adhesive system. I is worth noting that the crack growth rate accelerates
almost immediately from the onset of the load.
Fig.3.23. Crack increment for cyanoacrylate (Cyanoacrylate), acrylic mastic (Mastic) and
double face scotch (PSA) adhesives (left).
Fig.3.24. Crack speed characteristics. Vertical lines corresponds to the estimated time of
overall failure due to accelerating crack growth (right).
The observations from crack propagation are manifested in crack speed characteristics. It
must be pointed out that the value of the measurable crack speed is limited by time intervals
used between two measurements. Here measurement were taken every 45 seconds, thus the
maximum measureable speed is δa/45s (in practice ca. 50 mm/h).
Fracture energy
Fracture energy, Gc vs. crack speed, v are the data most required from fracture tests (Fig.3.25)
and are rarely presented in the literature in continuous form. Cyanoacrylate adhesive
possesses the best fracture characteristics, giving smooth, linear curves with small gradient.
The same linear relation can be observed for the mastic adhesive but for 2 times smaller
energy values. The PSA adhesive shows the worst characteristics. PSA was found most
sensitive to crack speed, promising the smallest energy and inhomogeneous energy transfer.
Although fully quantitative studies were not the main interest here, it must appreciated that for
the crack speed range presented, cyanoacrylate adhesive has about 200-260 Jm-2
fracture
energy, with mastic 25-40 Jm-2
and PSA 0.8-20 Jm-2
. This shows the abilities and versatility
of the CFT test in measuring fracture properties of the adhesive joints.
1 10
0.1
1
10
100
Cyanoacrylate
Mastic
PSA
Cra
ck
sp
eed
, v
(m
mh
-1)
Time, t (h)
0.1 1 10
0
5
10
15
20
25
30
Cra
ck
in
cre
men
t,
a (
mm
)
Time, t (h)
Cyanoacrylate
Mastic
PSA
FRACTURE IN ASYMMETRIC BONDED JOINTS
90
Fig.3.25. Fracture energy characteristics for the tested materials.
Nonlinear effects
In the CFT test, nonlinear behaviour and/or elasticity of the adhesive was manifested (see
Fig.3.26).
Fig.3.26. The example of creep behaviour (or adhesive elastic deformation) at the beginning
of the observed in mastic adhesive. On right, close up of elastic foundation region.
This effect can be related to many different molecular phenomenon. From a modelling
standpoint this could be approached with combination of parallel springs, possessing some
vertical stiffness and deforming under any applied load. In the case of a soft adhesive, where
the observed effects are particularly pronounced a process zone (see Fig.3.27) [194,200,201]
is created earlier. Process zone is defined as a zone in front of the crack tip affected by applied
load and in the study was not estimated. This effect should be studied because it can lead to a
wrong estimation of crack length and fracture energy. Basically we consider this zone (of
d2Δ/dt
2<0) as being due to adhesive time-dependent strain, rather than crack growth.
Nonlinear effects at the beginning of the test can be related to phenomena such as: Payne or
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
0
2
4
6
8
10
12
Cra
ck in
cre
me
nt,
a (
mm
)
Time, t (h)
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0.0
0.2
0.4
0.6
0.8
1.0
Experiment
a = 1.047 t0.2403
Cra
ck in
cre
me
nt,
a (
mm
)
Time, t (h)
1 10 100
10
100
Cyano
Mastic
PSA
Fra
ctu
re e
nerg
y,
Gc (
Jm-2
)
Crack speed, v (mmh-1)
Michał K. Budzik
91
Mullins effects, adhesive relaxation or adhesive creep under constant applied force. However,
presented here briefly elastic interpretation of the initial adhesive behaviour which may be
very promising, and could possibly allow direct fracture energy estimation using the classical
Dugdale model.
Fig.3.27. Schematic representation of the process zone.
3.3. Constant Displacement (Asymmetric Wedge) Test
Constant Displacement Test (CDT) is an asymmetric version of the popular Boeing Wedge
Test. The test setup was designed using Catia v5 environment and is schematically illustrated
in Fig.3.28. The rigid substrate (6) is fixed to the aluminium block/wall (7) using screws. The
pin (8) is used to place and assure vertical position of the sample. The wedge (3) of
aluminium of thickness, Δ is pushed by the screw based system (1 and 2) while guides assert
appropriate wedge position. If necessary vertical position of the entire system allows water
immersion tests. The test begins as the wedge is inserted manually to the desired position,
which results in the elastic deformation of the flexible plate (4) and the propagation of crack
continue at its self-determined rate. The test stops when strain energy stored within the
flexible adherend achieve value of the adhesive (5) minimum fracture energy, Gc. Within this
thesis novel crack length measurement method is reported. The upper surface of the flexible
beam was instrumented with series of strain gauges attached along the direction of crack
propagation. Strain gauges used were Vishay Micro-Measurements, reference EA-13-060LZ-
120/E (Vishay, Malvern, PA, USA) of nominal resistance 120Ω. Strain gauges were
connected in Wheatstone quarter-bridge circuit using Vishay Micromesures 2100 System
Multi Channel Signal Conditioner/ Amplifier with ten modules of Model 2120 B Strain
Gauge Conditioner/Amplifier, and one module of Model 2110 B Power Supply.
PROCESS
ZONE
FRACTURE IN ASYMMETRIC BONDED JOINTS
92
Fig.3.28. Schematic representation of the CDT test.
The CDT test metrological routine is shown in Fig.3.29.
Fig.3.29. CDT metrological circuit.
3.3.1. Data reduction method
Cantilever beam physical model and simple beam theory (SBT) mathematics were used to
analysis the Constant Displacement Test (see Fig.3.30).
Fig.3.30. Physical model of CDT.
a(t)
x
R
z Δ=const
Strain gauge Wheatstone
bridge
A/D
converter
PC
ε = f(t)
(Ω) (V)
1
2
3
4
5
6
7
8
4
3
6
Michał K. Budzik
93
Governing equation of SBT is:
(3.36)
With bending moment, M distribution:
(3.37)
Double integration of equation with appropriate boundary conditions yield the solution in
terms of plate deflection. The boundary conditions are:
(3.38)
(3.39)
plate deflection along the x:
(3.40)
Since displacement not force is applied to separate bonded plates, we can write:
(3.41)
Finally we obtain relation between crack length and wedge thickness, Δ:
(3.42)
Energy balance method was used to evaluate the energy release rate, G:
(3.43a)
(3.43b)
where terms corresponds to: work, W, done by force F to propagate the crack (=0), change in
elastic energy, U, due to crack propagation and adhesion energy to create new crack
increment, δa.
FRACTURE IN ASYMMETRIC BONDED JOINTS
94
The elastic energy is given by:
(3.44)
From (3.44) and (3.41):
(3.45)
Now, change of the elastic energy due to propagating crack – the energy release rate can be
found:
(3.46)
with
- geometrical second inertia moment of flexible adherend section.
From (3.43), we obtain (the minus sign is the convention used, since the energy is releasing-
giving out from the systems):
(3.47)
Energy release rate expression in asymmetric CDT:
(3.48)
It has to be pointed out that energy in asymmetric test is different from symmetric by the
factor of 2. In addition, like in the symmetric version of the test, G is very sensible to crack
length, a with scaling of -4th
power.
Crack length and energy release rate estimation from strain measurements
To obtain results in terms of the crack length and fracture energy strain measurements were
performed. Strain in the upper layer of flexible member (absolute value is given since upper
layer is submitted to compression):
(3.49)
With inversed beam radius, R(x)-1
:
Michał K. Budzik
95
(3.50)
Strain can found as:
(3.51)
Thus, only one strain gauge based at xє<0, a> is required to find out the actual crack position
and to measure crack length growth. Due to possible misalignment error e.g. inappropriate
strain gauge bonding, singular strain gauge properties or possible failure of the single gauge
using test can be unsuccessful or lead to misinterpretation. From (3.51) linear relation
between measured strain |εs(x)| and strain gauge position x, can be noted. This allows one
parameter linear regression analysis and thus a series of strain gauges were used.
(3.52)
Therefore, combining the data from the various strain gauges in their different relative
positions (x) with respect to the crack front, an accurate, and potentially continuous
assessment of crack length can be obtained.
Now, criteria for best α, (for n – strain gauges, with i for ith
gauge):
(3.53)
where:
|εs(x)|th - theoretical strain value at x
|εs(x)|exp - experimental value of strain at x.
This yields:
(3.54)
The minimum criteria is kept if, first derivation of the sum over α is equal 0, so:
(3.55)
The best estimation of α can be written:
FRACTURE IN ASYMMETRIC BONDED JOINTS
96
(3.56)
Thus, statistically, best crack length is:
(3.57)
Strain estimation will include the error of Δα and thus:
(3.58)
To calculate error of our estimation propagation of error method was used assuming no error
on x, thus:
(3.59)
with:
yi = |εs(xi)|exp.
We can write:
(3.60)
And:
(3.61)
Hence all y have equal weight, so Δy1= Δy2 =...Δyn=Δy:
(3.62)
The best estimation of the strain standard deviation is given for n-1 strain gauges:
(3.63)
Michał K. Budzik
97
From definition of α:
(3.64)
(3.65)
(3.66)
So:
(3.67)
And finally the estimation error, Δα:
(3.68)
Crack length error will be therefore:
(3.69)
So:
(3.70)
The energy release rate error:
(3.71)
Finally:
FRACTURE IN ASYMMETRIC BONDED JOINTS
98
(3.72)
3.3.2. Calibration of the CDT test using artificial crack tip test
Artificial Crack Tip Test technique was employed in order to estimate the accuracy of the
strain gauge technique developed here, without using an actual adhesive joint. Strain gauges
were bonded in place along the central line of the thin adherend, at values of x of 16, 26, 36
and 46 mm (Fig.3.31).
Fig.3.31. Artificial crack tip test principle.
Strain measurements were made for two wedge thicknesses, Δ - 4.6 and 9.7 mm, and six beam
lengths, aD: from 35-100 mm (measured directly using calliper gauge). The example of the
crack lengths estimation by a single gauges, using eq.3.51 are shown in the left of Fig.3.32.
On the right of the Fig.3.32 crack estimation-strain gauge position linear relation is shown. In
addition, eq.3.52 was verified. Results are illustrated in Fig.3.33 (left). Experimental results
are summarized in Table 3.3, and are shown in Fig.3.33 (right) as a crack length calibration
curve.
Screw system
a
Clamps
x1 x2
x3
x4
y
x
z
x
Michał K. Budzik
99
Fig.3.32. Verification of eq.3.51 (left). Estimated crack length value, a vs. strain gauge
position, x relation (right).
Fig.3.33. Verification of eq.3.52 (left). Crack length calibration curve, a vs. aD (right).
Table 3.3. Results of the crack length estimation with artificial crack tip test.
Direct measurement, aD
(mm)
Strain gauge estimate, a
(mm)
Standard deviation, σd(a)
(mm)
35 33.7 0.3
50 49.2 0.4
65 63.8 0.4
75 74.9 0.4
85 86.9 0.4
100 101.2 0.5
3.3.3. Surface treatment effect in CDT test
In order to verify the CDT method two different surface treatment of the aluminium plates
were employed. The aluminium surfaces were prepared either by simple abrasion or by
electrochemical treatment. Bondline thickness was maintained at ca. 0.35. Five longitudinal
strain gauges were fixed. The values of x1 to x5 were respectively 35, 45, 55, 75 and 85 mm.
Initially, two strain gauges are within the limits of the bonded region, and three without, as is
15 20 25 30 35 40 45 50
0.0003
0.0004
0.0005
0.0006
0.0007
0.0008
0.0009
0.0010
0.0011
0.0012 Experimental points
|s(x)|= 2.53668* 10 -5
x
R2
=0.99986
Ab
solu
t st
rain
, |
s(x)|
Gauge position, x (mm)
= 4.54 mm
a = 75 mm
30 40 50 60 70 80 90 100
30
40
50
60
70
80
90
100
a (m
m)
aD (mm)
a = 1.0017 aD
R2 = 99.96%
40 42 44 46 48
74.0
74.1
74.2
74.3
74.4
74.5
74.6
74.7
74.8
74.9
75.0
75.1
75.2
x1 = 16 mm
x2 = 26 mm
x3 = 36 mm
x4 = 46 mm
Cra
ck l
eng
th,
a (m
m)
Time, t (s)
15 20 25 30 35 40 45
74.0
74.2
74.4
74.6
74.8
75.0
Cra
ck
len
gth
est
imati
on
, a (
mm
)
Strain gauge position, x (mm)
a = 0.0227x + 73.7388
R2 = 99%
FRACTURE IN ASYMMETRIC BONDED JOINTS
100
shown in Fig.3.34. The wedge thickness, Δ was 4.5 or 9.7 mm. Two tests for each of the two
surface treatments were effected at 20 2°C, and at an ambient humidity of ca. 55% RH.
Fig.3.34. Schematic representation of tested sample.
Crack propagation
The evaluation of crack propagation kinetics can be virtually continuous with this strain gauge
technique, leading to detailed crack monitoring. Fig.3.35 represent crack length increment, δa,
vs. time, t, for the abraded and the PAA treated wedge tests.
0 50 100 150 200 250 300 350 400 450 500 550 600
0
2
4
6
8
10
Anodized
Abraded
Cra
ck
in
cre
men
t,
a (
mm
)
Time, t (h) Fig.3.35. Crack length increment, δa with time, t.
Δ
85
35
45
55
75
a(t0)=67 mm ladh(t0)=33 mm
y
x
z
x
R~εs-1
Michał K. Budzik
101
Noise on the measurement (electric) of a is typically less than 0.1 mm, and at worst less than
0.2 mm. This noise has been removed for aesthetic reasons, since it only essentially thickens
the graphic line. Error bars, shown for various values of time, t, were calculated using (3.69)
and those associated. As can be seen, estimated errors are similar for the two treatments and
reasonably independent of crack length, being of the order of ±0.5 mm. The general features
of wedge tests results are present in both cases, viz. continuously increasing crack length, a(t),
with time, t, and concomitant decrease of da(t)/dt, as strain energy release rate diminishes.
This is a typical trait of adhesive fracture, as first exemplified with elastomers many years ago
[e.g. 202,203]. Despite these overall trends, there are second order changes in gradient, with
slight ups and downs, which are presumably related to a degree of inhomogeneity of the
adhesive joint. Such effects are rarely reported in the literature, presumably quite simply
because most techniques for measuring crack length cannot be employed in continuous mode,
in contrast to the present method. Other advantage of using strain gauge technique is
possibility to check if any plastic deformation of the substrate occurs during the test. This will
be manifested by non-zero strain signal after finished test. The reconstituted joint led to
reproducible surface strain values, and did not reveal any permanent set. The most dangerous
situation is at the beginning of the test, when the crack length/wedge thickness ratio is the
smallest. Of course, as the crack progresses, this value, corresponding to the value of surface
strain at the moving crack front, quite rapidly decreases. Notwithstanding, e.g. for the PAA
treated joint, is ca. 0.26, which is well below the yield limit of aluminium 2024 alloy.
In a correctly designed test procedure, this potential exceeding of the elastic limit of the
adherend should, needless to say, be avoided.
Fracture
Fracture, or adhesion energy, Gc, was evaluated for continuously increasing a(t) using (3.48).
Corresponding crack growth rates, d(a)t)/dt, were also calculated from tangents to the a vs. t
curves. In Fig.3.36, Gc vs. v=d(a)t)/dt results are presented for the two systems corresponding
to simple abrasive of substrates and PAA treatment. The calculated error bars include
systematically errors on E, h, Δ as well as random errors. This is to show the potential
precision on Gc obtainable from crack length measurement with the strain gauge method
presented here. Without these errors error bar is of the size of the experimental point. Various
observations may be made. Firstly, overall the fracture energies are relatively low for this type
of joint. However, the adhesive used was a general-purpose, pure epoxy resin, and not a
specialised material (modified etc.), so this is not too surprising (see Table 1.2). Secondly,
clearly the PAA surface treatment gives a considerably higher fracture energy than simple
abrasive treatment, for any given crack growth rate. At the lower end of the rate scale studied,
fracture energy is ca. 80% greater for PAA, at the other extreme, ca. 20%. This suggests that
the relation between effective energy of adhesion and surface treatment is not simply
multiplicative, as often suggested for elastomers [202,203]. Superficially at least, fracture
appeared to be interfacial for the abraded aluminium surfaces, but distinctly cohesive within
the adhesive for the PAA pre-treatment. Again, this is not surprising as good adhesion to
aluminium generally requires adequate surface treatment. The results on real joints in this
context are presented to corroborate findings with the artificial wedge tests.
FRACTURE IN ASYMMETRIC BONDED JOINTS
102
1E-4 1E-3 0.01 0.1 1 10
60
70
80
90
100
110
120
Fra
ctu
re e
ner
gy
, G
c(J
m-2
)
Crack speed, v (mm h-1)
Abraded
Anodized
Fig.3.36. Fracture energy vs. crack speed characteristic of Al/Al bonded plates with pure
DGEBA resin after different surface treatment.
3.3.4. Adhesive joint with variable adhesion properties
Strain gauge method developed has been extended, in order to investigate crack growth
behaviour in the case of variable adherend surface pre-treatment. The basic idea is to alter the
surface treatment of one and the same adherend, but in separate zones, prior to bonding, with
a knowledge that one treatment is likely to produce markedly better adhesion than the other.
Simple measurement of the fracture energy on each surface is relatively easy and should hold
no surprises. Nevertheless, the fact that it is possible to vary surface pre-treatment on one and
the same adherend could prove useful in order to obtain more reliable comparative data on the
quality of adhesion different surface preparations and/or ageing conditions, eliminating, or at
least reducing, experimental scatter due to the use of separate joints (individual, slightly
different adherends, reproducibility of curing temperature and pressure, etc.). However,
principal aim here was to consider crack behaviour near the transition from one surface pre-
treatment to another.
In this study CFRP composite material with Eh=50 GPa was bonded to an aluminium plate to
form the joint. Adherends were bonded along 105 mm, as shown in Fig.3.37. Two strain
gauges were attached to the outer surface of the flexible, composite plate, along the centre line
and in the positions shown. The values of x1 and x2 were respectively 15 and 25 mm. The
figure suggests that one strain gauge initiates within the bonded length of the assembly, but in
practice, the adhesive bond was pre-cracked from the wedge end, so both gauges were
effectively in the de-bonded section. The adhesive used was a commercial epoxy resin
Araldite Cristal. Bondline thickness was maintained at 350 25 μm. Two different surface
Michał K. Budzik
103
treatments of the aluminium were used prior to bonding. In both cases, the aluminium to be
bonded was abraded with 1200 grade emery paper. This was the only treatment for parts of
the surface, and is represented by P (polishing). These parts were then carefully protected
with adhesive tape, and the remaining surface to be bonded was sandblasted, using SiO2 grit
(referred as SB). Finally, an aluminium wedge, of thickness Δ of 3 mm, was inserted. Tests
reported here were based on two separate assemblies, results being found reproducible.
Experiments were effected at 20 2°C, and at an ambient humidity of ca. 55% RH.
Fig.3.37. Geometry of asymmetric wedge test sample with strain gauges (dimensions in mm).
(a) Side view. (b) Top view showing position of the strain gauges. (c) Top view showing the
relative position of the wedge and one of the zones of sandblasting.
Crack length and fracture energy
Fig.3.38 represents an example of crack length, a, vs. time, t, for the composite/aluminium
assembly whilst fracture is occurring in the vicinity of the SB/P transition. It is usual in wedge
tests to observe an asymptotically decreasing crack speed, da/dt. This is due to a constantly
decreasing (strain) energy release rate, G. However, Fig.3.38 presents three distinctive
sections: from t=0 to ca. 20 hours, the usual decrease in da/dt is observed, and then from ca.
20 to ca. 40 hours, crack growth accelerates. Finally, from ca. 40 hours onwards, da/dt
decreases again, albeit with a couple of minor oscillations. These results are perhaps more
clear when presented as da/dt vs. t., as shown in Fig.3.39. The initial high crack speed near
t=0, and due to a relatively high value of G in the early stages of fracture, becomes attenuated,
only to pick up again from ca.20 hours, with a peak of ca. 0.6 mmh-1
near t=40 hours.
Thereafter, the crack speed again decays.
15
25
Δ
ladh(t0)=105 mm a(t0)=30 mm
Composite The adhesive
Aluminium
SB P
Wed
ge
Approximate Zone of Interest
y
x
z
x
a)
b)
c)
FRACTURE IN ASYMMETRIC BONDED JOINTS
104
Fig.3.38. Crack length, a, vs. time, t, for the composite/aluminium assembly in the vicinity of
the transition from SB to P surface treatment. The transition is occurring between ca. 20 and
40 hours (left).
Fig.3.39. Results of Figure 3.38 expressed as crack speed, da/dt, vs. time, t (right).
Clearly this ―U‖ shape of the da/dt vs. t plot is related to the surface treatment transition. In
order to understand the basic physics of what is occurring, we propose the following
explanation. Although the wedge test is usually considered to be 2 dimensional, in fact there
are some non-negligible 3D effects [177,204]. This phenomenon is essentially related to
anticlastic curvature of the bent adherend. We therefore assume, in the present case, that the
crack front is curved (convex towards the intact side of the fracture front) and may be
approximated to a circular arc of low profile, such that we may write , using the
nomenclature of Fig.3.40, R being the radius, δ the depth of the crack front, and b the joint
width, as before. The arc subtends a (small) angle of 2α. Consider the line AB, which cuts the
arc in two places and represents a transition between surface treatments of the aluminium
surface, sandblasting, SB, and polishing, P.
Fig.3.40. Model for curvilinear crack front traversing the frontier between SB and P treated
aluminium.
b
δ
y
SB P
A
B
R
α θ
aF*
a*
0 20 40 60 80 100 120
30
31
32
33
34
35
36
37
38
Sandblasted
Transition
Polished
Cra
ck
len
gth
, a (
mm
)
Time, t (h)
0 20 40 60 80 100
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Sandblasted
Transition
Polished
Cra
ck
sp
eed
, v
(m
m h
-1)
Time, t (h)
Michał K. Budzik
105
We take the SB treatment to be on the right hand side of AB. The central part of the arc, to the
left of the intersections with the line AB in the figure, subtends an angle 2α, where
. It is thus readily shown, for , that to a good approximation:
(3.73)
where represents the distance along the x axis between the wedge and the SB/P transition,
and is crack length, the asterisk denoting that we have, arbitrarily, defined crack length as
being the distance between the wedge and the fracture front taken at the joint edges. In other
words, the (projected) width of the crack front on the left of AB is given by
, with the remainder,
, on the right of AB, and thus,
by hypothesis, the former is in the P zone and the latter in the SB zone. With and
representing respectively fracture energy in the region pre-treated by sand-blasting and that
following simple polishing, we can write an expression for the (average) fracture energy, ,
whilst the crack front finds itself cut by line AB:
(3.74)
which is a function of crack length, , such that
and
, with intermediate values of
for
. Generally, as the crack
length, , increases, growth rate decreases since energy release rate, G ~ a-4
. However,
as the crack front encroaches on the zone of simple, polished surface treatment, the intrinsic
fracture energy decreases, since , such that for a given value of energy release rate
G, crack speed should increase, other things being equal. Thus, there are two antagonistic
effects, both being exacerbated by increasing a. If the decrease in intrinsic is more
significant than the effect of increased crack length, the crack will accelerate, contrary to the
behaviour of classic wedge tests, which decelerate due to diminishing G. Thus a graph of
crack length, a, versus time, t, will change from concavity towards the time axis to convexity,
and the junction of the concave and convex sections of the curve, at an inflexion point where
or possibly at an abrupt change of gradient
corresponds to
. Assuming that δ remains a constant, when , the crack front is totally on the
polished surface. Henceforth, is no longer a mixture of and , albeit smaller, and
simply equals . Thus crack growth rate will again decrease monotonically. Again, an
inflexion point may be expected. From these two inflexions on the graph of a vs. t, it should
thus be possible to estimate the depth of the curved crack front, δ. If the crack front negotiates
a transition from bonding on the polished treatment, P, to that on the sandblasted treatment,
SB, one may expect a similar effect. However, unfortunately, the added effect of fracture on
the sandblasted surface, thus leading to higher , will only decrease the crack growth rate
FRACTURE IN ASYMMETRIC BONDED JOINTS
106
even further than that due uniquely to increasing a, viz. G ~ a-4
. Thus, at best, one may expect
a slight reduction in da/dt, but no change in overall features of the curve from concavity to
convexity. Returning to Figures 3.38 and 3.39, we can see that the scenario described above,
and the scheme of Figure 3.40 can explain the observed features of crack speed, da/dt, vs.
time, t. Estimation of the positions of the inflexions in Fig.3.39 leads to two corresponding
values of a, and by difference, we estimate the crack depth, δ, to be ca. 2.3 mm. (Note that we
cannot reasonably estimate an error on da/dt, since the error bars correspond to position and
not to gradient). The value of 2.3 mm is quite plausible and entirely consistent with values
found earlier on a similar system [177]. Further work on a system with a transparent adherend
is envisaged, in order to corroborate this effect, although results from [177] already lends
credibility. However, the system will necessarily be different due to (at least one) different
surface treatment.
Fracture energy assessment
The data of Fig.3.38 have been used to calculate the fracture energy, , of the
aluminium/composite assembly in the vicinity of the surface treatment transition, where no
distinction is made as to whether this quantity is pure or , or whether (3.74) is
applicable. This has been done using the now standard equation for fracture energy, obtained
from an asymmetric wedge test, in which one adherend may be considered to be rigid viz.:
(3.75)
The evolution of vs. time, t, is shown in Fig.3.41.
0 20 40 60 80 100 120
125
150
175
200
225
250
275
300 Sandblasted
Transition
Polished
Fra
ctu
re E
nerg
y, G
c (J
m-2
)
Time, t (h)
Fig.3.41. Fracture energy, , vs. time, t, in the vicinity of the transition zone.
Michał K. Budzik
107
As in Fig.3.38, the two representative error bars at 0 and 100 hours correspond to estimates of
systematic errors on a, obtained using propagation of errors theory. Also as before, relative
errors, neglecting possible misestimates of Eh, Δ and h are far smaller, and would be difficult
to discern in the figure. Fig.3.41 shows anticipated behaviour: a decrease of with time as G
diminishes with crack growth, followed by a steeper decrease as the crack transfers from the
SB to the P treated aluminium surface. Finally, a further decrease with time occurs, but at
much lower values of and at a lower rate of change, because the crack front is entirely in
the P region.
Fig.3.42. Fracture energy, , vs. crack speed, v=da/dt, in the vicinity of the transition zone.
For , f=1 (I) and
for, f=0 (II). The intermediate cases (examples)
correspond to f=0.54 and 0.27, as shown schematically below (III).
Fig.3.42 represents fracture data in the more conventional manner, viz. vs. crack speed,
v=da/dt. The data at the top and at the bottom (accentuated by solid lines in the figure) follow
1E-4 1E-3 0.01 0.1
125
150
175
200
225
250
275
300 Sandblasted
Transition
Polished
Fra
ctu
re e
nerg
y, G
c (J
m-2
)
Crack speed, v (mm h-1)
I
III
II
Sandblasted Polished
I III
II
Crack growth
FRACTURE IN ASYMMETRIC BONDED JOINTS
108
the expected monotonic increase, which is well known for polymeric adhesives [201,205].
However, at first sight, this portrayal is surprising, since appears to be three-valued in
places. Indeed it is. The upper curve, or maximal fracture energy for a given fracture rate,
may be interpreted as the fracture energy, , and corresponds to fracture purely in the
SB region. Similarly, the lower curve is attributable to failure in the P zone, where
(minimal energy). The data points in the intermediate region apparently indicate a decrease in
fracture energy with increasing crack speed, which is highly unlikely under the present
conditions of stable crack growth. In fact, each data point in this region corresponds
effectively to a different type of assembly. The reason for this is that at each point, the relative
contribution from each of and is different, as given by (3.74) (or possibly a more
accurate version thereof). Since the and curves appear reasonably parallel, it is
reasonable to suppose that through each point, we should be able to pass a curve parallel to
the maximal and minimal curves. This has been done in Fig.3.42 for two cases to demonstrate
the principle (heavy dotted lines). To clarify the situation, we may write (3.74) in a simpler
form as:
(3.76)
where f represents the fraction of surface (failure front) SB involved in the separation, and
therefore (1 – f ) corresponds to the fraction of P. Let us consider (arbitrarily) the crack speed
of v=0.1 mmh-1
. As shown by the (light) dotted lines on the graph in Fig.3.42, this
corresponds to values of and respectively of ca. 265 and 152 Jm-2
. The equivalent
experimental value of for v=0.1 mm h
-1, corresponding to the intermediate case, in the
transition zone, and also shown with a dotted line, is equal to ca. 225 Jm-2
. Alternatively, we
may take the known value of joint width, b, the estimated value of crack front depth, δ (see
above), and together with (3.74) and (3.76) a knowledge of crack position, estimate f (ca.
0.54). Accepting the experimental values and above, (3.76) predicts a value of the
intermediate of ca. 215 Jm
-2. The procedure may be repeated, but for a crack speed, v, of
0.3 mmh-1
, and we find experimental and predicted values of respectively of 205 and 200
Jm-2
(f=ca. 0.27). (The relative data are also indicated by dotted lines in Fig.3.42). Given the
simplicity of the model for crack front shape, the agreement is really quite acceptable.
Fracture surfaces
The results presented in Fig.3.43-3.45 are a visual assessment of fracture surfaces, and
corresponds to a photograph of the side view of a composite/aluminium joint after fracture, in
the vicinity of the transition zone from SB to P treatment. The crack proceeded from the right
towards the left in the photograph. It is clearly visible that the fracture surface changes
radically at the frontier between the two surface treatments. The thin band between the
composite and the aluminium is the adhesive layer, which adheres to the latter when SB
treated, separation occurring at, or near, the adhesive/composite interface.
Michał K. Budzik
109
Fig.3.43. Photographs of side of fractured joint in the vicinity of the transition zone between
sandblasted (SB) and polished (P) surfaces. The fracture front comes from the right.
Fig.3.44. Photographs of side of fractured joint in the vicinity of the transition zone between
polished (SB) and sandblasted (P) surfaces. The fracture front comes from the left.
Fig.3.45. SEM micrographs of the side of surfaces near the fracture zone in the SB treated
region. (a) The macroscopically interfacial failure at the adhesive composite interface is in
fact a cohesive failure within the adhesive, but near the interface, whereas (b), the
adhesive/aluminium interface remains intact.
ALUMINIUM
COMPOSITE
ADHESIVE
ADHESIVE (a) (b)
COMPOSITE COMPOSITE
ADHESIVE
ALUMINIUM
ADHESIVE
ALUMINIUM
P SB P SB
New crack
initiation Initial crack
side
ALUMINIUM ALUMINIUM SB P SB P
COMPOSITE
COMPOSITE
ADHESIVE ADHESIVE
2 mm 2 mm
FRACTURE IN ASYMMETRIC BONDED JOINTS
110
However, when the fracture front enters the region of P surface treatment, there is a relatively
rapid deviation of crack path, with separation occurring at, or near, the adhesive/aluminium
interface. Opposite situation occurs when the crack is coming from polished to sandblasted
region, like in Fig.3.44. This suggests strongly that the SB treated aluminium presents better
adhesion to the adhesive than does the composite, but that the P surface has poorer adhesion.
This may be expected from the nature of the surface treatments, but corroboration from
energetic considerations follows a description of the transition behaviour. Fig.3.45 shows
details (pictured from the side) of the fracture zone near the SB treated aluminium, obtained
by scanning electron microscopy (SEM). It is clear that the macroscopically interfacial failure
at the adhesive/composite interface is, in fact, a cohesive failure within the adhesive, but near
the interface. This weakness in the interfacial region, or interphase, was first proposed by
Bikermann in the context of a weak boundary layer (WBL) [67], and later discussed by
Sharpe and also others [206,207]. It is also clear that the adhesive/aluminium interface
(interphase?) remains intact, when the latter has received the SB treatment.
Direct observation of double crack curvature phenomenon
Direct observation of the phenomenon of the double crack curvature was studied using
transparent flexible adherend - polycarbonate plate (PC). Variable, promising different
surface properties, surface treatment of aluminium plate was used prior to bonding. Primarily
aluminium was polished. This was the only treatment of the surface represented by P in
Fig.3.46.
Fig.3.46. Scheme of the experiment principle.
Camera
20 20 20 20
Camera
ladh(t0) = 80 a(t0)=35
PC
Aluminium
Δ =
5
z
x
y
x
P PAA P PAA
Michał K. Budzik
111
Finally, PAA treatment was made on entire sample protecting polished regions with
waterproof scotch tape, thus obtaining variable treatment. The anodized surface regions are
represented by PAA. After removing the tape, both PC and aluminium were lightly rinsed
with C2H5OH before bonding. Both plates were than bonded with epoxy adhesive, Araldite
Cristal plate leaving initial separation (crack length), a=35 mm. Measured adhesive thickness,
e was 0.4 mm. Two tests were run, thus 8 transition zones were observed. Two aluminium
wedges were used. The wedge thickness, Δ was changed during the test from 5 to 9 mm to
force crack propagation towards all transition zones. The scheme of the experiment is shown
in Fig.3.46.
Crack propagation in the vicinity of transition zone
Crack propagation was recorded with Canon D40 photo camera (from the side view) and with
micro USB camera Dino - top view. The interval time between two camera frames was set at
15 min. Experiments were effected at 23±2°C, and at an ambient humidity of ca. 55% RH.
Results are shown in Figs.3.47 and 3.48.
Fig.3.47. Crack propagation in the vicinity of the transition, STRONG/WEAK, zone. Crack is
coming from anodized (PAA) surface to polished (P). Arrows indicates regions of crack front
initiation.
Fig.3.48. Crack propagation in the vicinity of the transition, WEAK/STRONG, zone between
polished (P) and anodized (PAA) surfaces. Arrows indicates regions of crack front initiation.
PAA
P
CR
AC
K
P
PAA
CR
AC
K
FRACTURE IN ASYMMETRIC BONDED JOINTS
112
Based on results from direct observations, as well as from previous measurements, model of
crack propagation in the vicinity of transition zone was proposed (see Fig.3.49). Crack
propagation stages are detailed in Table 3.4.
Fig.3.49. Stages of crack propagation in the sample with variable adhesion properties.
Table 3.4. The crack propagation stages in the vicinity of transition zones.
a) Crack nucleation in STRONG* adhesion zone
b) Crack propagation, with increasing depth of the crack
c) STRONG – WEAK transition
d) Beginning of crack propagation in the WEAK**
zone with higher speed. Decrease of
crack curvature in the STRONG zone
e) Propagation of double crack curvature with an increase of curvature in the WEAK
zone
f) Crack establishment in the WEAK zone
g) Propagation of the crack in the weak zone with higher speed then in the STRONG
zone and lower energy consumption
h) Crack arrests close to the WEAK – STRONG transition. Crack propagation requires
new energy portion – e.g. increase of wedge thickness
i) Decrease of crack curvature without propagation of the front
j) Nucleation of the crack in the STRONG adhesion zone from almost straight crack
front
k) Beginning of crack propagation in STRONG zone
l) Normal propagation of the crack *STRONG – e.g. PAA or sandblasted,
**WEAK – e.g. polished
a) b) c) d) e) f)
g) h) i) j) k) l)
STRONG
WEAK
STRONG
Michał K. Budzik
113
Behaviour of crack front near transition zones
The behaviour of the crack front of the crack of length, a, vs. time, t, near the transition zone
separating strong adhesion S (e.g. sandblasted or anodized) and weak adhesion P (e.g.
polished) surface treatments on the aluminium surface prior to bonding was considered. The
validity of equation 3.74 was assumed describing the effective, or mean overall, fracture
energy, , when the (sharp, linear) transition of surface treatment traverses the curved
fracture front of Fig.3.39:
(3.77)
Fracture occurs when the strain energy release rate, G, is equal to , and the basic relation
for fracture is given by:
(3.78)
where U represents stored elastic energy, b is sample width and a is crack length. In the
asymmetric wedge test, is given by (3.74) which we write as
=ka-4
, where k=3ECΔ2h
3/8,
since it is only variations of a that are presently of interest.
From (3.74) and (3.78), we may write the threshold for fracture as:
(3.79)
Since the analysis leading to (3.74) is based on 2D analysis, it does not allow for any crack
front curvature, which amounts to variability of a=a(y), and there is therefore some doubt as
to what value of a should be used, although a* has previously been defined as the crack length
at the joint edge. Notwithstanding this detail for the moment, (3.79) is differentiated with
respect to time, t, and arranged to give:
(3.80)
where v=da*/dt=da/dt, assuming that it is permissible to equate the two derivatives (an
additive constant between them). For the moment, δ is constant.
FRACTURE IN ASYMMETRIC BONDED JOINTS
114
Now, when the crack front first reaches the S/P transition, a*=(a
*- δ). Evaluation (3.80) at this
point leads to:
(3.81)
Thus, since k, v and a are finite and positive, provided GcS>GcP, and dGcS/v>0 , which is
clearly the case, d2a/dt
2=∞. The experimental data suggest strongly an inflexion point,
corresponding to d2a/dt
2=0 , rather than a sharp change in da/dt (d
2a/dt
2=∞). This in turn
suggests that δ varies (if only slightly) as the S/P frontier is encroached upon, since this will
permit the denominator of the first term on the right hand side of (3.81), i.e. the term in
in (3.80) to remain finite. Clearly δ cannot decrease, or contact with the
transition will be lost, and so δ must increase. From a purely intuitive, physical, point of view,
this also seems reasonable, since adhesion is less good in the P region, and with a similar
local moment, or value of G, the anticlastic effect is likely to become exacerbated, increasing
δ. A sketch of the expected scenario is given in Fig.3.50. As the crack front encroaches onto
the zone of lower adhesion, separation occurs more readily, leading to a bubble-like failure
area.
Fig.3.50. Suggested sketch of crack front bubbling as it encounters the sharp transition
between SB and P treated aluminium.
The same consideration can be given to the case, occurring later, when the entire crack front
is just on the P region, i.e. when a*=aF
*, but simple substitution into (3.81) reveals that this
situation leaves d2a/dt
2 finite (and possibly 0) under expected conditions. Thus the
paradoxical multi-valued fracture energy curves may be explained. Clearly the model leading
to (3.74) has its limitations but the physical reason for the relatively smooth transition
between types of fracture behaviour occurring on what amounts to a step function in surface
treatment can be successfully explained and is fulfilled by the experimental observations.
S P
Michał K. Budzik
115
3.3.5. Effects of The Adhesive Compliance
The aim of this study was to consider effects of elastic foundation [135] and root rotation
[167,193,200] by using an extension of the strain gauge method previously presented. It must
be appreciated that the following is aimed at both a better appreciation of the intricacies of the
CDT test, and an analysis of the interpretation of results specifically in the context of the
strain gauge method adopted. An asymmetric wedge configuration was adopted in which both
adherends were from aluminium alloy. The flexible adherend was (initially) bonded to the
rigid member along length, ladh=60 mm. This left an initial effective crack length, a, as shown
(see Fig.3.50). Either 8 or 10 strain gauges were attached, along the central line of symmetry,
onto the outer surface of the flexible plate, with at least three being in front of the crack, or in
the so-called free zone, the remainder being in the (initially) bonded zone. Two adhesives
were used. One was a epoxy resin Bostik Araldite Cristal second acrylic mastic. Adhesives
thicknesses, e were maintained in the range of 0.2-0.8 mm. Prior to bonding, all substrate
surfaces were prepared by polishing and degreasing with acetone. In the case of bondline
thickness, e=0.2 mm, surfaces were subsequently sandblasted, with Salox Al2O3. For bondline
thickness, e=0.8 mm, and for the mastic joints, after polishing surfaces were treated by
phosphoric acid anodisation (PAA). Experiments were undertaken either at ambient
temperature, ca. 23°C, or elevated temperature, 50°C. Both conditions were achieved using
Memmert D 06061 Model 500 oven cabin (Memmert GmbH + Co. KG D-91107 Schwabach,
Germany).
Data reduction method
General deflection equation (simple beam theory - SBT) is 2-dimensional and does not allow
for any transverse effects (y in Fig.3.51). Also, due to use of simple beam theory, no
allowance is made for flexibility of the system near the crack front but in the bonded part (x >
a), particularly within the adhesive layer. This may give rise to some vertical displacement of
the flexible beam within the bonded zone, and thus lead to root rotation as was shown,
probably, in CFT test and was presented by other authors [173,208-210]. In Fig.3.52, a sketch
of a flexible beam, rigidly clamped (the brick wall) at x=a SBT (SBT-simple beam theory), but
bonded by a sufficiently soft adhesive as far as aW (W-Winkler) is presented.
FRACTURE IN ASYMMETRIC BONDED JOINTS
116
Fig.3.51. Schematic representation of samples tested.
With the insertion of a wedge of thickness, Δ, not only does the beam flex upwards but the
bonded area, aW<x<aSBT, also deforms as shown in Fig.3.52.
Fig.3.52. Schematic presentation of Winkler (aWinkler) and cantilever beam (aSBT) crack
lengths. The difference between these crack lengths gives the transition zone between open
crack and measurable elastic foundation effect.
Assuming the section, aW<x<aSBT, to be elastic and to possess foundation stiffness, k (Nm-2
),
in the z direction, a load of q (Nm-1
) can be represented as:
Δ
R
PROCESS
ZONE
aWinkler
aSBT
z
x
L
Δ
z
x
e h
a ladh
l
b
STRAIN GAUGES
y x
x = a
M
F z
x
k
Michał K. Budzik
117
(3.82)
Moments equilibrium equation (F is only acting shear force):
(3.83)
Force equilibrium equation:
(3.84)
The bending moment is given by:
(3.85)
where θ is the angle of Euler – Bernoulli beam model:
(3.86)
Application of standard equations describing the equilibrium of forces and couples and the
bending moment relation, M, leads to two 4th
order differential equations:
For 0< x<a:
(3.87)
with general solution:
(3.88)
And for : a< x<+∞:
(3.89)
with solution:
(3.90)
Equation 3.90 has only two constants of integration because the possible solution in is
physically unrealistic, and has therefore been eliminated. The constant, or wave number, λ, is
given by:
FRACTURE IN ASYMMETRIC BONDED JOINTS
118
(3.91)
In the present case, k is interpreted as:
(3.92)
where νadh and Eadh are respectively Poisson’s ratio and Young’s modulus of the adhesive, and
e is adhesive thickness. The bracketed terms in νadh are to allow for essentially plane strain
conditions existing in a thin adhesive layer (e/b<<1). The correction is not negligible, since an
adhesive could easily have a Poisson’s ratio of 0.4, leading to ca. twice the value of k given
by the simpler, often used, expression
.
To find out 6 constants of (3.88) and (3.90) we need derivations up to the 3th
order for both,
free and bonded zone:
For 0<x<a:
(3.93)
(3.94)
(3.95)
(3.96)
For a<x<∞:
∞
(3.97)
∞
(3.98)
∞
(3.99)
∞
(3.100)
Michał K. Budzik
119
Required boundary conditions are:
For x=0:
(3.101)
(3.102)
Finally, from continuity conditions when x=a, we have:
(3.103)
(3.104)
(3.105)
(3.106)
Therefore constants are:
(3.107)
(3.108)
(3.109)
(3.109)
(3.110)
(3.111)
By deriving the standard beam equation one further time with respect to x, and using
equilibrium of moments (dM/dx+F=0 with F as the (shear) force exerted by the wedge on the
beam), we have:
FRACTURE IN ASYMMETRIC BONDED JOINTS
120
(3.112)
Thus:
(3.113)
when equation (3.96) has been substituted. Now, elastic energy is:
(3.114)
and using (3.109), we obtain expression:
(3.115)
Which may is used to find out the strain energy release rate, G:
(3.116)
Substituting second moment of inertia, Ih:
(3.117)
With , Ih – inertia moment defined before.
Also useful are expressions for surface strain of the upper surface of the flexible adherend, εs,
which we find from derivation of equations 3.88 and 3.90 and the constants. It is in the main
negative, therefore we use its absolute value. Strain in free zone:
(3.118)
(3.119)
(3.120)
Strain in bonded zone :
Michał K. Budzik
121
∞
(3.121)
∞
(3.122)
For comparison, strain equation of SBT theory is:
(3.123)
Note that, in the limit of (infinite rigidity, k, within the intact joint), equation (3.120)
becomes equation of simple beam theory (3.123) as would be expected.
To compare both models ratio θ, is introduced, representing the ratio of surface strains, for
given values of x and a, expected from the SBT and the Winkler (W) interpretations:
(3.124)
Thus we may see that there is a simple proportionality between the two estimated values of
strain in the free zone, independent of x, yet dependent on crack length a (λ, from equation
3.91 is a constant of the system). Clearly, for high values of λa (rigid adhesive and/or large
crack length), , and little distinction exists between the two models. However, for
smaller values of λa, the difference may be significant. By rearrangement of SBT equation,
we may isolate crack length, aSBT, as estimated from SBT, as a function of (measured) surface
strain, , at a given x (and given Δ and h). From (3.124), we see that the same strain
(measurement) will give rise to a different estimate of crack length, aW, from the Winkler
theory:
(3.125)
We take as the real, or best estimate of crack length and finally, we may also define the
ratio η, representing the ratio of estimates of fracture energy obtained from the two
approaches:
(3.126)
Again, as , the ratio of fracture energy estimates tends towards unity, but for lower
values of λaW, SBT tends to overestimate strain energy release rate, and therefore fracture
energy.
FRACTURE IN ASYMMETRIC BONDED JOINTS
122
Results and interpretation
The experimental set-up described in previous section allows to obtain values of strain of the
flexible adherend on the unbonded side along the centre line, and at distances xi from the
inserted wedge. Due to continuous monitoring, it is possible to follow the evolution of
(surface) strain with time. Although time dependency is not the principal aim of the present
study, an example for a joint made with acrylic mastic is given in Fig.3.53 (left), in
which vs. time, t, is given for 8 different strain gauges (here numbered 1-8).
Fig.3.53. Left: surface strain, |εs(x)| , at various values of x, vs. time, t. Right: |εs(x)| vs. x for a
given time shown by the bold vertical dotted line on the left part of the figure.
Strain measurements are reported over three minutes. What is observed over the first ca. 20
seconds can be construed as transient behaviour during loading. Thereafter, there is some
crack growth, or possibly viscoelastic accommodation of the adhesive, exemplified by
changing strain, the system approaching stability towards three minutes. More significant is
the fact that strain clearly depends on the position of the strain gauge in question. The greatest
strain is at xi=45 mm (strain gauge 3), and the least at xi=91 mm (strain gauge 8), with the
others being intermediate. If we take a given value of time, here ca. 0.025 h we can plot
vs. xi as shown in Fig.3.53 (right). It can be seen how the strain at first increases, more
or less linearly, from xi=0 (where is necessarily 0, as shown by the initial point, not in
fact, a measurement) up to its maximum, and then decreases beyond xi=45 mm. Up to 45 mm,
we have strain in the unbonded (or debonded) area, linearly increasing in agreement with
equation (3.119) (or its simple beam theory counterpart). Somewhere in the range between 45
and 67.5 mm, there will (presumably) be a maximum, which should correspond to the crack
front. Beyond that value of x, surface strain would be expected to become immediately zero,
if simple beam theory is applied. However, it can be seen that strain only slowly decreases
with x, strongly suggesting an effect of elastic foundation.
Having seen semi-quantitatively the effect of elastic foundation, the objective now is to obtain
the best interpretation of the results quantitatively, and thus obtain a good estimate of crack
length. Other advantages of the elastic foundation treatment will also become manifest.
A
A
A-A
0.00 0.01 0.02 0.03 0.04
0.0000
0.0002
0.0004
0.0006
0.0008
0.0010 x
1= 29 x
2= 37 x
3= 45 x
4= 59
x5= 67.5 x
6= 75.5 x
7= 83.5 x
8= 91
Str
ain
, S
(x)
Time, t(h)
0 20 40 60 80 100
0.0000
0.0002
0.0004
0.0006
0.0008
0.0010
Str
ain
,
s(x
)
x (mm)
FREE BONDED
CR
AC
K
Michał K. Budzik
123
Interpretation of strain gauge measurements
The theory presented has been applied quantitatively to results obtained both with the epoxy
resin and with the acrylic mastic and essential findings presented in Fig.3.54-3.58 These
represent static results, i.e. crack growth rate is sufficiently slow to be neglected during
measurement. Various features are present in these figures to a greater or lesser extent. In all
cases, experimental results show that (negative) surface strain initially increases acceptably
linearly from the origin, corresponding to the position of the wedge. This observation is in
agreement both with the SBT model and the Winkler theory and is explained by the linearly
increasing bending moment. Following this, strain decreases with x, in Fig.3.54 very abruptly
becoming effectively zero, suggesting to a good approximation the validity of the SBT.
Fig.3.54. Results of (negative) surface strain, |εs(x)| vs. distance, x, from wedge for the epoxy
adhesive with a 0.2 mm thick bondline (ambient temperature) (right).
Fig.3.55. As for Fig.3.53, but at 50°C (left).
It may be considered that this case, with a rigid adhesive and thin bondline, shows the most
inflexible elastic foundation. On heating the same system to 50°C (Fig.3.55), the gradient of
the initial linear portion decreases, and what is more, the reduction of (negative) strain after a
maximum, is more gradual. Clearly the modulus of the adherend (aluminium alloy) does not
change significantly over this temperature range of less than 30°C, but that of the epoxy
adhesive does. Thus, despite the use of a thin bondline (its thermal expansion is negligible
here), reduction in elastic modulus is clearly sufficient to change observed behaviour.
Reduced absolute surface strain in the unbonded region can be attributed to the root rotation
allowed by the bonded portion, and follows from the Winkler treatment. Figs.3.56 and 3.57
show results for a 0.8 mm bondline. At ambient temperature, some effect of elastic foundation
may be seen, due to the fourfold increase in joint thickness. This is, however, limited. At
50°C, the initial gradient is reduced and the drop in strain after the maximum more gradual, in
keeping with the findings with a 0.2 mm bondline. There is even some (very slight and
therefore possibly insignificant) suggestion that surface strain becomes positive (i.e. negative
strain becomes negative in the figure).
0 10 20 30 40 50 60 70 80 90
0.0000
0.0002
0.0004
0.0006
0.0008
0.0010
0.0012
0.0014
0.0016
0.0018
0.0020
0.0022
Str
ain
, -
s(x)
x (mm)
Experiment
Winkler
SBTcorrected
SBT
= 2.8 mm
aWinkler
= 54.5 mm
aSBT
= 56.5 mm
= 0.55
0 10 20 30 40 50 60 70 80 90 100
0.0000
0.0002
0.0004
0.0006
0.0008
0.0010
0.0012
0.0014
0.0016
0.0018
0.0020
0.0022
Str
ain
, -
s(x)
x (mm)
Experiment
Winkler
SBTcorrected
SBT
e = 0.2 mm
T = 50o C
= 2.8 mm
aWinkler
= 55 mm
aSBT
= 63.5 mm
= 0.11
FRACTURE IN ASYMMETRIC BONDED JOINTS
124
Fig.3.56. Results of (negative) surface strain, -εs(x), vs. distance, x, from wedge for the epoxy
adhesive with a 0.8 mm thick bondline (ambient temperature) (right).
Fig.3.57. As for Fig.3.56, but at 50°C (left).
In Fig.3.58, we see a very marked effect of the elastic foundation. After the maximum of
(negative) strain, there is a gradual decrease over more than 20 mm. Given the comment about
negative strain above for the 0.8 mm thick epoxy resin, it may at first appear surprising that
this is not observed here, but closer inspection of the theory shows that the distance at which
positive surface strain appears is dependent on the wavenumber, λ, which in turn depends on
the adhesive Young’s modulus, Eadh. A smaller value of λ implies a larger wavelength, and
thus positive strain appears further from the crack front, too far to be visible in our case.
0 20 40 60 80 100
0.0000
0.0002
0.0004
0.0006
0.0008
0.0010
0.0012
0.0014
0.0016
0.0018
0.0020
0.0022 Experiment
Winkler
SBTcorrected
SBT
Str
ain
, -
s(x
)
x (mm)
MASTIC
e = 0.4 mm
= 2.6 mm
aWinkler
= 53 mm
aSBT
= 74.3 mm
= 0.0477
Fig.3.58. Results of (negative) surface strain,-εs(x), vs. distance, x, from wedge for the acrylic
mastic with a 0.4 mm thick bondline (ambient temperature).
The results presented in Figs.3.54-3.58 may be analysed in the light of the theory presented
above. Beam, or crack length, aSBT, as obtained from the SBT interpretation implies an
encastré attachment, as if the bonded portion of the joint were embedded in a brick wall
(Fig.3.52). This leads to overestimation of crack length, compared to the Winkler theory
(aW<aSBT). The physical reason is that in reality, the bonded region presents some flexibility,
0 10 20 30 40 50 60 70
0.000
0.001
0.002
0.003
0.004
0.005
0.006
Str
ain
, -
s(x)
x (mm)
Experiment
Winkler
SBTCorrected
SBT
e = 0.8 mm
= 3.2 mm
aWinkler
= 35.8 mm
aSBT
= 40 mm
0 20 40 60 80 100
0.0000
0.0005
0.0010
0.0015
0.0020
0.0025
0.0030
Str
ain
, -
s(x)
x (mm)
Experiment
Winkler
SBTCorrected
SBT
e = 0.8 mm
T = 50oC
= 3.2 mm
aWinkler
= 52.5 mm
aSBT
= 59.3 mm
= 0.1451
Michał K. Budzik
125
allowing root rotation. Attachment of the free beam is thus less restrained (the beam is not
exactly parallel to the substrate at the crack front, as is assumed in SBT) leading to lower
values of (absolute) surface strain in the former. This in turn suggests a lower applied bending
moment (larger R in Fig.3.52) which, in the case of SBT implies a higher value of crack
length, a. Mathematically, we consider (3.124), which gives the ratio of .
The ratio θ was already included in (3.125) in order to estimate the Winkler crack length, aW,
but assuming SBT, we use the simpler equation, viz.:
(3.127)
in which θ has been neglected. (Note that (3.125) and (3.129) are independent of x since εs(x)
is linear in x). From (3.125) and (3.129), we may calculate , which
quantifies the overestimation of crack length from SBT theory. As may be seen in Fig.3.59,
for large values of the product λaW, the difference is relatively slight, the ratio being close to
unity, but for small values (i.e. short crack for a given value of λ) it can be very significant
(aW has been taken as the true crack length). In any dynamic experiment, this means that the
overestimation of crack length will actually decrease, since crack length itself is increasing.
Fig.3.59. Crack length overestimation due to assumption of SBT
From the experimental results, it is possible to use regression analysis to obtain the best fits of
the data to both SBT and Winkler interpretations. For the former, only the initial linear data
are used. For the latter, all data are used and therefore the requisite value of λaW, for a given
system (crack length included) can be found. This was done using a Matlab® programme.
Returning to Figs.3.54-3.58, three calculated lines were added, viz. SBT, Winkler, and
SBTcorrected. The first (open circles) is simply application of linear regression to the
experimental points assumed to correspond to the separated part of the wedge test. The second
(continuous heavy line) corresponds to application of the Winkler theory. This, of course, is
0 20 40 60 80
1.0
1.2
1.4
1.6
1.8
2.0
1
/3
aW
FRACTURE IN ASYMMETRIC BONDED JOINTS
126
valid over the entire joint, unbonded and bonded portions. It can be seen that obverse surface
strain is expected to become momentarily positive, after the decrease from its maximum
(absolute) value corresponding to the crack front (except in the case of the soft acrylic mastic,
Fig.3.58, where the dip is off scale). This effect cannot be truly confirmed from our results,
but a strong suggestion is nevertheless present in Figs.3.56 and 3.57. Within the limits of
experimental error, the trend of surface strain behaviour upstream of the crack front correlates
well with theory. This is particularly the case for the results obtained with the acrylic mastic.
As can be seen from the data inset in Figs.3.54-3.58, the SBT and Winkler estimates of crack
length are different, as expected, with aW<aSBT. The more supple is the elastic foundation, the
greater is the difference. The third line, SBTcorrected, corresponds to a simple calculation of the
expected strain/position relationship, using SBT but assuming the Winkler estimate of crack
length to be valid. As can be seen, this treatment overestimates strain, only slightly for a rigid
elastic foundation, but quite considerably so for a flexible foundation.
In Fig.3.60, a microscopic side view of an acrylic mastic bond is presented, both before and
after insertion of the wedge. It can be seen that the initial thickness of the adhesive of 0.4 mm
increases after wedge insertion.
Fig.3.60. Zone near crack front, on bonded side, in acrylic mastic bond, before (upper) and
after (lower) wedge insertion. Extension of the adhesive layer perpendicular to the interface is
due to imposed displacement and effect of elastic foundation.
The wedge being on the left hand side, it can be seen that the extension of the mastic, without
fracture, increases from right to left.
Effect on strain energy release rate
Clearly, the aim of the wedge test is to estimate the fracture energy, principally in mode I, of
adhesive bonds, Gc, or equivalently the critical strain energy release rate for given conditions.
In this study, actual fracture has been avoided in the main since the aim was to evaluate
Δ
Δ
Michał K. Budzik
127
different interpretations of the basic test. As such, values of fracture energy for the systems
studied are not presented here, but rather compare estimates of strain energy release rates
corresponding to fracture energies if the crack were to propagate. In order to compare
equivalents, we first consider the five systems corresponding to Figs.3.54-3.58 with their
assumed correct cracks length, viz. those calculated using Winkler theory. Values of the ratio
of energy release rate , as given by equation 3.126, are presented in Fig.3.61.
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Epoxy, e = 0.2 mm
Epoxy, e = 0.2 mm, 50oC
Epoxy, e = 0.8 mm
Epoxy, e = 0.8 mm, 50oC
Mastic, e = 0.4 mm
Fig.3.61. Ratio, (equation (18)) of SBT to Winkler values of the strain energy
release rate using Winkler values of the crack length, aW.
It is clear that in all cases, use of SBT leads to an overestimate of strain energy release rate,
and that this effect is seriously increased by the supple nature of the adhesive layer. However,
this is perhaps an unfair comparison. The SBT approach already presents an error, which may
be significant if the adhesive layer is soft. It produces an estimate of crack length, aSBT, which
is too large. Thus we are justified in comparing the strain energy release rates of the two
approaches using calculations based on respective estimates of crack length. (3.126) becomes:
(3.128)
(3.129)
where refers to the fact that each calculation of G uses its corresponding value of the crack
length, a. Equation 3.129 make use of introduced before factor θ, in order to avoid using
different crack lengths in the same formula. Values of are presented in Fig.3.62, and show
convincingly that the two errors in SBT are effectively self-compensating, at least in the cases
studied. These results show that there are still things to be learnt from the wedge test in the
vicinity of the crack front. This approach should prove promising to look at such phenomena
as crazing and cavitation in the adhesive layer.
FRACTURE IN ASYMMETRIC BONDED JOINTS
128
0.0
0.2
0.4
0.6
0.8
1.0
Epoxy, e = 0.2 mm
Epoxy, e = 0.2 mm, 50oC
Epoxy, e = 0.8 mm
Epoxy, e = 0.8 mm, 50oC
Mastic
~
Fig.3.62. Ratio, (equation (20)) of SBT to Winkler values of the strain energy
release rate, each using own values of crack length, a.
Effects of deformation of the adhesive layer perpendicular to the interface, using an approach
first suggested by Winkler in the 19th
century where here analyzed. The Winkler model (W)
was used in conjunction with a strain gauge method, previously reported, allowing detailed
information of strain within the flexible member of an asymmetric wedge test to be accurately
determined. A comparison between the two approaches, W and SBT was made. An important
finding was that when calculating crack length from the SBT treatment, using only data from
the free part of the adhesive bond, this approach led systematically to overestimations. If
corrected crack lengths are used, then strain energy release rate is overestimated. However,
when using SBT with its own erroneous estimate of crack length, the errors incurred would
appear to be self-compensating, at least in the cases studied! It should be noted that use of
SBT may indicate a crack length beyond the actual value, where the adhesive is significantly
strained but not broken. The effect may be compared to crazing or cavitation. Findings do not
invalidate use of SBT analysis, which is reassuring, but the present treatment should lead
potentially to more insight into fracture, in the deformed, but still intact adhesive zone.
3.3.7. Use of the CDT test for assessment of curing time
The theory discussed in the previous section was used to study the behaviour of the structure
bonded with the long time curing adhesive. Basically, as an adhesive cures, it is expected to
get harden and thus its modulus increases. Change of parameter λ, or simply change of the
adhesive mechanical properties, with time was expected, thus giving λ=f(t) or more strictly,
Eadh=f(t). The flexible adherend used was CFRP composite with Eh=75 GPa. The rigid or
thick plate was Hydronalium AA5754. Prior to bonding the aluminium surface was polished
and then anodized using the described PAA process. The composite plate was abraded, using
400 grit paper, then wiped in acetone and carefully dried in a stream of warm air (max. 50oC).
Plates were bonded using Araldite GY784BD epoxy adhesive. The measured adhesive
thickness was e=0.31 mm and found constant along the bonded zone. Crosslinking was
effected at ambient temperature (ca. 23°C) for 24 hours under a low constant pressure of ca. 4
Michał K. Budzik
129
bar (to avoid air trapping and voids) and at ca. 55 % RH. The crack length, which was kept
constant during the test, was a=65 mm while the plates were bonded along ladh=45 mm. Six
strain gauges were attached to the exposed side of thin substrate. Three in the so-called free
zone, and three in the bonded zone. To avoid any crack propagation a thin aluminium wedge
of thickness Δ=1.6 mm was used. The wedge was inserted, between separated plates, to the
desired position (2 mm from the CFRP plate extremity) and then, after ca. 5 seconds
removed, preventing any crack propagation. This operation was repeated for two samples
every day for 15 consecutive days.
Fig.3.63. Schematic representation of tested samples. Bonded and free zones are separated
with dash line.
Strain behaviour
The test results of the tests performed are shown in Figs.3.64-3.66 in terms of the flexible
plate strain change. Visible change, increase in free, decrease in bonded zone, of the strain is
related to the change of the adhesive state. Although this change can come from many
different phenomena, we assume that during crosslinking the adhesive changes its mechanical
properties. Fig.3.65 correspond to the measured strain distribution (circles) along the joint, at
the beginning and end of the test. Using Matlab® Fitting tool, the Winkler model strain was
added (dashed and solid line). Experimental and Winkler model curves agree well, thus
proving that the method presented can be successfully used to study effects of the adhesive
crosslinking.
BONDED
ZONE
FREE
ZONE
STRAIN GAUGES
x
y
b
32.5
42.5
52.5
69.5
75.5
82.5
FRACTURE IN ASYMMETRIC BONDED JOINTS
130
Fig.3.64. Instrumented wedge test results for the adhesive subjected to slow crosslinking.
Vertical line indicates position of the crack front. Note evolution of the strain with time
(arrows indicating the direction of change) in both zones.
Fig.3.65. Interpretation of experimental data (points) with Winkler based model at the
beginning (solid line) and end of the test (dashed line). Vertical line represents crack position.
Fig.3.66 considers the situation as described previously, this time adding SBT and Winkler
model estimations of the crack length. It must be appreciated that the change of the adhesive
state, due to reticulation, when the joint is treated as an encastré (SBT) is manifested by the
decrease of the crack length. In Winkler, change of the adhesive state is seen via a change of
the λ parameter – with time increase, adhesive stiffness does the same.
Process zone after
2 days
0 10 20 30 40 50 60 70 80 90 100 110
0.00000
0.00005
0.00010
0.00015
0.00020
0.00025
0.00030
0.00035
0.00040
0.00045
Str
ain
, -
S(x
)
x (mm)
t= 2 days
Winkler, t = 2 days
t = 15 days
Winkler, t = 15 days
FREE ZONE
BONDED
ZONE
Process zone
after 11 days
0 10 20 30 40 50 60 70 80
0.00000
0.00005
0.00010
0.00015
0.00020
0.00025
0.00030
0.00035
Str
ain
, -
S(x
)
x (mm)
t = 12 days
t = 8 days
t = 6 days
t = 2 days
BONDED
ZONE
FREE
ZONE
Michał K. Budzik
131
Fig.3.66. Comparison between Winkler and SBT model. Vertical lines represent: real or
Winkler crack length (solid), crack length obtained with SBT estimation after 15 days of
curing (dashed), and SBT after 2 days of curing (dotted).
Curing time estimation
From Figs.3.67-3.69, the time required to cure the adhesive can be estimated. It should be
noted that all of the parameters presented here can be used to evaluate curing time. Strain
gauges based in the bonded zone provide estimation of the exact size of the process zone as
well as the value of the characteristic adhesive parameter, λ. This is even more valuable, since
the CDT employed was successfully performed in non-destructive mode. Returning to
Figs.3.67-3.69, the curing time using Winkler and SBT approach is manifested in different
forms as mentioned previously. Using the Winkler model, we observe changes of the λ
parameter which correspond to the adhesive state or more clearly to the size of the process
zone (see Fig.3.67).
Fig.3.67. Change in the adhesive parameter, λ, during crosslinking. Three stages of
crosslinking can be found (left).
Fig.3.68. Apparent change of the crack length estimated using simple beam theory, aSBT, as a
function of time (right).
2 4 6 8 10 12 14
65
67
69
71
73
75
77
Cra
ck
len
gth
, a
SB
T(m
m)
Time, t (days)
2 4 6 8 10 12 14
0.1
0.2
0.3
0.4
0.5
0.6
m
m-1
)
Time, t (days)
dλ/dt↑ dλ/dt=0
STAGE I STAGE II STAGE III
a = 65 mm
aSBT = 67 mm
aSBT = 76.5 mm
0 10 20 30 40 50 60 70 80 90 100 110
0.00000
0.00005
0.00010
0.00015
0.00020
0.00025
0.00030
0.00035
0.00040
0.00045 t = 2 days
SBT, t = 2 days
Winkler, t = 2 days
t = 15 days
SBT, t = 15 days
Winkler, t = 15 days
Str
ain
, -
S(x
)
x (mm)
FRACTURE IN ASYMMETRIC BONDED JOINTS
132
The change of the wave number is however not linear. This, in fact, stays in good agreement
with some recent theory [e.g. 222]. As was shown, polymer crystallization gradient varies
with time. It is small in the initial stage (stage I), than accelerates (stage II) to a certain
crystallization speed (known as critical crystallization speed, vc). After achieving vc,
crystallization gradient decrease, and after short time reaches 0 (stage III). In Fig.3.67 similar
situation can be observed, related to the change of λ. However the marked acceleration, in
stage II, is not understood.
Fig.3.69. Variation of Energy Release Rate, G, with curing time. Vertical line represents the
end of the crosslinking period.
Fig.3.70. Schematic presentation of change of crack length during crosslinking. Not that the
real crack length remains constant, whereas estimated crack length changes due to neglect of
adhesive properties in SBT.
Δ
PROCESS
ZONE
Δ
z
x
aWinkle
r aSBT↓
R↓
|εs|↑ PROCESS
ZONE
t =
0
t = ∞
aSB
T
aWinkler
R
z
x
|ΔaSBT|>0
2 4 6 8 10 12 14
1.8
2.0
2.2
2.4
2.6
2.8
3.0
3.2
En
erg
y R
ele
ase
Rate
, G
I (J
m-2
)
Time, t (days)
Michał K. Budzik
133
In Fig.3.68 SBT crack length variation with time is presented. In the SBT treatment, the
change of the adhesive state (crosslinking), from soft to rigid is manifested by the apparent
decrease in crack length (or more correctly SBT model crack length overestimation). This
effect was expected and stays in agreement with previous statements when using the Winkler
model. The observed change in apparent crack length is linearly dependent on time. Also
energy release rate evolution, shown in Fig.3.69, was found approximately linear. The energy
release rate for both SBT and Winkler models was found to be the same, when crack lengths
applied corresponded to the given model estimation. Therefore, the curing time can be
estimated from both models (SBT or Winkler), although intuitively the Winkler model is
closer to the truth in this situation (Fig.3.70). The change of adhesive properties was observed
during the adhesive crosslinking using the strain gauge technique and interpretation with the
Winkler model. The length of the process zone decrease with time from ca. 60 to 5 mm (see
also Fig.3.65) is explained by the change of the adhesive properties with time. It must be
emphasized that the actual crack length remained unaltered throughout, since the conditions
of wedge insertion were carefully controlled. 11 days curing time can be finally recommended
as a curing cycle for a given adhesive system.
3.3.8. Temperature effects on fracture using the CDT test
One of the aims of the thesis was development of the fracture test, which can be successfully
applied for in situ, continuous measurements in complex mechanical-environmental loading
conditions. Therefore, the CDT test developed was performed to study fracture at ambient
(ca. 24oC) and elevated temperature (40
oC). In this test, two aluminium plates were used.
Prior to bonding, aluminium surfaces were polished and sandblasted with Salox Al2O3 grit.
The adhesive used was commercial epoxy resin Araldite Cristal. Plates, of now standard
lengths, thicknesses and widths, were bonded along ladh=70 mm. Crosslinking was effected at
ambient temperature (ca. 23°C) for 24 hours under 0.3 bar pressure and at ca. 55 % RH, and
dried for 24 hours at 50oC in ventilated Memmert D 06061 Model 500 oven cabin (Memmert
GmbH, Schwabach, Germany). Thickness of the adhesive layer was e=0.2 mm. Initial crack
length, was a=35 mm. Such prepared samples were left for 5 days in ambient conditions to
avoid any secondary effects of curing. Nine strain gauges were attached to the outer surface,
as shown in Fig.3.71. One additional strain gauge was left mechanically unloaded, thus
serving as a dummy gauge (to register temperature effect only). Before the test, samples were
precracked with a 3 mm thick wedge. Wedges of thickness, Δ=4.8 and 6.8 mm, were then
inserted to separate the two bonded plates. In order to observe fracture at elevated
temperature, the CDT setup and Dino Microcamera (used to register in situ fracture features
in different conditions) were placed in the Memmert oven cabin. Four tests were carried out.
Each of the tests started in ambient or room conditions (T=24oC, RH=60%). The crack was
then left to propagate until its speed reached ca. 0.01 mmh-1
. Subsequently the temperature
was increased to T=40oC with the heating rate of 0.5
oC/min. The sample was left in this
conditions to the moment of crack propagation arrest.
FRACTURE IN ASYMMETRIC BONDED JOINTS
134
Fig.3.71. Schematic representation of tested sample.
Microscopic features of fracture in different conditions
Figs.3.72-3.74 are visual assessments of the crack propagation and fracture in both tested
conditions. The first remark must be made about multiple crack fronts. In ambient conditions,
the crack propagates via bridges created between the substrates which are particularly visible
in the bottom left of Fig.3.72. In ambient conditions these bridge zones, as observed, occur
and thus could be attributed to the process zone. Propagation observed in elevated
temperature conditions takes place in a different manner. The crack propagates through the
created fibrils of polymer (see Fig.3.73), which span the space between aluminium plate and
the adhesive layer, resulting in craze-like zones [215,216] (Figs.3.72 and 3.73). These zones,
as was observed, can expand to about 10 mm in length. This particular, elevated temperature
propagation, feature was found only in globally defined process zones. This means that the
strain recorded within the craze zone was higher than after in respect to the wedge position. In
addition, this zone was found to possess the ability to post cure and recover elastic properties
after a decrease of temperature, providing zones of certain strength. This is shown in Fig.3.74.
It was observed, that while removing (from between separated plates) the wedge at elevated
temperature the plates returns to parallel, initial, positions in respect to rigid adherend. The
situation looked different when the wedge was left between separated plates and the entire
system was cooled down to ambient temperature. Thus, removing the wedge did not change
the flexible member position (Fig.3.74). This strongly suggest that some modification to the
adhesive occurred.
b
BONDED
ZONE
FREE
ZONE
STRAIN GAUGES
86.5
95
79
23
x
y
31
39
46.5
55
64
71
Michał K. Budzik
135
Fig.3.72. Crack propagation sequences recorded in room temperature condition, T=24
oC (left)
and at the elevated temperature T=40oC (right).
Fig.3.73. Feature of fracture in elevated temperature – the adhesive crazes in the bonded zone
(indicated with arrows).
Fig.3.74. Sample cooled down with wedge between separated plates. Wedge was removed
after cooling down.
Results of strain measurements
In the following, new and intriguing results have been obtained. Discussion presented here is
not definitive since several points are still unclear and require further investigative work.
ALUMINIUM
ALUMINIUM
ADHESIVE
100 μm
ALUMINIUM
ADHESIVE
ALUMINIUM
ALUMINIUM
ADHESIVE
200 μm
T = 24oC
TIM
E
T = 40oC CRACK DIRECTION
FRACTURE IN ASYMMETRIC BONDED JOINTS
136
Figs.3.75-3.78 refer to the results of the direct strain gauge measurements. As can be noted (as
also stated in previous studies) process zone length increase with increase of temperature.
Also the adhesive elastic properties (we assume that change of temperature does not affect the
aluminium properties) change with temperature. Of particular interest here is the possibility to
obtain results and crack position while two parameters that are changing with time, λ and
crack length a.
Fig.3.75. Strain profiles of the sample tested in ambient temperature at the beginning and end
of the test. The crack propagation from solid to dotted line (left).
Fig.3.76. Strain profiles of the sample tested in elevated temperature at three stages:
beginning, middle and at the end of the test. Solid. dashed and dotted line corresponds to the
crack position at the given stage respectively (right).
Fig.3.77. Strain profiles of the samples tested in ambient temperature at the beginning and end
of the test for the 6.8 mm wedge. The crack propagation from solid to dotted line (left).
Fig.3.78. Strain profile of the sample tested in elevated temperature at three stages: beginning,
middle and at the end of the test. Solid, dashed and dotted line corresponds to the crack
position at the given stage respectively (right).
Graphical, precise, crack length estimation
Experimentally obtained strain evolution, and crack propagation in time (illustrated in
Fig.3.79a) are the most important findings obtained in this study. When the crack is passing a
strain gauge position, the maximum value of strain response (since the corresponding bending
moment has his maximum at this moment of time) is expected (here it is for 95 mm long
crack and strain gauges bonded at this position, with the respect to the wedge).
0 10 20 30 40 50 60 70 80 90 100 110
0.0000
0.0005
0.0010
0.0015
0.0020
= 6.8 mm
T = 40oC
t = 0
t = 24 h
t = 170 h
Su
rfac
e st
rain
, -
s
x (mm)
0 10 20 30 40 50 60 70 80 90
0.0000
0.0005
0.0010
0.0015
0.0020
t = 0
t = 24 h
Su
rface s
train
, -
s
x (mm)
= 6.8 mm
T = 24oC
0 10 20 30 40 50 60 70 80 90
0.0000
0.0005
0.0010
0.0015
0.0020
t = 0 h
t = 48 h
t = 100h
Su
rfac
e st
rain
, -
s
x (mm)
= 4.8 mm
T = 40oC
0 10 20 30 40 50 60 70 80 90
0.0000
0.0005
0.0010
0.0015
0.0020
t = 0 h
t = 22 h
Su
rfac
e st
rain
, -
s
x (mm)
= 4. 8 mm
T = 24oC
Michał K. Budzik
137
Fig.3.79. a) Strain evolution with progressing crack. Signals from three strain gauges located
at: x1=23 mm and x6=64 mm (already passed by the crack) and x10=95 mm (passed during the
test). Last gauge signal presents evolution of signal in neighbourhood of the gauge. Estimated
process zone, or elastic foundation zone of ca. 14 mm. b) Fracture surface after the test with
visible process zone (left), close-up of the process zone (right).
Knowing the exact position of all strain gauges used, finding the maximum value of the
recorded strain (for the gauge at defined xi) we can calibrate measurements to give the exact
position of the crack front at any time, t, for any adhesive and his state, thus, crack length,
a=xi at this moment. This was successfully achieved and an example is shown in Fig.3.79a.
Moreover, using such data, we can precisely estimate the size (length) of the process zone,
which in the present study was ca. 14 mm length. The direct observation of the surface after
fracture supports experimental results (Fig.3.79b). Left of a Fig.3.79b is close up of the
deformed part of the adhesive, in fact, showing some details of ductile fracture. The crack
lengths recorded during the study are summarized in Fig.3.80. It must be pointed out that for
the same sample, crack length decreased when the test was stopped at elevated temperature,
and started once more at ambient (after cooling down). This is possibly due to post cure and
-εs(x=95 mm)=max
-εs(x=95 mm)=min
ELASTIC FOUNDATION ZONE
a)
b)
ca.14 mm
77.5 80.0 82.5 85.0 87.5 90.0 92.5 95.0 97.5 100.0 102.5 105.0
0.0000
0.0002
0.0004
0.0006
0.0008
0.0010
0.0012
0.0014
0.0016
0.0018
0.0020
x1 = 23 mm
x6 = 64 mm
x10
= 95 mm
Su
rface s
train
, -
s(xi)
Crack length, a (mm)
= 6.8 mm
T = 40oC
FRACTURE IN ASYMMETRIC BONDED JOINTS
138
recovery of certain zones within the aluminium/the adhesive interface (also, the typical
fracture energy value should decrease with increasing temperature).
Fig.3.80. Crack length evolution during the test (note the joint behaviour in Fig.3.74).
Kinetics of crack propagation
Crack kinetics measurements are shown in Figs.3.81-3.83 for the situation when the wedge of
thickness, Δ=6.8 was used (representative for all tests made). What can be easily noted is
stable, slow crack growth at ambient temperature and fast, unstable crack growth at elevated
temperature. The structure, bonded with the tested adhesive, submitted only to 16oC of
temperature, will suddenly fail, although while testing without mechanical loading, will resist,
probably, for hundreds if not thousand of hours.
Fig.3.81. Crack propagation in time for different test conditions (left).
Fig.3.82. Crack propagation kinetic in ambient test condition (right).
Comparing crack speed characteristics (Figs.3.82 and 3.83), differences between propagation
in different conditions are even more pronounced. Stable crack slowing down may be
observed at ambient temperature, while five specific crack speed regions (marked in Fig.3.83)
were found for the elevated temperature propagation. Primarily (stage I), the crack slow down
1E-3 0.01 0.1 1 10 100
0
5
10
15
20
25
30
35
= 6.8 mm, T = 24oC
= 6.8 mm, T = 40oC
Cra
ck
in
cre
men
t,
a (
mm
)
Time, t (h)
0.01 0.1 1 10
0.01
0.1
1
10
100
1000
= 6.8 mm
Cra
ck
sp
eed
, d
a/d
t (m
m h
-1)
Time, t (h)
T = 24oC
60
65
70
75
80
85
90
95
100
105
110
= 6.8 mm
T = 40oC
= 6.8 mm
T = 24oC
= 4.8 mm
T = 40oC
Cra
ck l
eng
th,
a (m
m)
t = 0
t = end
= 4.8 mm
T = 24oC
ADHESIVE RECOVERY
Michał K. Budzik
139
as would be expected in any condition. Second stage (II) shows some crack length
acceleration, possibly due to additional, temperature driven, stress relief. In third stage (III),
the crack slows down once more after stress relief. Fourth stage (IV) can be treated as a
transition stage, between normal, elevated temperature propagation speed, and acceleration
due to approaching sample extremity (stage V) with some signs of stick-slip propagation. In
fact, acceleration in stage V can be provided using entire Winkler model, in which, deflection
is written as a symmetric equation with stress growth at both elastic foundation extremities.
Fig.3.83. Crack propagation kinetics in elevated temperature. Five specific regions were
noted: I. Linear speed decrease, normal for propagation at constant temperature, II. Increase
of the crack speed, possibly due to the adhesive stress relief, III. Normal, stable crack
propagation. IV. Transition, stick-slip like zone, V. Acceleration of the crack close to the
sample extremity.
Fracture energy comparison, Gc
The previous findings were also emphasized in fracture energy analysis. Figs.3.84 and 3.85
show energy release rate as functions of time and crack speed respectively. Although Fig.3.84
does suggest normal – stable energy release Fig.3.85 is more unusual. Here, second order
changes are more pronounced and once more, five specific zones of crack propagation in
40 60 80 100 120 140 160
0.01
0.1
T = 40oC
Cra
ck s
pee
d,
v (
mm
h-1
)
Time, t (h)
= 6.8 mm
I II III IV V
1 10 100
0.01
0.1
1
T = 40oC
Cra
ck s
pee
d,
v (
mm
h-1
)
Time, t (h)
= 6.8 mm
FRACTURE IN ASYMMETRIC BONDED JOINTS
140
elevated temperature can be found. Accordingly, and contrary to the accepted model, some
part of the graph shows decrease of fracture energy with increase of crack speed. This may
show, that some of the environmental conditions, providing crack acceleration, are more
important than simple, mechanical considerations.
Fig.3.84. Energy Release Rate change with time (left).
Fig.3.85. Change of the energy release rate with the crack speed. Stages I-V like in Fig.3.83
(right).
In Fig.3.86, three dimensional visualisation of the entire crack propagation process is
presented. Such data are presented for the first time within this thesis, and are mainly due to
use of the CDT technique developed here.
Fig.3.86. Energy- Crack- Speed curve for elevated temperature propagation. Zone IV of ca.
12.5 mm length.
The complex problem described (but not solved here), requires separate, more complex and
detailed studies. However, provisionally, in agreement with theory (time-temperature
superposition) [223], properties of the polymers e.g. elevated temperature fracture energy can
7580
85
90
95
100
105
30
40
50
60
70
80
90
100
110
0.1
1
Ene
rgy
Rel
ease
Rat
e, G
c (J
m-2
)
Crack speed, v (mm h-1 )
Crack length, a (m
m)
T = 40oC
= 6.8 mm
0.01 0.1 1 10 100
20
40
60
80
100
120
140
160 = 6.8 mm, T = 24
oC
= 6.8 mm, T = 40oC
En
erg
y R
ele
ase
Rate
, G
c (
J m
-2)
Crack speed, v (mm h-1)
I
II III IV
V
0 20 40 60 80 100 120 140 160 180 200
20
40
60
80
100
120
140
160
= 6.8 mm, T = 24oC
= 6.8 mm, T = 40oC
En
erg
y R
ele
ase
Rate
, G
c (
J m
-2)
Time, t (h)
Michał K. Budzik
141
be shifted from e.g. ambient temperature fracture. Herein, for a given environment the
following formula can be written:
(3.130)
where:
v – crack speed
α , β – system constants
γ – environment dependent energy release rate.
Using part of Fig.3.85 (ambient condition curve and stage III of elevated temperature
propagation – stable crack growth at elevated temperature) and following equation 3.130, the
shifted fracture energy vs. crack speed can be plotted (for elevated temperature) as shown in
Fig.3.87. Such based data shows, that in the case studied, the energy required to propagate the
crack at elevated temperature is ca. 3.6 times smaller than for the same adhesive system but
tested in ambient conditions.
0.01 0.1 1 10 100
40
50
60
70
80
90
100
110
120
130
140
150
Gc = 29.88 v
0.1146+ 25.2
Gc = 29.88 v
0.1146+ 90.15
T = 24oC
T = 40oC
En
erg
y R
ele
ase
Rate
, G
c (J
m-2
)
Crack speed, v (mm h-1)
= 6.8 mm
Fig.3.87. Interpretation of temperature effect with highlighted region used for elevated
temperature interpretation.
Testing of adhesive joints in hostile environments has only rarely been reported. Generally,
environment impact was limited to joint behaviour analysis after aging. This can easily lead a
to mismatch with the real problem of the joint reliability. Although after degradation, decrease
of material properties can be observed [224-226], the critical changes are achieved after many
hours of extreme exposure. When structures are loaded both, environmentally and
mechanically, the majority will not survive until degradation moment, e.g. few degrees
heating can critically change the system behaviour. Accordingly, change of the joint operating
temperature can provide many changes in mechanical properties of the system. Complex,
simultaneous mechanical-environmental tests are therefore required. However complex
loading conditions imply many phenomena. Despite complexity it was proven that the
technique developed here, taking advantage of standard strain gauges, is reliable and can be
used for such studies. In addition, the wedge test is particularly suitable for the assessment of
structural adhesive fracture energies. However, this test is often analysed using simple beam
FRACTURE IN ASYMMETRIC BONDED JOINTS
142
theory (SBT), whereby it is assumed that one or both adherends may be treated as encastré
cantilever beams, in their unbonded portions. However, when the adhesive layer is
particularly deformable either due to its low elastic modulus or significant thickness or both,
the encastré part may be far from rigid, leading to an effect of root rotation, or elastic
foundation. This must be taken into account and is part of developed here CDT technique.
Michał K. Budzik
143
s
t n
2
1
Chapter 4. MODELLING OF STRESSES IN ASYMMETRIC ADHESIVE
JOINTS
Finite element analysis approach was introduced to study asymmetric test configuration and
its effects. Different geometrical aspects of the flexible member and the adhesive were
analyzed. Moreover, it must be pointed out, that all elements of the modelled bonded joint
undergoes classical, proportional Hook's law.
4.1. FEM model
Bonded adhesive joint was designed using Cast3m (Development Team, Laboratoire de
Mécanique Systèmes et Simulation, Commissariat Français à l'Energie Atomique) program in
3-dimensional Cartesian geometry with a high level interpreted language: GIBIANE.
Finite Elements (FE)
CUB8 elements were used for the adhesive layer design. They main feature are eight corner
nodes with eight Gauss points. CUB8 corresponds to solid three dimensional elements and are
perfectly suited for elastic analysis. The plate was represented by the COQ4 elements which
are four nodes, four Gauss points elements devoted to the Mindlin-Reissner plates. COQ4
gives advantage of taking transverse shear into account (nb. analogically to Timoshenko’s
beam element). It must be appreciated that both introduced elements takes advantage from
square shape function. This is important aspect for estimation of searched values between the
element nodes. Some of the FE features are summarized in Table 4.1.
Table 4.1. Details of finite elements used in the studies.
Element Presentation Stress or Force Deformation
COQ 4
4 nodes, 3 D element
Local orientation:
N11- normal force in 11
N22 - normal force in 22
N12 - normal force in 12
M11- bending moment in 11
M22- bending moment in 22
M12 - bending moment in 12
V1 - shear force in 1
V2 - shear force in 2
Local orientation:
εss – plane elongation in ss
εtt – plane elongation in tt
γst – shearing in st
γtn – shearing in tn
γsn – shearing in sn
ρss – curvature in ss
ρtt – curvature tt
ρst – curvature st
CUB 8
8 nodes, 3 D element
Global orientation:
ζxx - stress in x
ζyy - stress in y
ζzz - stress in z
ηyz - stress in yz (shear)
ηxz - stress in xz (shear
ηxy- stress in xy (shear
Global orientation:
εxx – plane elongation in x
εyy – plane elongation in y
εzz – plane elongation in z
γyz – shearing in yz
γxz – shearing in xz
γxy – shearing in xy
z
x
y
FRACTURE IN ASYMMETRIC BONDED JOINTS
144
To cast the results within the global coordinate system x, y, z stresses and deformations of
shell element has to be changed. That was made using following the formulas shown below.
(4.1)
(4.2)
(4.3)
(4.4)
(4.5)
(4.6)
(4.7)
(4.8)
(4.9)
(4.10)
(4.11)
(4.12)
where:
h – thickness of modelled shell element
k = 1 for the upper plate surface
= 0 for the middle plane
= -1 for the bottom plate surface.
Mesh
The mesh was built from 50x50 (length x width) finite elements (FE) with increasing length
of the element and constant element width equal to the adhesive width divided by number of
elements, thus sb=b/50. Length of the element was increasing, starting from the crack front
(for plate in both directions), with the first element being of length s1=100 μm (see Fig.4.1).
Michał K. Budzik
145
Fig.4.1. Details of designed mesh.
Boundary conditions
At x=0 constant displacement, Δ was imposed (see Fig.4.1). Thus:
(4.13)
The bottom plane of the adhesive was modelled as built in (brick in Fig 4.1) so thus rigid
adherend was not modelled in the studies. All movements: displacements or rotations are
blocked.
(4.14)
(4.15)
(4.16)
where u with appropriate index is beam displacement in the given direction.
The plate-adhesive interface was modelled using COLLER procedure (operator), which
defines shell (or plate) - volume elements junction. This procedure allows the rotation from
shell elements (e.g. COQ4) to the volume element (e.g. CUB8) nodes. Function allows to
bond two surfaces leaving possibility for the edge rotations and displacements without
sn=sn-1+(Σsn-1)/ladh
h e
ladh (50 FE) a (50 FE)
b (
50 F
E)
z
x
x
y
zadh
s1
FRACTURE IN ASYMMETRIC BONDED JOINTS
146
separation between bonded elements. This fact is very important while studying asymmetric
geometry with one relatively flexible member. One important parameter was taken into
account, the flexible adherent eccentricity with respect to the adhesive layer. Normally,
bended plate is represented in finite element by its mean plane, so that no interfacial stresses
exists. In practice the bottom side of the plate is in contact with the adhesive which implies
different load transfer than from mean plate (e.g. shear and results). Adhesive displacement is
therefore:
(4.17)
where ux, uz are local beam displacements along x and z and θy is the local beam rotation. This
eccentricity generates many coupling between vertical reaction and bending momentum,
membrane effects and others.
4.2. Finite Element Analysis
In order to study different systems which can be used within the experiment, the plate-
adhesive configuration was varying during the finite element analysis. Parameters, their
notation and range which were studied are summarized in Table 4.2.
Table 4.2. Parameters tested during studies.
Parameter Designation Value Unit
Crack depth δa 0-3.5 mm
Plate Thickness h 1-5 mm
Adhesive thickness e 0.01-1 mm
Plate Young modulus E 5-200 GPa
Adhesive Young modulus Eadh 0.01-5 GPa
Plate Poisson coefficient ν 0.01-0.45 -
Adhesive Poisson coefficient νadh 0.1-0.499 -
Length of the free zone a 10-100 mm
Length of the bonded zone ladh 20 mm
Width of the plate b 5-50 mm
Imposed displacement Δ 1-9 mm
4.2.1. Stress state with the straight crack front
For experimental results interpretation common assumption is that the crack front is straight,
and does not change shape during propagation. Stress distribution, for the cleavage, in- and
anti-plane loading modes within the adhesive and crack front are shown in Fig.4.2. As can be
seen, stress distribution is heterogeneous along the crack front, with high concentration in the
middle of the adhesive width of mode I and II. However, near to the sample edge certain
mode III stress should be noted.
Michał K. Budzik
147
σzz
THE ADHESIVE
CRACK FRONT
-10 -5 0 5 10
0
50
100
150
200
250
300
350
400
Str
ess,
zz (M
Pa
)
y (mm)
σxz
-10 -5 0 5 10
30
35
40
45
50
55
60
Sh
ea
rin
g s
tre
ss,
xz (
MP
a)
y (mm)
σyz
-10 -5 0 5 10
-40
-30
-20
-10
0
10
20
30
40
Te
ari
ng
str
ess,
yz (
MP
a)
y (mm)
Fig.4.2. Stress state within the adhesive and at the straight crack front for three common
loading modes (input data like in Table 4.3).
4.2.2. Stress mixity
As was found previously, different stresses exists within the adhesive layer and at the crack
front. In order to study stress mixity in the studies, two, stress driven stress-mixity factors are
introduced:
a) II/I stress mixity (in-plane shearing component), defined as:
FRACTURE IN ASYMMETRIC BONDED JOINTS
148
(4.18)
b) III/I stress mixity (tearing, anti-plane shearing component):
(4.19)
where I, II, III corresponds to fracture modes.
During former analysis (section 4.2.1) it was found that the in-plane shearing stress, play the
most important role in the middle of the sample. Therefore, mode II/I was analyzed only at
this point. Contrary, tearing stress can be neglected in the middle of the sample width
(Fig.4.2), but the value of tearing stress at the sample board can play important role. Therefore
sample board was chosen for further mode III/I stress analysis. All data (constants) used for
the FEA studies of mode mixity are collected in Table 4.3.
Table 4.3. Mode mixity test panel.
Parameter Notification Value Unit
Plate Thickness h 1.6 mm
Adhesive thickness e 0.2 mm
Plate Young modulus E 70 GPa
Adhesive Young modulus Eadh 4.5 GPa
Plate Poisson coefficient ν 0.3 -
Adhesive Poisson coefficient νadh 0.45 -
Length of the free zone a 35 mm
Length of the bonded zone ladh 20 mm
Width of the plate b 25 mm
Imposed displacement Δ 4 mm
a) In-plane shearing contribution
The effects of the geometry on the modes II/I mixity are illustrated in Figs.4.3-4.7. It must be
appreciated, that the adhesive properties, not bonded plate, are playing dominant role in stress
II/I mixity. However some variation can be found for all of tested parameters of the plate,
generally change in mode mixity is small (please note the vertical scale). One of the important
findings is the crack length effect (Fig.4.3). Because of high heterogeneity of stress mixity at
the crack front for small crack lengths, initial crack of the length ca. 30 mm can be
recommended and was used in experimental part of the thesis. From geometrical aspects (we
are most of the time forced to used material of given properties) important is plate width.
Highest stress homogeneity is obtained for wider plates, although, global mode II/I mixity
also increases. In the studies these effects were balanced by using plate of 25 mm width.
Michał K. Budzik
149
Fig.4.3. Slight increase of the shearing component with crack length.
Fig.4.4. Plate width effect on I/II mode mixity (left). Plate width effect n I/II mode mixity
(right).
Fig.4.5. Plate Young modulus effect (left). Adhesive Young modulus effect (right).
20 40 60 80 100 120 140 160 180 200
10
12
14
16
18
20
(
%)
Plate Young modulus, E (GPa)
0 1 2 3 4 5
2
4
6
8
10
12
14
Adhesive Modulus, Eadh
(GPa)
-0.4 -0.2 0.0 0.2 0.4
11
12
13
14
15
b=5
b=20
b=50
y/b
10 20 30 40 50
11.8
12.0
12.2
12.4
12.6
12.8
13.0
13.2
13.4
13.6
Width, b (mm)
10 20 30 40 50 60 70 80 90 100
12.6
12.8
13.0
13.2
13.4
13.6
Crack length, a (mm)
CRACK LENGHTS RANGE IN
THE EXPERIMENTS
FRACTURE IN ASYMMETRIC BONDED JOINTS
150
Fig.4.6. Plate thickness effect (left). Adhesive thickness effect (right).
Fig.4.7. Poisson ratio effect of the plate (left) and the adhesive (right).
Anti-plane shearing (tearing) stress contribution
Results of the FEM analysis of the mode III/I mixity at the sample board are shown in
Figs.4.8-4.13.
Fig.4.8. Increase of the tearing component with the crack length.
10 20 30 40 50 60 70 80 90 100
6
8
10
12
14
16
18
20
22
(
%)
Crack length, a (mm)
0.1 0.2 0.3 0.4
12.6
12.8
13.0
13.2
13.4
13.6
13.8
14.0
(
%)
Plate Poisson ratio,
0.1 0.2 0.3 0.4 0.5
0
5
10
15
20
25
30
Poisson ratio, adh
1 2 3 4 5
12.0
12.5
13.0
13.5
14.0
14.5
15.0
15.5
Plate thickness, h (mm)
0.2 0.4 0.6 0.8 1.0
8
10
12
14
16
18
20
22
24
Adhesive thickness, e (mm)
Michał K. Budzik
151
Fig.4.9. Tearing stress factor distribution along the normalized sample width (left) and
variation of the mode mixity with the sample width (right).
Fig.4.10. Plate Young modulus effect (left). Adhesive Young modulus effect (right).
Fig.4.11. Plate thickness effect (left). Adhesive thickness effect (right).
1 2 3 4 5 6 7 8 9 10
10.5
11.0
11.5
12.0
12.5
13.0
13.5
14.0
(
%)
Plate thickness, h (mm)
0.0 0.5 1.0 1.5 2.0
12
13
14
15
16
17
(
%)
Adhesive thickness, e (mm)
20 40 60 80 100 120 140 160 180 200
8
10
12
14
16
18
(
%)
Plate Young modulus, E (GPa)
0 1 2 3 4 5
0
2
4
6
8
10
12
14
(
%)
Adhesive Young modulus, Eadh
(GPa)
-0.4 -0.2 0.0 0.2 0.4
0
10
20
30
40
50
60
b = 5 mm
b = 20 mm
b = 50 mm(
%)
y/b5 10 15 20 25 30 35 40 45 50
6
8
10
12
14
16
(
%)
Plate width, b (mm)
FRACTURE IN ASYMMETRIC BONDED JOINTS
152
Fig.4.12. Poisson ratio effect of the plate (left) and the adhesive (right).
In the modes III/I mixity near the adhesive side board (contrary to modes II/I mixity) more
important are plate properties. This is result of well known membrane effects. In fracture tests
often plates are used instead of rectangular bars. Such adherend geometry implies many
additional effects, like rotations at the plate extremities etc. which finally results in additional
stress components which are transferred by the adhesive layer. This effects arises when the
used plate is of small rigidity ratio to the rigidity of the adhesive. It must be appreciated that
choosing right sample geometry to be tested can be crucial in many aspects and should be
somehow balanced process. Change of any of tested parameters impose additional effects on
the load transfer in the system and can cause additional phenomena to occur. Sample
geometry should be easy to produce and enable comparison with the tests which were made
up to the time. All tested sample features and geometry used within the thesis are detailed in
Table 4.4. Geometrical effects on the stress state is detailed in Table 4.5.
Table 4.4. Features of the sample used within the thesis.
Parameter Notation Value Unit
Crack length a 40-105 mm
Width b 25 mm
Plate Young modulus E 50-77 GPa
Plate thickness h 1.6 mm
Plate Poisson ratio ν 0.05-0.3 -
Adhesive Young modulus Eadh 0.02-4.5 GPa
Adhesive thickness e 0.2-0.8 mm
Adhesive Poisson ratio νadh 0.45-0.49 -
Imposed displacement Δ 1.6-9.7 mm
Table 4.5. Features of chosen geometry.
y\mode component η (δa = 0 – δa = 1.5) (%) κ (δa = 0 – δa = 1.5) (%)
y = 0 13.3 – 13.5 0
y = ± 12.5 20 – 16 (but small stress level) 11 - 9
0.1 0.2 0.3 0.4
0
10
20
30
40
50
60
70
(%
)
Plate Poisson ratio,
0.1 0.2 0.3 0.4 0.5
0
5
10
15
20
25
(
%)
Adhesive Poisson ratio, adh
Michał K. Budzik
153
4.2.3. Anti-plane shearing effect on fracture structure
Figure 4.13 illustrates the mode III/I stress mixity, obtained with FEM analysis, in the
adhesive near the sample board. It must pointed out, that the stress level, when the crack
length is small, is at this stage very high (see Fig.4.2).
6 8 10 12
0
2
4
6
a = 10 mm
a = 50 mm
a = 100 mm
(
%)
y (mm)
Fig.4.13. Tearing component at different stages of crack propagation (FEM analysis).
As was noted previously [e.g. 217], cleavage-tearing stress mixity can create river patterns
phenomenon in brittle materials, i.e. epoxies. This effect was observed during microoptical
studies of the fracture surfaces of the Epidian6 (DGEBA) adhesive (Fig.4.14a). Epidian6 was
tested in the CRT test on slow strain rate tensile machine. All tested samples (also with MMT
nanoparticles) indicates the river pattern phenomenon near the side board. This characteristic,
microstructural, feature was never recorded in the adhesive joints test and within this study is
related to the FEM analysis which indicates high modes III/I mixity near the side board of the
adhesive. In Fig.4.14 a and b graphical analysis of the fracture surface is made. Using river
patterns found at the fracture surface, detailed studies of the stress state at the crack front can
be deduced. This short study indicates curved crack front. This findings stays with agreement
with the finite element studies (e.g. Fig.4.2) where high mode I stress concentration is based
in the middle of the sample width. While this stress is decreasing, moving from the centre of
the plate, mode III – anti-plane shearing plate increases. Mode I concentration can result with
the crack front curvature (studied in the next section), while mode III/I mixity in the side of
the adhesive result in river patterns phenomenon, which was observed during direct,
microscopic, observations.
FRACTURE IN ASYMMETRIC BONDED JOINTS
154
Fig.4.14. Optical micrographs of river patterns observed during the test on macro a) and
microscale b). Graphical interpretation of the surface features.
4.2.4. Crack depth
As can be seen from Fig.4.2 some stress heterogeneity occurs when the straight crack front.
Mode I stress, which drive the crack propagation is concentrated in the centre of the adhesive
width. This can result in creation of the curved crack front, where the stress state will be more
homogenous. One of the variable introduced to the analysis was crack depth, δa. Crack depth
is defined as a difference between crack length at the middle of the sample and crack length at
edge (see Fig.4.14) [204], thus δa=acentre-aside. The origin of the phenomenon is unknown but
mostly referred to the anticlastic bending effect [177,218,219]. It must also appreciated that in
bonded joints stress-strain state varies along the crack front, from plane-strain in the middle
(at y=0) to plane-stress on the side of samples (at y=± b/2). Crack length of the crack with
curved front can be defined as:
(4.20)
where n is parameter defining crack front shape.
In this study n=2 was assumed, thus giving parabolic crack front. Crack front curvature is
important parameter for the crack length estimation but often ignored by researchers. The
energy release rate, G from definition is strongly crack dependent function [204]. Ignoring
crack front curvature effect can lead to misinterpretation of the results. It was found that
energy release rate can be overestimated to ca. 40% [177]. Direct evidence of the parabolic
crack front is shown on Fig.4.15 and was found by the author while studying in all test for all
tested samples.
a)
LO
CA
L C
RA
CK
DIR
EC
TIO
NS
GLOBAL CRACK DIRECTION
MODE I
MO
DE
III
b)
PR
OV
ISIO
NA
L C
RA
CK
FR
ON
T
400 μm 100 μm
Michał K. Budzik
155
Fig.4.15. Curved crack front for two bonded with epoxy adhesive aluminium plates. Picture
made with micro camera system (Dino-Lite).
In order to study crack front curvature sample with features detailed in Table 4.6 was
analyzed.
Table 4.6. Crack depth test panel.
Parameter Designation Value Unit
Crack length a 35 mm
Width b 25 mm
Plate Young modulus E 70 GPa
Plate thickness h 1.6 mm
Plate Poisson ratio ν 0.3 -
Adhesive Young modulus Eadh 4.5 GPa
Adhesive thickness e 0.2 mm
Adhesive Poisson ratio νadh 0.45 -
Imposed displacement Δ 4 mm
Results of the qualitative analysis, made with Cast3M code are shown in Fig.4.16 in form of
three nominal stresses and their distribution in the adhesive layer. More quantitative results of
stress distribution along the crack length are shown in Figs. 4.17-4.19. Stress at two positions
were analyzed: at the middle of the width of the adhesive, and at the adhesive board (right
side of the figures). Qualitative comparison reveals that homogeneity of the stress increase
while the crack depth is increasing up to some moment, in the present study ca. 1.5 mm. From
quantitative studies can be found, that the same stress level is obtained when the crack depth
is 2.75 mm (red bold line), or 0.8 mm (blue bold line) for cleavage and in-plane shearing
modes. Homogeneity of the mode III stress is never obtained. Direct microscopic
observations yield that the crack length is ca. 2.3 mm deep (green bold line). This value is of
course closer to the mode I stress homogeneity, mode which is driving the crack. Thus, the
difference between FEM analysis and direct observation (ca. 0.45 mm) may be suspected to
come from the stress heterogeneity or from the complex stress state along the crack front.
Crack front
curvature
δa
Wed
ge
25
25
FRACTURE IN ASYMMETRIC BONDED JOINTS
156
σzz σxz σyz
Fig.4.16. Finite Element Analysis result. Stress state in the adhesive layer. In columns:
cleavage, in- and anti-plane shearing stress respectively. In rows arising crack depth, from 0-
1.5 mm. Arrows indicate change of stress level.
δa=0
δa=0.5
δa=1
δa=1.5
Michał K. Budzik
157
Fig.4.17. Cleavage stress distribution. Opening stress, ζzz vs. crack depth, δa (left) curves
showing linear relation on sample board and in the middle (right). Solid line indicate crack
depth for which stress at the width centre and side are of the same value, dashed line for
experimental value.
Fig.4.18. In-plane shearing stress distribution along the crack length (left).Shearing stress at
the crack front vs. crack depth indicating linear relations on side and in the middle of the
sample (right). Solid line indicate crack depth for which shearing stress at the sample centre
and side are of the same value.
Fig.4.19. Anti-plane shearing stress along sample width (left). Tearing stress at the crack front
vs. crack depth indicating linear relations on side and in the middle of the sample (right). Bold
lines indicate crack depth for the situation described previously.
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
0
10
20
30
40
50
60
Tea
rin
g s
tres
s,
yz (
MP
a)
Crack depth, a (mm)
yz
(y = 0) = 14.627 a+36.193
R2 = 99.96%
yz
(y = 12.5) = 0
y = 0
y = ± 12.5 mm
-10 -5 0 5 10
-40
-30
-20
-10
0
10
20
30
40
a = 0 mm
a = 0.5 mm
a = 1 mm
a = 1.5 mm
Teari
ng
str
ess
, y
z (
MP
a)
y (mm)
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
30
40
50
60
70
80
90
y = 0
y = ± 12.5 mm
xz
(y = 12.5) = 14.627 a+36.193
R2 = 99.99%
xz
(y = 0) = -6.936 a+54.4123
R2 = 99.99%
Crack depth, a (mm)
Sh
eari
ng
str
ess
, x
z (M
Pa)
-10 -5 0 5 10
30
35
40
45
50
55
60
a = 0 mm
a = 0.5 mm
a = 1 mm
a = 1.5 mm
Sh
eari
ng
str
ess
, x
z (M
Pa)
y (mm)
0.8
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
0
50
100
150
200
250
300
350
zz
(y=±12.5) = 93.282 a - 27.239
R2 = 99.28 %
zz
(y=0) = -50.089 a + 367.488
R2 = 99.98 %
Crack depth, a (mm)
Str
ess
,
zz (
MP
a)
y = 0
y = ± 12.5 mm
-10 -5 0 5 10
0
50
100
150
200
250
300
350
400
a = 0 mm
a = 0.5 mm
a = 1 mm
a = 1.5 mm
Str
ess
,
zz (M
Pa)
y (mm)
2.75
FRACTURE IN ASYMMETRIC BONDED JOINTS
158
Final studies were made in order to find out effects of the crack depth impact on the stress
mixity along the crack front. This results are shown in Figs.4.20 and 4.21. Like previously
analysis were made for the centre of the adhesive for modes II/I mixity, and for the adhesive
board in case of modes III/I mixity. As can be seen, mode mixity is not changed by the crack
depth, although some relation can be found. Linear increase of the modes II/I mixity can be
seen (Fig.4.20) but with very small (0.2%) change for the analyzed crack depth range.
Contrary, modes III/I decreases with crack depth increase with power low. This change is
3.5% for the analyzed crack depth range.
Fig.4.20. Effect of the crack depth on stress mixity factors distribution. Shearing stress
component (left). Tearing mode component (right).
Fig.4.21. Slight, linear increase of the shearing component with the crack depth increase in
the middle of the tested sample (left). Decrease of the tearing stress component with the crack
depth at the side of the sample (right).
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
13.4
13.5
13.6
a, (mm)
= 0.001382a + 0.1338
R2 = 99.82 %
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
9.0
9.5
10.0
10.5
11.0
11.5
12.0
12.5
(
%)
a (mm)
= -0.02421a0.5268
+0.1253
R2 = 99.69 %
-10 -5 0 5 10
10
11
12
13
14
15
16
17
18
19
20
21
a = 0 mm
a = 1.5 mm
y (mm)
-10 -5 0 5 10
0
2
4
6
8
10
12 a = 0 mm
a = 1.5 mm
(
%)
y (mm)
Michał K. Budzik
159
DISCUSSION OF THE TESTS
Three new, asymmetric bonded joints tests were introduced: Constant Rate Test (CRT) – with
constant deflection rate loading condition, Constant Force Test (CFT) with constant force
loading condition and Constant Displacement Test (CDT) – with constant displacement
loading condition. Results were interpreted using physical cantilever beam model and
mathematical simple beam theory. Validation of the test procedures was made using proposed
Artificial Crack Tip Test. Accurate and continuous measurements of crack propagation in the
adhesive joints were introduced. Particularly in the CDT test novel instrumented wedge
technique was introduced. For the three tests original setups were designed (using Catia v5
environment) and built. Winkler elastic foundation model was extended and used in adhesive
bonding joints. Original tests applications were proposed. Fracture of the bonded joints was
studied in micro and macro scale using the developed metrological procedures as well as
atomic force microscope (AFM) and scanning electron microscope (SEM) techniques. The
asymmetric joints geometry was analysed using finite element method analysis (FEA).
Comparison of the tests
The three tests introduced in this work promise precise and correct crack increment
measurements. The errors of the crack increment were of marking point size and were
removed from the graphs for aesthetic reasons. Systematic relative errors of the methods
presented are listed in Table C.1.
Table C.1. Possible parameters error.
Test Relative error (%)
Eh b h F Δ α
CRT
CFT
CDT
max 10
max 10
max 10
0.2
0.2
0.2
max 3.1
3.1
3.1
0.02
0.02
-
0.5
0.5
max 1.7
-
-
max 2.5
Estimation
procedure
3 point
bending test,
representative
population
Calliper gauge,
representative
population
Calliper gauge,
representative
population
Force sensor
calibration,
electric noise
Sensor
calibration,
for CDT
calliper gauge
Statistical,
electric
noise
Fig.C.1 illustrates crack length calibration curves of the three new tests obtained in the
artificial crack tip test. All tests possess linear characteristic of the estimated crack length, a
vs. reference crack length, aD, so that:
(C.1)
with α – constant.
FRACTURE IN ASYMMETRIC BONDED JOINTS
160
40 50 60 70 80 90 100
40
50
60
70
80
90
100
110
Est
imate
d c
rack
len
gth
, a
(m
m)
Reference crack length, aD (mm)
aCDT
=1.0017 aD
R2 = 99.96 %
aCFT
=1.069 aD
R2 = 99.92 %
aCRT
=1.0089 aD
R2 = 99.97 %
Fig.C.1. Crack length calibration curve of the tests.
The best precision is assured by the strain gauge technique, where the estimated crack length
can be treated as the real crack length. The less precise test is constant force. The CFT test
overestimates apparent crack length by 6.9 %.
Crack propagation behaviour
a) CRT test: in the test a constant displacement rate of the bonded plate is maintained during
the entire test. The CRT characteristic feature is linear relation between separation rate and
the crack speed. Thus one test will give one fracture energy for one crack speed, although the
separation rate will be changed during testing. The CRT test is susceptible to the adhesive
properties and highly rate dependent [220]. When the adhesive was brittle, like Epidian 6 or
cyanoacrylate the test was performed without problem. However, when the adhesive was
more ductile and tough the test could not be performed, leading to the substrates plastic
deformation which is important limitation in the applications of the CRT test.
b) CFT test: the test starts by applying constant force to the flexible plate extremity. The
initial crack growth rate is low, and increasing with time which can be critical in certain cases.
Due to the crack acceleration this test unstable, particularly when brittle adhesives are tested,
e.g. for dentistry. In this test viscoelastic properties of the adhesive and flexible substrate are
important, since at the beginning there is a possibility of the adhesive creep. The initial value
of applied force should be estimated using different method.
c) CDT test: the test starts with the wedge insertion causing adhesive break. Crack growth rate
is maximum at the test beginning and decreasing with time. Therefore the basic results and
information about adhesive fracture toughness can be found after relatively short time,
Michał K. Budzik
161
although the propagation may continue even several months [221]. The test gives possibility
of long time durability testing although can particularly suffers of the adhesive relaxation.
The comparison of the crack increment behaviour, δa in time, t for the tests studied is
illustrated schematically in Fig.C.2.
C
rack
in
cre
men
t,
a
Time, t
CRT
CFT
CDT
Fig.C.2. Crack increment behaviour of the new developed tests.
Tests applicability
a) CRT: The design is limited by the tensile machine adaptation. The designed feature (size
250x250x30mm) enable only horizontal sample position. The adaptation of facilities for
environmental tests are more complicated than for CDT and CFT tests. The test can be
performed on any flexible substrate. Environmental test are possible, although displacement
and force sensors require correct protection. Testing brittle adhesives is preferred. Tough
adhesives should be avoided due to possibility of the flexible adherend plastic deformation (at
least in the configuration studied).
b) CFT: The designed setup enable horizontal and vertical sample positioning. Size of the
entire set up is compact (size 200x200x300 mm including water tank) but not as small as for
the CDT. Test can be performed on any flexible substrate. Environmental tests can be
performed, although displacement sensor needs to be protected correctly. The test was found
not suitable to very brittle adhesives where the critical damage appeared faster (ΔtCD) than the
interval time (ΔtIT) between two measurements (see Fig.C.3). Electrical circuit with crack
increment feedback would be the best in this test.
FRACTURE IN ASYMMETRIC BONDED JOINTS
162
Fig.C.3. Schematic representation of situation from CFT test when damage time, ΔtCD is
shorter than interval time, ΔtIT.
c) CDT: Designed setup was compact (60x60x250 mm) thus giving operational advantage. In
addition the test can be performed on any flexible substrate to which strain gauge can be
attached. Any environmental test can be performed, although strain gauges needs to be
protected correctly. All systems could be tested includes brittle and ductile adhesives, variable
surface treatments, crosslinking adhesives etc. Interpretation of the results is very precise
although the metrology is the most complicated from all presented tests.
Tests requirements
All introduced tests are using standard testing laboratory equipment. It must be pointed out
that to run CRT and CFT tests special equipment is bought at once. In the CDT test the
flexible adherend needs to be instrumented each time with new, at least two strain gauges,
although some limited reutilization of the instrumented flexible substrates is possible.
Required equipment to run single experiment is detailed in Table C.2.
Table C.2. Special equipment required to run a single test.
CRT CFT CDT
Devoted setup/
features complicity
Tensile machine
Force sensor
Displacement sensor
required/limited
design
required
required
required
required/easy
-
-
required
required/medium
-
-
-
ΔtIT
Cra
ck
in
cre
men
t,
a
Time, tΔtCD
Michał K. Budzik
163
Strain gauges
Wheatstone bridge
Signal conditioning
Post processing
-
-
required
required
-
-
required
required
required/at least 2
required
required
required
Major advantages
Features need to be
bought only once,
fast testing
Features need to be
bought only once,
multichannel
Fast accurate testing,
multichannel
Major drawbacks
Required expensive
tensile machine, one
machine – one tested
sample
- Expensive
Wheatstone bridge,
limited reusability of
samples
Summary of the experimental results
Aluminium anodization
1. Electrochemical Phosphoric Acid Anodization (PAA) applied to two different aluminium
alloys gives different surface topographies – more developed in Al-Cu alloy, although Al-Mg
alloys was affected. The alumina layer was found thicker on Al-Cu.
2. Employed PAA process gives rise of the adhesion energy over simply abraded of ca. 60 %
as was proved in the CDT test. In addition PAA change the fracture nature from adhesive to
cohesive.
Constant Rate Test
1. The constant rate test was found sensitive enough to study microstructural effects in the
adhesive joints.
2. The clay nanoparticle-modified epoxy adhesive gives higher fracture energy as well as
smaller crack growth rate over the pure adhesive. The results were confirmed with the fracture
surface studies.
3. Crack propagation rate and fracture energy of the adhesive joint are rate dependant.
4. The advantage of the epoxy/nanoparticle reinforced system is more exposed when plate
deflection rates used in the test are low.
5. The fast crack rates in the asymmetric joint geometry provides river pattern phenomenon at
the sample side boards.
Constant Force Test
1. CFT test was found precise and reliable for crack length estimations.
2. CFT test was employed successfully for testing of the adhesive joints providing precise,
continuous fracture observations.
FRACTURE IN ASYMMETRIC BONDED JOINTS
164
3. The crack speed during the test accelerates to unidentified asymptote (sound of speed in the
given material?).
4. The crack speed and deflection rate were found linearly related.
5. The cyanoacrylate adhesives are characterized be the highest and the least crack speed
dependant fracture energy.
6. The observed nonlinear effects were assigned to creep or elastic deformation of the
adhesive.
Constant Displacement Test
1. CDT test was found precise and reliable for crack length estimations.
2. Surface treatment effect was successfully studied with the CDT test.
3. Double crack curvature phenomenon in the vicinity of the adhesion transition zone was
observed and successfully explained. In addition crack front behaviour in the studied
condition was proposed.
4. Observations of the crack path in the vicinity of the adhesion transition zone were made
illustrating that the crack is able to change the path from the interface to the cohesive fracture
without breaking the adhesive.
5. The interpretation of the soft adhesives behaviour was made using adopted to the adhesive
joints Winkler model.
6. Extended theory was successfully applied to study of the adhesive in metastable state.
7. Energy release rate was derivated from the Winkler model giving better inside into the
fracture mechanics of the adhesive joints.
8. The simple beam theory was found to overestimate the apparent crack length, although not
incorporating errors to energy release rate. The energy release rate derivated from Winkler
and SBT theory was found equal at least in situation studied.
10. The SBT model was found sufficient to study macroscale behaviour of the adhesive joints.
Finite Element Analysis
1. The proposed model was found reliable to study stress state as a function of the sample
geometry.
2. Proposed plate geometry provides ca. 13.3 % and 0% of the interfacial shearing and tearing
stress contribution respectively in the middle of the crack tip. The rest is opening stress.
3. The high value of the tearing stress near the side boards of the sample was found to be
responsible for the river patterns phenomenon.
4. The crack opening stress distribution provides the crack front curvature.
Michał K. Budzik
165
CONCLUSIONS
1. The novel tests designed in this work are promising in that they offer accurate, reliable
and continuous crack propagation observations.
2. The confirmation of this reliability was assessed by the simple beam theory (SBT) and
the Winkler elastic foundation model. It was found that the SBT overestimates the
apparent crack length which required correction for the experimental crack length.
However the extended Winkler model proved that this overestimation which appeared
to require correction is selfcompensating in terms of the strain energy release rate.
3. Among different materials systems studied in the Constant Force Test the
cyanoacrylate-bonded-joints gave the most promising results.
4. The novel Constant Rate Test allowed precise estimation of the improvement of the
adhesion efficiency in the aluminium joints bonded with nano-particle-reinforced
adhesive compared to the pure epoxy (30% improvement in fracture energy and 25%
slower crack propagation).
5. In all the tests performed using the different materials systems the existence of the
curved crack front was confirmed. This phenomenon helped to explain the fracture
behaviour at the boundary of the weak/strong adhesion zones.
6. The finite element method and analysis were found useful in studying the stress state
in the asymmetric bonded joints and allowed the explanation of the two phenomena:
the formation of the crack front curvature and the river patterns.
FRACTURE IN ASYMMETRIC BONDED JOINTS
166
PERSPECTIVES
The following future tasks can be proposed:
1. Performing of durability tests for different materials systems used in the shipbuilding
industry where lack of the information about the durability and reliability of the
adhesive joints is main reasons of their limited applications.
2. Simplification of the existing (developed Constant Rate Test, Constant Force Test and
Constant Displacement Test) and their interpretation.
3. Development of the new, fracture mechanics based tests for particular applications,
e.g. axissymmetric samples.
4. Fracture modelling using finite element method and finite difference method –
although adhesion is a very complex phenomenon the finite element based methods
can provide many step forward in this field. For example studying adhesive joints
under complex mechanical loading can be very expensive or impossible but can be run
with finite element code. In addition fracture criteria as well as viscoelastic properties
of the adhesives and their effect on the structure behaviour can be considered.
5. Fractography based studies – the fractographic analysis is a very important tool for the
future adhesive bonding design and can provide big step forward in this field. The
knowledge about the microscale behaviour of the adhesive bonding has rarely been
reported.
6. The extended studies of new adhesives and surface treatment using Atomic Force
Microscope and Scanning Electron Microscope should be performed giving the
possibility of direct measuring of the adhesion forces.
Michał K. Budzik
167
APPENDICES
Appendix 1. Mode II contribution in Constant Displacement Test
Since the CDT test, in principle, invoke a degree of mode II fracture in addition to the
principal mode I. Simple analysis is presented showing, that the former contribution is
negligible.
Consider the geometry shown in Fig.A.1.1, which represents two thick substrates bonded such
that failure is occurring parallel to the interface, in mode II, due to the application of force, Fx
(the lower adherend is considered fixed). If the force is stationary, but the crack grows by an
increment δa, the elastic strain energy density in the volume abh (where b is width) is
reduced, since Fx decreases, although the volume involved increases to (a+δa)bh. The net
result is a reduction in overall stored energy, and it is this that drives the crack in mode II,
requiring energy GII b δa.
Fig.A.1.1. Close-up of the beam/wedge contact zone.
From an energy balance, it can be shown that the mode II fracture energy, GII, is given by:
(A.1)
With, compliance:
(A.2)
Since:
(A.3)
Therefore:
(A.4)
α F
Fx = F sinα
Fz = F cosα
FRACTURE IN ASYMMETRIC BONDED JOINTS
168
Now consider the wedge/beam contact zone magnified in Fig.A.1.2. Assuming friction to be
negligible, the vertical force, Fz is, in fact, the vertical component, Fcosα, of the contact force,
F , normal to the beam (although the difference is very small for small α).
Fig.A.1.2. Model for mode II fracture
Similarly, there also exists a (small) horizontal component, Fx.
(A.5)
where:
(A.6)
From simple beam theory:
(A.7)
For situation studied (Fig.A.1.1):
(A.8)
We obtain:
(A.9)
Thus:
(A.10)
For small values of α valid is relation:
(A.11)
with:
(A.12)
Fx
a δa
h
Michał K. Budzik
169
we obtain:
(A.13)
The energy release rate is:
(A.14)
Finally in the present context, we may estimate the relative importance of mode II fracture
compared with mode I, from the ratio and the mode mixity rate from wedge – beam contact:
(A.15)
The worst case corresponds to the following values: h=1.6 mm, Δ=9.7 mm and a=75 mm.
These values lead to a ratio of mode II to mode I fracture energies of ca. 1.5 x 10-6
. Clearly,
neglect of mode II in this asymmetric test is of no consequence, although it may be with other
geometries.
FRACTURE IN ASYMMETRIC BONDED JOINTS
170
Appendix 2. Friction dissipation in Constant Displacement Test
During CDT test, due to propagating crack, wedge/beam friction can occurs. This situation
can leads to possible friction energy dissipation. Simple analysis is presented showing, that
the wedge/beam friction in particular cases should not be neglected .
Consider situation like in Fig.A.2.1. The crack is propagating towards x, changing the radius
of the beam curvature, R and length of bended arm S. The point A due to crack growth moves
to the A’ position. Because beam and wedge stays in touch during entire process some energy
is lost for additional friction work.
Fig.A.2.1. CDT test and the close-up of the wedge/beam friction region.
From the SBT:
(A.13)
Δ
a
s α
x
z
Fx
Φ
A'
Fz
A
α
F
R
Michał K. Budzik
171
(A.14)
The change in bended arm:
(A.15)
(A.16)
(A.17)
(A.18)
Finally:
(A.19)
For incremental growth of crack, δa:
(A.20)
(A.21)
Since along the x, crack growth is δa, beams slides on the wedge, from A to A':
(A.22)
The normal force at the wedge/beam contact:
(A.23)
Assuming Guillaume Amontons low of friction:
(A.24)
FRACTURE IN ASYMMETRIC BONDED JOINTS
172
We can write:
(A.25)
Work of the friction forces (on A-A' way) is:
(A.26)
Giving:
(A.27)
The strain energy:
(A.28)
(A.29)
(A.30)
(A.31)
Finally the work of adhesion forces is |δU|-Φ|δs|. The relative importance of friction
dissipation compared with energy release rate:
(A.32)
The worst case corresponds to the following values: aluminium wedge-aluminium plate
system, μ=1.2, using wedge of thickness, Δ=4 mm, and assuming a=35 mm, the friction
component can be ca. 5.5% of the strain energy.
Michał K. Budzik
173
Appendix 3. List of publications
Table A.1. List of publications
JOURNALS FROM JCR LIST
Journal: Title: Authors: Year
International
Journal of
Adhesion and
Adhesives
Accurate and Continuous
Adhesive Fracture Energy
Determination using an
Instrumented Wedge Test
M. Budzik , J. Jumel , K. Imielińska,
M.E.R.Shanahan
2009
Journal of
Adhesion
Fracture in Composite/
Aluminium Joints of Variable
Adhesive Properties
M. Budzik , J. Jumel , K. Imielińska,
M.E.R.Shanahan
2009
Journal of
Adhesion Science
and Technology
Effect of Adhesive Compliance
in the Assessment of Soft
Adhesives with the Wedge Test
M. Budzik , J. Jumel , K. Imielińska,
M.E.R.Shanahan
2010
(accept
ed)
OTHER REVIEWED JOURNALS
Material
Engineering
Properties of Al/CFRP composite
adhesive bonding
M. Budzik , J. Jumel , K. Imielińska
2007 in
Polish
Material
Engineering
Crack propagation in the variable
adhesion properties condition
M. Budzik , J. Jumel , K. Imielińska,
M.E.R.Shanahan
2009 in
Polish
Advances in
Material Science
Fracture of aluminium joints
bonded with epoxy adhesive
reinforced by MMT nanoparticles
M. Budzik , R.Pilawka,J. Jumel , K.
Imielińska, M.E.R.Shanahan
2009
BOOK CHAPTERS:
Modern
technologies and
energy conversion
ed. W.
Przybylskiego
Defect detection in Al/CFRP
composite bonded joints using
ultrasonic technique
M.Budzik 2008 in
Polish
CONFERENCES, ABSTRACTS, POSTERS:
Mechanics –
Gdańsk 2007 –
Poland
Defect detection in Al/CFRP
composite bonded joints using
ultrasonic technique
M. Budzik , J. Jumel , K. Imielińska 2007
Euradh – Oxford-
Great Britiain
A strain gauge technique to
monitor crack propagation in
wedge test
M. Budzik , J. Jumel , K. Imielińska,
M.E.R.Shanahan
2008
Colloque national
MECAMAT
Ecole de
mécanique des
matériaux-
Aussois-France
Essais de fissuration sur adhésif
au moyen d'un essai de clivage en
coin instrumente, mise en œuvre
et application.
M. Budzik , J. Jumel , K. Imielińska,
M.E.R.Shanahan
2009
Swissbonding
2009-Zurich-
Switzerland
A novel technique for precise
crack length measurement in the
DCB or wedge test
M. Budzik ,J. Jumel , K. Imielińska,
M.E.R.Shanahan
2009
Colloque national
MECAMAT
Ecole de
mécanique des
matériaux-
Aussois-France
Adhesive monitoring with
instrumented wedge test
M. Budzik ,J. Jumel , K. Imielińska,
M.E.R.Shanahan
2010
FRACTURE IN ASYMMETRIC BONDED JOINTS
174
Appendix 4. Gantt chart of thesis progress
Michał K. Budzik
175
REFERENCES
[1] Kaelble D.H., Physical Chemistry of Adhesion, 1975, Wiley.
[2] Kestelman V., Veslovsky R., Adhesion of Polymers, 2002, The McGraw-Hill.
[3] Griffith A.A., Philos. Trans.R.Soc., 1920, A221, 163-197.
[4] Erdogan F., Sih G.C., ASME Journal of Basic Engineering, 1963, 85D, 519–527.
[5] Parmigiani J.P., Thouless M.D., Engineering Fracture Mechanics, 2007, 74, 2675-2699.
[6] Chen B., Dillard D.A., Int. J. Adhes. Adhes., 2001, 21,357- 368.
[7] Shu L., Yuh J. Ch., Xiankui Z., 2004, 41, 6147–6172.
[8] Blackman B.R.K., Kinloch A.J., Paraschi M., Teo W.S., Int. J. Adhes. Adhes., 2003, 23,
293–305.
[9] Steinbrecher G., Buchman A., Sidess A., Sherman D., Int. J. Adhes. Adhes., 2006, 26,
644–650.
[10] Mazza P.P.A., Martini F., Sala B., Magi M., Colombini M.P., Giachi G., Landucci F.,
Lemorini C., Modugno F., Ribechini E., Journal of Archaeological Science, 2006, 33, 9,
1310-1318.
[11] The Adhesive and Sealant Council, ascouncil.org.
[12] Broxterman E., Adhesive and Sealant Council’s 2001, Convention and Exposition, 2001,
New Orleans, USA.
[13] Petrie E.M., Handbook of Adhesives and Sealants, 2000, The McGraw-Hill.
[14] Pate K.D., in Adhesion science and engineering, Ed. Pocius A.V., 2002, Elsevier.
[15] Täljsten B., Proceedings of the International Symposium on Bond Behaviour of FRP in
Structures, 2005, China.
[16] Wegman R.F., Tullos T.R., Handbook of adhesive bonded structural repair, 1992, Noyes
Publications.
[17] Baker A., Composite Structures, 1999, 47, 431-443.
[18] Avram J.B., Thesis, 2001, Air-Force Institute of Technology, USA.
[19] National Materials Advisory Board, National Research Council, Structural Adhesives
with Emphasis on Aerospace Applications, 1976, Marcel Dekker Inc.
[20] Mays G.C., Hutchinson A.R., Adhesives in Civil Engineering, 1992, Cambridge
University Press.
[21] Kecsmar J., Thesis, 2003, University of Southampton, UK.
[22] Clarke J.L., Structural design of polymer composites: Eurocomp Design Code and
handbook, 1996, Taylor and Francis.
[23] Cantrill J., Kapadia A., Pugh D., Proc. Instn. Mech. Engrs. Part M, 2004, Professional
Engineering Publishing.
[24] http://www.superyachttimes.com.
[25] Educational Materials Bonding/Adhesives Textbook, FEICA (The Association of
European Adhesives Manufacturers), feica.com.
[26] Täljsten B., FRP Strengthening of existing concrete structures – design guideline, 2004,
Luleå University of Technology, Third ed.
[27] Teng J.G., Smith S.T., Yao J., Chen J.F., Construction and building materials, 2002, 17,
447-462.
FRACTURE IN ASYMMETRIC BONDED JOINTS
176
[28] Vilnay O., The International Journal of Cement Composites and Lightweight Concrete,
1998, 10, 2, 73-78.
[29] Roberts T.M., The Structural Engineering, 1989, 67, 12, June 20, 229-233.
[30] Täljsten B., Thesis, 2002, Luleå University of Technology, Sweden.
[31] Täljsten B., Journal of Materials in Civil Engineering, 1997, 206-212.
[32] Pesic N., Pilakoutas K., Composites Part B: Engineering, 2003, 34, 4, 327-338.
[33] Henostroza G.H., Adhesion en Odontologia Restauradora, 2003, Asociacion
Latinoamericana De Operatoria Dental Y Biomateriales.
[34] Kinloch A.J., Adhesion and Adhesives science and technology, 1987, Kluwer.
[35] Irfan M.H., Chemistry and technology of thermosetting polymers in construction
applications, 1998, Springer Verlag.
[36] Hussey B., Wilson J., Structural Adhesives Directory and Datebook, 1996 Springer.
[37] Goulding T.M., in Handbook of adhesive technology ed. Pizzi A., Mittal K.L., 2003,
Marcel Dekker.
[38] Czub P., Bończa-Tomaszewski Z., Penczek P., Pielichowski J., Chemia i technologia
zywic epoksydowych, 2002, WNT.
[39] Bascom W.D., Cottington R.L., Timmons C.O., 1976, Naval engineers journal, 73.
[40] Kajiyama M., in Adhesion science and engineering, Editor Pocius A.V., Elsevier 2002.
[41] Mittal, K.L., Polyimides, Synthesis, Characterization, and Applications, 1984, Plenum
Press.
[42] Ablebond 71-1, Technical Data Sheet, 2003, Ablestik Laboratories.
[43] Licari J.J., Swanson D.W., Adhesive Technology for electronic application, 2005,
William Andrew Publishing.
[44] Schneberger G.L., Adhesives in manufacturing, 1983, Marcel Dekker.
[45] Frisch K.C.Jr., in Adhesion science and engineering, Editor Pocius A.V., 2002, Elsevier.
[46] Bragole R. A., Urethans in Elastomers and Coatings, 1973, Technomic, Westport, Conn.
[47] Ellis C., The Chemistry of Synthetic Resins, 1935, Reinhold.
[48] Gerhardt C.F., Ann. Chem., 1853, 87, 159.
[49] Schroder, Prinzhorn, Kraut K., Ann. Chem., 1869, 150, 1.
[50] Baeyer A., Dtsch. Chem. Ges.,1872, 5, 25.
[51] Kleeberg W., Ann. Chem., 1891, 263-283.
[52] Adams R.D., Comyn J., Wake W.C. Structural adhesives joints in engineering, 2nd
edition, 1997, Chapman and Hall.
[53] Lampman H.F., Reidenbach F., Engineered Materials Handbook. ASM International,
1995, Rogers Corporation, 157- 286.
[54] Stokes E.H., 40th SAMPE Symposium, 1995, 40(1), 59.
[55] McDonald R.A., Quereshi S.P., 41st SAMPE Symposium, 1996, 41(2), 1573.
[56] Ostberg D.T. 41st SAMPE Symposium, 1996, 41(2), 1459.
[57] Moreau W.M., Semiconductor Lithography, Principles, Practices, and Materials, 1988,
Plenum.
[58] Seymore R.B., Carraher C.E., Polymer Chemistry: An Introduction, 1988, Marcel
Dekker.
Michał K. Budzik
177
[59] Gardziella A., Pilato L.A., Knop A., Phenolic Resins: Chemistry, Applications,
Standardization, Safety and Ecology, 2000, Springer.
[60] Bain J.W., Hopkins D.G., J. Physic. Chem. 1925, 29, 88.
[61] Borroff E.M., Wake W.C., Trans. Inst. Rubber Ind.,1949, 25, 190.
[62] Gent A.N., Schultz, J., J. Adhes., 1972, 3, 281.
[63] Wake W.C., Applied Science, 1976, 65-71.
[64] Packham D.E., Adhesion Aspects of Polymeric Coatings, Ed. K.L. Mittal, 1983, Plenum.
[65] Wenzel R. N., Ind. Eng. Chem., 1936, 28, 988.
[66] Derjaguin B. V., Krotova N.A., Doklady, 1948, 61, 849.
[67] Bikerman J.J., The science of adhesive joints, 1961, Academic Press.
[68] Shanahan M.E.R., Rubber World, 1991, 205, 28.
[69] Columbia.edu
[70] Shukla R. K., Mencinger N. P., Solid State Technology, 1985, 67–74.
[71] Żenkiewicz M., Adhezja i modyfikowanie warstwy wierzchniej tworzyw
wielkocząsteczkowych, 2000, WNT.
[72] Briggs D., Industrial adhesion problems, 1985, Wiley.
[73] Tan N., Chin B., Lim K. H. K., Bourdillon A. J., Hewlett Packard Journal, 1998.
[74] Voyutskii S. S., Autohesion and Adhesion of High Polymer, 1963, Wiley.
[75] www.specialchem4adhesives.com
[76] Vasenin R.M., Vysokomol. Soedin., 1960, 2, 851.
[77] Comyn J., Adhesion Science, 1997, The Royal Society of Chemistry.
[78] Briggs D., Kendal C.R., Polymer, 1979,20,1053F.
[79] Jacobasch H.J., Freitag K.H., Acta Polymerica, 1979, 30, 453.
[80] Owen D.K., J. Appl. Polym. Sci. 1975, 19, 265.
[81] Petrucci R.H., Harwood W.S., Herring F.G., General Chemistry, 2002, Prentice-Hall.
[82] Miessler L. M., Tar D. A., Table of discoveries attributes the date of publication/release
for the Lewis theory as 1923, 1991, Pearson Prentice-Hall.
[83] Epley T.D., Drago R.S. J., Paint Technology, 1969, 41, 500.
[84] Hertz H., J. Reine Angew. Math, 1882, 92, 156.
[85] Hilderbrand J.H., Scott R.L., Regular solutions, 1962, Prentice – Hall.
[86] Persson B.N.J., Tosatti E., Journal of Chemical Physics, 2001, 115, 12, 5597-5610.
[87] Minford D.J., Hanbook of aluminium bonding technology and data, 1993, Marcel
Dekker.
[88] Hagemaier D.J., End Product Nondestructive Evaluation of Adhesive Bonded Metal
Joints, 1990, ASM International.
[89] Schneberger G.L., Adhesives in Manufacturing, 1983, Marcel Dekker.
[90] Kinloch A.J., Adhesion and Adhesives, 1987, Chapman and Hall.
[91] Thelen E., in Symposium on Adhesives for Structural Applications, ed. Bodnar M.J.,
1966, Interscience.
[92] Knipe R., Advanced Materials and Processes, 1997.
[93] Davies R.J., Kinloch A.J., in Adhesion ed. K.W. Allen, 1989, Elsevier Applied Science.
[94] Bijlmer P.F.A., J. Adhes., 1973, 5, 319-331.
[95] Levi, D.W., J. Appl. Polym. Sci., 1977, 32, 189-199.
[96] Kinloch A.J., Smart N.R., J. Adhes., 1981, 12, 23-35.
FRACTURE IN ASYMMETRIC BONDED JOINTS
178
[97] Brewis D.M., Comyn J., Tegg J.L., Int. J. Adhes. Adhes., 1980, 1, 35-39.
[98] Minford J.D., Org. Coat. Appl. Polym. Sci. Proc., 1982, 47, 189-193.
[99] Bishopp J.A., Sim E.K., Thompson G.E., Wood G.C., in Adhesion, Ed. K.W. Allen,
1989, Elsevier Applied Science.
[100] Selwood P.G., Maddison A., Sheasby P.G., UK Patent No. GB 2 139 540 A, Published
14th Nov. 1984.
[101] Treverton J.A., Armor M.P., Bosland A., Corros. Sci., 1992, 33, 1411-1426.
[102] Maddison A., Critichlow G.W., Proc. 16th Annual Meet. Adhes. Soc., Feb. 1993, 73-
76.
[103] Wegman R.F., Bodnar W.M., Duda E.S., Bodnar M.J., Adhesives Age 1967, 10, 22-26.
[104] Minford J.D., Adhesives Age, 1978, 21, 17-23.
[105] Packham D.E., Bright K., Malpass B.W.J., Appl. Polym. Sci. 1974, 18, 3237-3247.
[106] Schneberger G.L., Chemical and Electrochemical Conversion Treatments,
Encyclopedia of Chemical Technology, 1981, Wiley.
[107] United States Patent 3081214: Method of bonding polyethylene to corona discharge
treated polyethylene terephthalate
[108] Njegic A., Beevers A., Proc. ASE 85 Conf., 1985, 349-354.
[109] Pijpers A.P., Meier R.J., Journal of Electron Spectroscopy and Related Phenomena,
2001, 121, 1-3,299-313.
[110] Shanahan M.E.R., Bourges-Monnier C., Int. J. Adhesion and Adhesives, 1996, 16, 129-
135.
[111] Pike R.A., Patarini V.H., Zatorski R., Lamm F.P., Int. J. Adhes. Adhes., 1992, 12, 227-
231.
[112] Blackman B.R.K., Kinloch A.J. Watts J.F., Composites, 1994, 25, 5, 332-341.
[113] Egito F.D., Pure & Appl. Chem., 1990, 62, 9, 1699-1708.
[114] Uddin M.N., Koizumi K., Murata K., Sakata Y., Poly.Degrad.Stab., 1997, 56, 37.
[115] Doyle D.J., Structural Adhesives Engineering III—Conference Proceedings, PRI
Adhesives Group, 1992.
[116] Novak I., Poliak V., Int. Polym. Sci. Tchnol., 1993, 20, 5.
[117] Pleuddemann E.P., Silane Coupling Agents, 1982, Plenum Press.
[118] Schmueser D.W., J. Eng. Mater. Technol., 1990, 112, 321.
[119] ASTM D897-08 - Standard Test Method for Tensile Properties of Adhesive Bonds.
[120] Anderson G.P., Chandapeta S., DeVries K.L., in Adhesively Bonded Joints: Testing,
Analysis, and Design, ed. Johnson S., 1988, ASTM, STP-981.
[121] ASTM C297-04 - Standard Test Method for Flatwise Tensile Strength of Sandwich
Constructions.
[122] ASTM D2095-6 (2008) - Standard Test Method for Tensile Strength of Adhesives by
Means of Bar and Rod Specimens.
[123] Anderson G.P., DeVries K.L., in Adhesion and Adhesives, ed. Patric R., 1988, Marcel
Dekker.
[124] ASM-Handbook, Vol. 8, ed. Kuhn H. and Medlin D., 2000, 109-110.
[125] ISO 11003-1, Adhesives - Determination of shear behaviour of structural adhesives --
Part 1: Torsion test method using butt-bonded hollow cylinders.
Michał K. Budzik
179
[126] ASTM E229-97 Standard Test Method for Shear Strength and Shear Modulus of
Structural Adhesives (Withdrawn 2003).
[127] Yang Q.D., Thouless M.D., Ward S.M., Int. J. Solids Struct., 2001, 38, 3251-3262.
[128] ASTM D1002-05 Standard Test Method for Apparent Shear Strength of Single-Lap-
Joint Adhesively Bonded Metal Specimens by Tension Loading (Metal-to-Metal).
[129] Duong C.N., Wang C.H., Composite Repair: Theory and Design, 2007, Elsevier.
[130] Diang S., Kumosa M., Engineering Fracture Mechanics, 1994, 47, 503-519.
[131] Baker A.A., Jones R., Bonded repair of aircraft structure, 1988, Martinus Nijhoff
Publishers.
[132] Anderson G.P., DeVries K.L., Sharon G., in Adhesive Joints, Formation,
Characterization, and Testing, ed. Mittal K.L., 1984, Plenum Press.
[133] ASTM D3528 - 96(2008) Standard Test Method for Strength Properties of Double Lap
Shear Adhesive Joints by Tension Loading.
[134] Volkersen O., Die Niektraftverteilung in Zugbeanspruchten mit Konstanten
Laschenquerschritten, Luftfahrtforschung, 15, 1938, 41 - 47.
[135] Winkler E., Die Lehre von der Elasticität und Festigkeit, Teil 1, 2. H . Dominicus,
Prague, 1867 (as cited in L. Fryba, History of Winkler Foundation, Vehicle System
Dynamics, supplement, 24, 7-12, 1995).
[136] Choupani N., International Journal of Adhesion & Adhesives, 2009, 29,761–773.
[137] Griffith A.A., Philosophical Trans. Roy. Soc., 1920, 221A, 163 – 198.
[138] Orowan E., Reports on Progress in Physics XII, 1948, 185–232.
[139] Irwin G., Journal of Applied Mechanics, 1957, 24, 361–364.
[140] Erdogan E., International Journal of Solids and Structures, 2000, 27, 171–183.
[141] Xiao F., Hui C. Y., Kramer E. J., J. Mater.Sci. 1993, 28, 20, 5620.
[142] Cotterell B., Rice, J.R., Int. J. Fractr., 1980, 16, 155-169.
[143] Chen B., Dillard D.A., Int. J. Solids Struct., 2001, 38, 6907-6924.
[144] Chai H., Int. J.Fract., 1987, 32, 211-213.
[145] Lin C., Liechti K.M., J. Adhes., 1986, 24, 101-121.
[146] Sundararaman V., Davidson, B.D., Int. J. Solids Struct., 1997, 34, 7, 799-817.
[147] Pirondi A., Nicoletto G., Engineering Fracture Mechanics, 2006, 73, 16, 2557-2568.
[148] Blackman B.R.K., Kinloch A.J., Wang Y., Williams J.G., J. Mater. Sci., 1996, 31,4451-
4466.
[149] Thouless M.D., Acta. Metall.Muter., 1990, 38, 1135-1140.
[150] Chen B., Dillard D.A., Int. J. Adhes. Adhes., 2001, 21, 357-368.
[151] Chen B., Dillard J.G., Dillard D.A., Clark, R.L. Jr., J. Adhes., 2001, 75,405-434.
[152] Blackman B.R.K., Dear J.P., Kinloch A.J., Macgillivray H., Wang Y., Williams J.G.
Yayla P., J. Mater. Sci., 1996, 31, 4467-4477.
[153] Shu L., Yuh J.C., Zhu X., International Journal of Solids and Structures, 2004, 41,
6147–6172.
[154] ASTM D1062-08 Standard Test Method for Cleavage Strength of Metal-to-Metal
Adhesive Bonds test.
[155] ASTM D3807-98 (2004) Standard Test Method for Strength Properties of Adhesives in
Cleavage Peel by Tension Loading (Engineering Plastics-to-Engineering Plastics).
FRACTURE IN ASYMMETRIC BONDED JOINTS
180
[156] ASTM D3433-99 (2005) Standard Test Method for Fracture Strength in Cleavage of
Adhesives in Bonded Metal Joints.
[157] Sener J.Y., Ferracin T., Caussin L., Delannay F., International Journal of Adhesion &
Adhesives, 2002, 22, 129–137.
[158] Nairn J.A., International Journal of Adhesion & Adhesives 20 (2000) 59-70.
[159] Ikegami K., Fujii T., Kawagoe H, Kyogoku H., Motoie K., Nohno K., Sugibayashi T.,
Yoshida F., Int. J. Adhesion and Adhesives, 1996, 16, 219-226.
[160] Baldan A., J. Mat. Sci., 2004, 39, 4729 – 4797.
[161] Ashcroft I.A., Hughes D.J., Shaw S.J., International Journal of Adhesion & Adhesives,
2001, 21, 87-99.
[162] ASTM D3762-03 Standard Test Method for Adhesive-Bonded Surface Durability of
Aluminium - The Wedge Test.
[163] Ripling E.J., Mostovoy S., Patrick R.L., Mater. Res. Standards., ASTM Bull, 1964, 4,
129–134.
[164] Mostovoy S., Ripling E.J., J. Appl. Polym. Sci., 1966, 10, 1351–1371.
[165] Wiederhorn S.M., Shorb A.M., Moses R.L., J. Appl. Mech., 1968, 39, 1569–1572.
[166] Mostovoy S., Ripling E.J., J. Appl. Polym. Sci., 1969, 13, 1083–1111.
[167] Kanninen M.F., Int. J. Fracture, 1974, 10, 415–430.
[168] Kollek H., Int. J. Adhes. Adhes., 1985, 5, 75–80.
[169] Whitney J.M., Compos. Sci. Technol., 1985, 23, 201–19.
[170] Blackman B.R.K., Dear J.P., Kinloch A.J., Osiyemi S., J. Mater. Sci. Lett. 1991, 10,
253–256.
[171] Meiller M., Roche A.A., Sautereau H., J. Adhes. Sci. Technol., 1999, 13, 773–788.
[172] Blackman B.R.K., Hadavinia H., Kinloch A.J., Paraschi M., Williams J.G., Eng. Fract.
Mech., 2003, 70, 233–248.
[173] Sargent J.P., Int. J. Adhes. Adhes., 2005, 25, 247–256.
[174] Aglan H., Abdo Z., J. Adhes. Sci. Technol., 1996, 10, 183–98.
[175] Ural A., Zehnder A.T., Ingraffea A.R., Engineering Fracture Mechanics, 2003, 70, 93–
103.
[176] Davidson R., Proc. Soc. Photo-Opt. Instrum. Eng. – Int. Soc. Opt. Eng., 1988, 814,
479–489.
[177] Popineau S., Gautier B., Slangen P., Shanahan M.E.R., J. Adhes., 2004, 80, 1173–1194.
[178] Nayeb-Hashemi H., Swet D., Vaziri A., Measurement, 2004, 36, 121–129.
[179] Hwang H.Y., Kim B.J., Chin W.S., Kim H.S., Lee D.G., J. Adhes. Sci. Technol. 2005,
19, 1081–1111.
[180] Crosley P.B., Ripling E.J., J. Test. Eval., 1991, 19, 24–28.
[181] Rigby R.P., Packham D.E., Int. J. Adhesion and Adhesives, 1995, 15, 61 -71.
[182] Budzik M.K., Jumel J., Imielińska K., Inżynieria Materiałowa, 2008, 982-986.
[183] Castaings M., Hosten B., Kundu T., NDT&E International., 2000, 33, 377.
[184] Choi J., Thesis, 2003, Martin-Luther-Universitat Halle-Wittenberg.
[186] Critchlow G.W., Brewis D.M., Int. J. Adhesion and Adhesives, 1996, 16, 255-275.
[185] Bard A.J., Encyclopedia of electrochemistry of the elements, 1973, Marcel Dekker.
Michał K. Budzik
181
[187] Parkhutik V.P., Shershulsky V.I., Journal of Physics D - Applied Physics, 1992, 25,
1258–1263.
[188] Vanderlinden B., Terryn H., Vereecken J., Journal of Applied Electrochemistry, 1990,
20, 798–803.
[189] Nielsch K., Thesis, 2002, Martin-Luther-Universit¨at Halle-Wittenberg.
[190] Wehrspohn R.B., Li A.P., Nielsch K., Muller F., Erfurth W., Gosele U., in Oxide Films
ed. Hebert K.R., Lillard R.S., Mac Dougall B.R., 2000, Marcel Dekker.
[191] Timoshenko S., Strength of materials, 1930, von Nostrand Company.
[192] Alho J.M., Spencer B.D., in Statistical Demography and Forecasting, 2005, Springer.
[193] Williams J.G., J. Adhes., 1993, 41, 225–239.
[194] Cotterell B., Hbaieb K., Williams J.G., Hadavinia H., Tropsa V., Mech. Mater., 2006,
38, 571–584.
[195] Zhai L.L., Ling G.P., Wang Y.W., Int. J. Adhes. Adhes., 2007, 28, 23–28.
[196] Zhai L.L., Ling G.P., Li J., Wang Y.W., Materials Letters, 2006, 60, 3031–3033.
[197] Zhang H., Tang L.-C., Zhang Z., Friedrich K., Sprenger S., Polymer, 2008, 49, 3816–
3825.
[198] Kowalczyk K., Spychaj T., Progress in Organic Coatings, 2008, 62, 425–429.
[199] Giannakas A., Spanos C.G., Kourkoumelis N., Vaimakis T., Ladavos A., European
Polymer Journal, 2008, 44, 3915–3921.
[200] Williams J.G., Hadavinia H., Journal of the Mechanics and Physics of Solids, 2002, 50,
809 – 825.
[201] Maugis D., Barquins M., J. Phys. D., 1978, 11, 1989-2033.
[202] Hadavinia H., Kinloch A.J., Little M.S.G., Taylor A.C., Int. J. Adhes. Adhes., 2003,
23, 449-461.
[203] Destrebecq J.F., Grediac M., Sierra-Ruiz V., Compo. Sci. Tech., 2007, 67, 707-719.
[204] Jumel J., Shanahan M.E.R., J. Adhesion, 2008, 84, 788-804.
[205] Gent A.N., Petrich R.P., Proc. Roy. Soc., 1969, A310, 433-448.
[206] Sharpe L.H., J. Adhesion, 1972, 4, 51-64.
[207] Maguire J.F., Talley P.L., Lupkowski M., J. Adhesion, 1994, 45, 269-290.
[208] Inoue Y., Kobotake Y., Appl. Sci. Res., 1959, A8, 321.
[209] Kaelble D.H., J. Rheol., 1960, 4, 45.
[210] Gardon J.L., J. Appl. Polym. Sci., 1963, 7, 643.
[211] Hashemi S., Kinloch A.J., Williams J.G., Proc. Roy.Soc. London., 1990, A347, 173-
199.
[212] Williams J.G., J. Comp.Mater., 1987, 21, 330-347.
[213] Williams J.G., Comp.Sci.Tech., 1989, 35, 367-376.
[214] Kinloch A.J., Lau C.C, Williams J.G., Int. J. Fracture, 1994, 66, 45-70.
[215] Sha Y., Hui C.Y., Ruina A., Kramer E.J., Acta mater., 1997, 45, 9, 3555-3563.
[216] Tang T., Hui C.Y., Retsos H.G., Kramer E.J., Engineering Fracture Mechanics, 2005,
791–805.
[217] Hull D., Fractography: Observing, Measuring and Interpreting Fracture Surface
Topography, 2002, Cambridge University Press.
[218] Davidson B.D., Schapery R.A., J. Compos. Mater., 1988, 22, 641–656.
[219] Davidson B.D., J. Compos. Mater., 1990, 24, 1124–1137.
FRACTURE IN ASYMMETRIC BONDED JOINTS
182
[220] Du J., Thouless M.D., Yee A.F., Acta Materialia, 2000, 13, 3581-3592.
[221] Armstrong K.B., Int. J. Adhes. Adhes., 1997, 17, 89-105.
[222] Chenal J.M., Chazeau L., Guy L., Bomal Y., Gauthier C., Polymer, 2007, 48, 1042-
1046.
[223] Ferry J.D., Viscoelastic properties of polymers, 3rd
edition, 1980, John Wiley and Sons.
[224] Fernando M., Harjoprayitno W.W., Kinloch A.J., Int. J. Adhesion and Adhesives, 1996,
16, 113-119.
[225] Chiang Ch., Ma Ch., Wang F.-Y., Kuan H.-Ch., European Polymer Journal, 2003, 39,
825–830.
[226] Buch X., Shanahan M.E.R., Polymer Degradation and Stability, 2000, 68, 403-411.
Michał K. Budzik
183
LIST OF FIGURES
Fig.1.1. Global adhesive industry by region, 2007 share demand[11]. .................................... 12
Fig.1.2. World Adhesive (including sealants) demand in 2007[2]. .......................................... 12
Fig.1.3. Structural adhesive market in 2002[12]. ..................................................................... 13
Fig.1.4. JAS Grippen bonded primary elements[15]. ............................................................... 14
Fig.1.5. Bridge repaired and strengthened with CFRP patch (left). ......................................... 18
Fig.1.6. Zaragoza Bridge Pavilion. Bridge build totally from cement reinforced with GFRP
(right). ....................................................................................................................................... 18
Fig.1.7. One of the examples of structural bonding applications in medicine[25]................... 18
Fig.1.8. Structural adhesive application in dentistry. ............................................................... 19
Fig.1.9. Basic acrylic reaction[36]. .......................................................................................... 20
Fig.1.10. Cyanoacrylate monomers structure, with R is usually alkyl group[35]. ................... 21
Fig.1.11. Ethylene oxide (oxirane). .......................................................................................... 21
Fig.1.12. Bisphenol A-epichlorohydrine reaction[38].............................................................. 22
Fig.1.13. Amine – curing mechanism of epoxies: a) initial step, formation of a secondary
amine and more hydroxyl groups; b) formation of tertiary amine; c) continued crosslinking
through reaction of hydroxyl groups[38]. ................................................................................ 22
Fig.1.14. The chemical structure of BMI and MDA[40]. ........................................................ 23
Fig.1.15. a) Preparation of polyamide precursor, b) curing of polyamides by imidization[43].
.................................................................................................................................................. 24
Fig.1.16. Urethane group. ......................................................................................................... 25
Fig.1.17. Basic polyurethane polymerization reaction[45]. ..................................................... 26
Fig.1.18. Typical mechanism for a urethane adhesive bonding covalently to a polar
surface[45]. ............................................................................................................................... 26
Fig.1.19. Polymerization of phenol with excess of formaldehyde[52]. ................................... 27
Fig.1.20. General structure for a linear silicon polymer[43]. ................................................... 28
Fig.1.21. van der Waals interactions[69]. ................................................................................. 33
Fig.1.23. Schematic presentation of good and bad wetting. ..................................................... 35
Fig.1.24. Adhesive – substrate interdiffusion[75]. ................................................................... 35
Fig.1.25. The adhesive joint model, a) metal – metal joint, b) chain presentation[88]. ........... 39
Fig.1.26. Schematic presentation of metallic surface[89]. ....................................................... 40
Fig.1.27. Presentation of polymer surface[89]. ........................................................................ 40
Fig.1.28. Effect of surface pretreatment on the performance of aluminium joints with a
toughened epoxy adhesive and subjected to ageing in water[90]. ........................................... 45
Fig.1.29. Specimen configuration for adhesive tensile testing. ................................................ 48
Fig.1.30. Napkin ring test. ........................................................................................................ 50
Fig.1.31. Single lap test configuration (ASTM D1002). .......................................................... 50
Fig.1.32. Double-lap shear adhesive specimen configurations (ASTM D3528). .................... 51
Fig.1.33. Graphical presentation of shear lag model. ............................................................... 51
Fig.1.34. Peel stress distribution in shear lap geometry. .......................................................... 52
Fig.1.35. Joint in mode I fracture (F – applied force, Δ – vertical displacement, a – crack
length, δa – crack increment). .................................................................................................. 53
Fig.1.36. ASTM D1062 cleavage test of adhesive joints (left). ............................................... 55
FRACTURE IN ASYMMETRIC BONDED JOINTS
184
Fig.1.37. ASTM D3807 cleavage test specimen (right). .......................................................... 55
Fig.1.38. ASTM D3433 test specimen (left). ........................................................................... 56
Fig.1.39. ASTM D3433 TDCB test specimen (right). ............................................................. 56
Fig.1.40. D3762 wedge test configuration for durability testing.............................................. 57
Fig.2.1. Asymmetric test sample. ............................................................................................. 60
Fig.2.2. Through Transmission Method device. ....................................................................... 65
Fig.2.3. Schematic diagram of the pore formation at the beginning of the anodization. Stage1:
formation of barrier oxide on the entire area, stage 2: local field distributions caused by
surface fluctuations, stage 3: creation of pores by field-enhanced or/and temperature-enhanced
dissolution, stage 4: stable pore growth.................................................................................... 67
Fig.2.4. Topography of AA2024-T3 before (left) and after anodization (right). ..................... 68
Fig.2.5. Alumina layer obtained on Al-Cu alloy. ..................................................................... 68
Fig.2.6. Topography of AA5754-H111, before (left) and after (right) anodization. ................ 68
Fig.2.7. Symmetric and asymmetric adhesive joints. ............................................................... 69
Fig.2.8. Block scheme o sample fabrication. ............................................................................ 69
Fig.2.9. The constancy of the bondline maintained by PTFE spacers...................................... 69
Fig.2.10. Flight-view of the side-camera observations made during the tests. ........................ 70
Fig.3.1. Energy Release Rate principle (Δ-separation distance, F-applied force, a-crack length,
δa-crack length increment, U-stored elastic energy). ............................................................... 71
Fig.3.2. Constant Rate Test sample (a) and principle of the test (b). ....................................... 72
Fig.3.3. CRT test routine. ......................................................................................................... 73
Fig.3.4. Constant Rate Test physical interpretation.................................................................. 73
Fig.3.5. Scheme of the artificial crack tip test. ......................................................................... 76
Fig.3.6. CRT calibration curve. ................................................................................................ 77
Fig.3.7. Specimen deflection vs. time for two adhesive systems and two displacement rates. 78
Fig.3.8. Crack increment vs. time plots. Nonlinear graph profile (in square) followed by quasi-
constant crack growth (vc~const) (left). Nonlinear part of the graph at the onset of crack
growth (t=0-1.5h) (right). ......................................................................................................... 79
Fig.3.9. Macrographic view of fracture surfaces: River patterns regions (inside white boards)
and three distinct crack speed sections indicated with white arrows: I – fast crack speed
(probably during wedge insertion), II – medium – decreasing crack speed (probably nonlinear
part of Fig.3.8), III – stable crack growth (linear part of Fig.3.8). ........................................... 79
Fig.3.10. Average minimum fracture energy for aluminium bonded with reinforced and pure
epoxy adhesive (left). ............................................................................................................... 79
Fig.3.11. Crack growth rate vs. deflection rate for the two systems: with and without
nanoparticles (right). ................................................................................................................. 79
Fig.3.12. Optical micrographs of fracture surface of E6 resin, after lower (left) and higher
crack growth rate (right). .......................................................................................................... 80
Fig.3.13. Optical micrographs of fracture surface of E6 with nanoparticles, after lower (left)
and higher crack growth rate (right). ........................................................................................ 80
Fig.3.14. SEM fractographs of pure DGEBA epoxy for different crack speeds regimes (in
columns) and for different magnifications (in rows: 200-500-1000x). .................................... 81
Michał K. Budzik
185
Fig.3.15. SEM fractographs of epoxy reinforced with 5% of MMT 1-D nanoparticles for
different crack speeds (columns) and magnifications (rows). Arrows indicates nanoparticles.
.................................................................................................................................................. 82
Fig.3.16. SEM fractographs made at the middle of the sample (left) and near the side edge
(right). Arrows indicates features of the river patterns phenomena......................................... 83
Fig.3.17. Side view of Fig.3.16 made using optical microscope. ............................................. 83
Fig.3.18. Schematic representation of the CFT test. ................................................................ 84
Fig.3.19. Block scheme of metrological circuit........................................................................ 84
Fig.3.20. Constant Force Test physical model. ........................................................................ 84
Fig.3.21. Artificial crack test in Constant Force Test configuration. ....................................... 87
Fig.3.22. Linear force – deflection relation (left) and crack length calibration curve (right). . 87
Fig.3.23. Crack increment for cyanoacrylate (Cyanoacrylate), acrylic mastic (Mastic) and
double face scotch (PSA) adhesives (left). ............................................................................... 89
Fig.3.24. Crack speed characteristics. Vertical lines corresponds to the estimated time of
overall failure due to accelerating crack growth (right). .......................................................... 89
Fig.3.25. Fracture energy characteristics for the tested materials. ........................................... 90
Fig.3.26. The example of creep behaviour (or adhesive elastic deformation) at the beginning
of the observed in mastic adhesive. On right, close up of elastic foundation region. .............. 90
Fig.3.27. Schematic representation of the process zone. .......................................................... 91
Fig.3.28. Schematic representation of the CDT test. ................................................................ 92
Fig.3.29. CDT metrological circuit. ......................................................................................... 92
Fig.3.30. Physical model of CDT. ............................................................................................ 92
Fig.3.31. Artificial crack tip test principle................................................................................ 98
Fig.3.32. Verification of eq.3.51 (left). Estimated crack length value, a vs. strain gauge
position, x relation (right). ........................................................................................................ 99
Fig.3.33. Verification of eq.3.52 (left). Crack length calibration curve, a vs. aD (right). ........ 99
Fig.3.34. Schematic representation of tested sample.............................................................. 100
Fig.3.35. Crack length increment, δa with time, t. ................................................................. 100
Fig.3.36. Fracture energy vs. crack speed characteristic of Al/Al bonded plates with pure
DGEBA resin after different surface treatment. ..................................................................... 102
Fig.3.37. Geometry of asymmetric wedge test sample with strain gauges (dimensions in mm).
(a) Side view. (b) Top view showing position of the strain gauges. (c) Top view showing the
relative position of the wedge and one of the zones of sandblasting. .................................... 103
Fig.3.38. Crack length, a, vs. time, t, for the composite/aluminium assembly in the vicinity of
the transition from SB to P surface treatment. The transition is occurring between ca. 20 and
40 hours (left). ........................................................................................................................ 104
Fig.3.39. Results of Figure 3.38 expressed as crack speed, da/dt, vs. time, t (right). ............ 104
Fig.3.40. Model for curvilinear crack front traversing the frontier between SB and P treated
aluminium. .............................................................................................................................. 104
Fig.3.41. Fracture energy, , vs. time, t, in the vicinity of the transition zone. ................... 106
Fig.3.42. Fracture energy, , vs. crack speed, v=da/dt, in the vicinity of the transition zone.
For , f=1 (I) and for, f=0 (II). The intermediate cases (examples)
correspond to f=0.54 and 0.27, as shown schematically below (III). ..................................... 107
FRACTURE IN ASYMMETRIC BONDED JOINTS
186
Fig.3.43. Photographs of side of fractured joint in the vicinity of the transition zone between
sandblasted (SB) and polished (P) surfaces. The fracture front comes from the right. .......... 109
Fig.3.44. Photographs of side of fractured joint in the vicinity of the transition zone between
polished (SB) and sandblasted (P) surfaces. The fracture front comes from the left. ............ 109
Fig.3.45. SEM micrographs of the side of surfaces near the fracture zone in the SB treated
region. (a) The macroscopically interfacial failure at the adhesive composite interface is in
fact a cohesive failure within the adhesive, but near the interface, whereas (b), the
adhesive/aluminium interface remains intact. ........................................................................ 109
Fig.3.46. Scheme of the experiment principle. ....................................................................... 110
Fig.3.47. Crack propagation in the vicinity of the transition, STRONG/WEAK, zone. Crack is
coming from anodized (PAA) surface to polished (P). Arrows indicates regions of crack front
initiation. ................................................................................................................................. 111
Fig.3.48. Crack propagation in the vicinity of the transition, WEAK/STRONG, zone between
polished (P) and anodized (PAA) surfaces. Arrows indicates regions of crack front initiation.
................................................................................................................................................ 111
Fig.3.49. Stages of crack propagation in the sample with variable adhesion properties. ....... 112
Fig.3.50. Suggested sketch of crack front bubbling as it encounters the sharp transition
between SB and P treated aluminium. .................................................................................... 114
Fig.3.51. Schematic representation of samples tested. ........................................................... 116
Fig.3.52. Schematic presentation of Winkler (aWinkler) and cantilever beam (aSBT) crack
lengths. The difference between these crack lengths gives the transition zone between open
crack and measurable elastic foundation effect. ..................................................................... 116
Fig.3.53. Left: surface strain, |εs(x)| , at various values of x, vs. time, t. Right: |εs(x)| vs. x for a
given time shown by the bold vertical dotted line on the left part of the figure. ................... 122
Fig.3.54. Results of (negative) surface strain, |εs(x)| vs. distance, x, from wedge for the epoxy
adhesive with a 0.2 mm thick bondline (ambient temperature) (right). ................................. 123
Fig.3.55. As for Fig.3.53, but at 50°C (left). .......................................................................... 123
Fig.3.56. Results of (negative) surface strain, -εs(x), vs. distance, x, from wedge for the epoxy
adhesive with a 0.8 mm thick bondline (ambient temperature) (right). ................................. 124
Fig.3.57. As for Fig.3.56, but at 50°C (left). .......................................................................... 124
Fig.3.58. Results of (negative) surface strain,-εs(x), vs. distance, x, from wedge for the acrylic
mastic with a 0.4 mm thick bondline (ambient temperature). ................................................ 124
Fig.3.59. Crack length overestimation due to assumption of SBT ......................................... 125
Fig.3.60. Zone near crack front, on bonded side, in acrylic mastic bond, before (upper) and
after (lower) wedge insertion. Extension of the adhesive layer perpendicular to the interface is
due to imposed displacement and effect of elastic foundation. .............................................. 126
Fig.3.61. Ratio, (equation (18)) of SBT to Winkler values of the strain energy
release rate using Winkler values of the crack length, aW. ..................................................... 127
Fig.3.62. Ratio, (equation (20)) of SBT to Winkler values of the strain energy
release rate, each using own values of crack length, a. .......................................................... 128
Fig.3.63. Schematic representation of tested samples. Bonded and free zones are separated
with dash line. ......................................................................................................................... 129
Michał K. Budzik
187
Fig.3.64. Instrumented wedge test results for the adhesive subjected to slow crosslinking.
Vertical line indicates position of the crack front. Note evolution of the strain with time
(arrows indicating the direction of change) in both zones. .................................................... 130
Fig.3.65. Interpretation of experimental data (points) with Winkler based model at the
beginning (solid line) and end of the test (dashed line). Vertical line represents crack position.
................................................................................................................................................ 130
Fig.3.66. Comparison between Winkler and SBT model. Vertical lines represent: real or
Winkler crack length (solid), crack length obtained with SBT estimation after 15 days of
curing (dashed), and SBT after 2 days of curing (dotted). ..................................................... 131
Fig.3.67. Change in the adhesive parameter, λ, during crosslinking. Three stages of
crosslinking can be found (left). ............................................................................................. 131
Fig.3.68. Apparent change of the crack length estimated using simple beam theory, aSBT, as a
function of time (right). .......................................................................................................... 131
Fig.3.69. Variation of Energy Release Rate, G, with curing time. Vertical line represents the
end of the crosslinking period. ............................................................................................... 132
Fig.3.70. Schematic presentation of change of crack length during crosslinking. Not that the
real crack length remains constant, whereas estimated crack length changes due to neglect of
adhesive properties in SBT. .................................................................................................... 132
Fig.3.71. Schematic representation of tested sample.............................................................. 134
Fig.3.72. Crack propagation sequences recorded in room temperature condition, T=24oC (left)
and at the elevated temperature T=40oC (right). .................................................................... 135
Fig.3.73. Feature of fracture in elevated temperature – the adhesive crazes in the bonded zone
(indicated with arrows). .......................................................................................................... 135
Fig.3.74. Sample cooled down with wedge between separated plates. Wedge was removed
after cooling down. ................................................................................................................. 135
Fig.3.75. Strain profiles of the sample tested in ambient temperature at the beginning and end
of the test. The crack propagation from solid to dotted line (left). ......................................... 136
Fig.3.76. Strain profiles of the sample tested in elevated temperature at three stages:
beginning, middle and at the end of the test. Solid. dashed and dotted line corresponds to the
crack position at the given stage respectively (right). ............................................................ 136
Fig.3.77. Strain profiles of the samples tested in ambient temperature at the beginning and end
of the test for the 6.8 mm wedge. The crack propagation from solid to dotted line (left). .... 136
Fig.3.78. Strain profile of the sample tested in elevated temperature at three stages: beginning,
middle and at the end of the test. Solid, dashed and dotted line corresponds to the crack
position at the given stage respectively (right). ...................................................................... 136
Fig.3.79. a) Strain evolution with progressing crack. Signals from three strain gauges located
at: x1=23 mm and x6=64 mm (already passed by the crack) and x10=95 mm (passed during the
test). Last gauge signal presents evolution of signal in neighbourhood of the gauge. Estimated
process zone, or elastic foundation zone of ca. 14 mm. b) Fracture surface after the test with
visible process zone (left), close-up of the process zone (right). ........................................... 137
Fig.3.80. Crack length evolution during the test (note the joint behaviour in Fig.3.74). ....... 138
Fig.3.81. Crack propagation in time for different test conditions (left). ................................ 138
Fig.3.82. Crack propagation kinetic in ambient test condition (right). .................................. 138
FRACTURE IN ASYMMETRIC BONDED JOINTS
188
Fig.3.83. Crack propagation kinetics in elevated temperature. Five specific regions were
noted: I. Linear speed decrease, normal for propagation at constant temperature, II. Increase
of the crack speed, possibly due to the adhesive stress relief, III. Normal, stable crack
propagation. IV. Transition, stick-slip like zone, V. Acceleration of the crack close to the
sample extremity. ................................................................................................................... 139
Fig.3.84. Energy Release Rate change with time (left). ......................................................... 140
Fig.3.85. Change of the energy release rate with the crack speed. Stages I-V like in Fig.3.83
(right). ..................................................................................................................................... 140
Fig.3.86. Energy- Crack- Speed curve for elevated temperature propagation. Zone IV of ca.
12.5 mm length. ...................................................................................................................... 140
Fig.3.87. Interpretation of temperature effect with highlighted region used for elevated
temperature interpretation....................................................................................................... 141
Fig.4.1. Details of designed mesh........................................................................................... 145
Fig.4.2. Stress state within the adhesive and at the straight crack front for three common
loading modes (input data like in Table 4.3). ......................................................................... 147
Fig.4.3. Slight increase of the shearing component with crack length. .................................. 149
Fig.4.4. Plate width effect on I/II mode mixity (left). Plate width effect n I/II mode mixity
(right). ..................................................................................................................................... 149
Fig.4.5. Plate Young modulus effect (left). Adhesive Young modulus effect (right). ........... 149
Fig.4.6. Plate thickness effect (left). Adhesive thickness effect (right).................................. 150
Fig.4.7. Poisson ratio effect of the plate (left) and the adhesive (right). ................................ 150
Fig.4.8. Increase of the tearing component with the crack length. ......................................... 150
Fig.4.9. Tearing stress factor distribution along the normalized sample width (left) and
variation of the mode mixity with the sample width (right). .................................................. 151
Fig.4.10. Plate Young modulus effect (left). Adhesive Young modulus effect (right). ......... 151
Fig.4.11. Plate thickness effect (left). Adhesive thickness effect (right). .............................. 151
Fig.4.12. Poisson ratio effect of the plate (left) and the adhesive (right). .............................. 152
Fig.4.13. Tearing component at different stages of crack propagation (FEM analysis). ....... 153
Fig.4.14. Optical micrographs of river patterns observed during the test on macro a) and
microscale b). Graphical interpretation of the surface features. ............................................. 154
Fig.4.15. Curved crack front for two bonded with epoxy adhesive aluminium plates. Picture
made with micro camera system (Dino-Lite). ........................................................................ 155
Fig.4.16. Finite Element Analysis result. Stress state in the adhesive layer. In columns:
cleavage, in- and anti-plane shearing stress respectively. In rows arising crack depth, from 0-
1.5 mm. Arrows indicate change of stress level. .................................................................... 156
Fig.4.17. Cleavage stress distribution. Opening stress, ζzz vs. crack depth, δa (left) curves
showing linear relation on sample board and in the middle (right). Solid line indicate crack
depth for which stress at the width centre and side are of the same value, dashed line for
experimental value. ................................................................................................................. 157
Fig.4.18. In-plane shearing stress distribution along the crack length (left).Shearing stress at
the crack front vs. crack depth indicating linear relations on side and in the middle of the
sample (right). Solid line indicate crack depth for which shearing stress at the sample centre
and side are of the same value. ............................................................................................... 157
Michał K. Budzik
189
Fig.4.19. Anti-plane shearing stress along sample width (left). Tearing stress at the crack front
vs. crack depth indicating linear relations on side and in the middle of the sample (right). Bold
lines indicate crack depth for the situation described previously. .......................................... 157
Fig.4.20. Effect of the crack depth on stress mixity factors distribution. Shearing stress
component (left). Tearing mode component (right). .............................................................. 158
Fig.4.21. Slight, linear increase of the shearing component with the crack depth increase in
the middle of the tested sample (left). Decrease of the tearing stress component with the crack
depth at the side of the sample (right). ................................................................................... 158
Fig.C.1. Crack length calibration curve of the tests. .............................................................. 160
Fig.C.2. Crack increment behaviour of the new developed tests. .......................................... 161
Fig.C.3. Schematic representation of situation from CFT test when damage time, ΔtCD is
shorter than interval time, ΔtIT. ............................................................................................... 162
Fig.A.1.1. Close-up of the beam/wedge contact zone. ........................................................... 167
Fig.A.1.2. Model for mode II fracture .................................................................................... 168
Fig.A.2.1. CDT test and the close-up of the wedge/beam friction region. ............................. 170
FRACTURE IN ASYMMETRIC BONDED JOINTS
190
LIST OF TABLES
Table 1.1. General comparison of joining methods[13]. .......................................................... 13
Table 1.2. Values of measured Critical Fracture Energies[39]. ............................................... 23
Table 1.3. Common properties of structural adhesives. ........................................................... 29
Table 1.4. Surface topography influence on peel energy[64]. .................................................. 30
Table 1.5. Wetting conditions................................................................................................... 35
Table 1.6. Bond energy of some of the common interactions. ................................................. 37
Table 1.7. Commercial silane coupling agents[13]. ................................................................. 46
Table 2.1. AA5754 chemical composition. .............................................................................. 61
Table 2.2. Physical and mechanical properties of AA5754-H111 alloy. ................................. 61
Table 2.3. AA2024 chemical composition (supplier data). ...................................................... 61
Table 2.4. Properties of AA2024-T3 (supplier data). ............................................................... 62
Table 2.5. Properties of the CFRP composites. ........................................................................ 62
Table 2.6. Used polycarbonate properties (supplier data). ....................................................... 63
Table 2.7. Bending test results. ................................................................................................ 64
Table 2.8. Materials properties from TTM measurement. ....................................................... 64
Table 2.9. Aluminium surface treatment procedures. .............................................................. 66
Table 3.1. Results of artificial crack tip test. ............................................................................ 77
Table 3.2. Results of artificial crack tip test for CFT configuration. ....................................... 88
Table 3.3. Results of the crack length estimation with artificial crack tip test. ........................ 99
Table 3.4. The crack propagation stages in the vicinity of transition zones........................... 112
Table 4.1. Details of finite elements used in the studies. ....................................................... 143
Table 4.2. Parameters tested during studies. .......................................................................... 146
Table 4.3. Mode mixity test panel. ......................................................................................... 148
Table 4.4. Features of the sample used within the thesis. ...................................................... 152
Table 4.5. Features of chosen geometry. ................................................................................ 152
Table 4.6. Crack depth test panel. .......................................................................................... 155
Table C.1. Possible parameters error. ..................................................................................... 159
Table C.2. Special equipment required to run a single test. ................................................... 162
Table A.1. List of publications ............................................................................................... 173