User manual | /smash/get/diva2:517104/FULLTEXT01.pdf

Experimental Studies of Cold Roll
Bonded Aluminum Alloys
Steinar Lauvdal
Materials Science and Engineering
Submission date: August 2011
Supervisor:
Bjørn Holmedal, IMTE
Norwegian University of Science and Technology
Department of Materials Science and Engineering
Declaration
Declaration
I hereby declare that everything in this report is produced by the author himself. None of the
work presented here has been published before and in those cases other work or theories
has been touched or presented, this has been credited to the authorized source. Further on I
give my word that everything, to the extent of my knowledge, is carried out accordingly to
the laws of the Master of Science, Architecture and Engineering Exam at the Norwegian
University of Science and Technology.
Trondheim, August 9th 2011
________________________________
Steinar Lauvdal
III | P a g e
III
IV
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
IV | P a g e
Steinar Lauvdal
// Abstract
Abstract
This master’s thesis is based on experimental studies of the parameters influencing cold roll
bonding (CRB) of the aluminum alloys AA1200 and AA3103,in the work hardened and
annealed condition. The effect on the bond strength from the preparations parameters as
degreasing agent, scratch brushing and exposure time for oxide growth is investigated in
comparison to former studies. Further the effect of rolling speed and effect from
contributing factors from the different testing methods is discussed. Three different
methods for testing the bond strength are used. One of them was established during this
study and was named Tensile Bond Strength Test (TBST). A final investigation of the fracture
surfaces and bond interface in a scanning electron microscope (SEM) was carried out to
analyze the bond mechanism and distribution of fractured oxides.
The TBST is testing the direct bond strength with no peel or shear forces involved. It also
only requires a fraction of the sample material for testing and any roll bonded sample is
applicable for this test. These are the huge advantages with the test method. The test
method is however still naive, and suffers from a series of challenges. The current test
ranged is from 4MPa to 40MPa, but with potential for a large range expansion. Further are
bond damaging effects, caused by the machining, reducing the accuracy of the
measurements and compromising “grooving”; a measure taken for increasing the test range
above 40MPa.
The strain rate at which the samples were tested, showed to have strong influence on the
measured bond strength. Much higher than the effect of any work hardening on either of
the alloys. The preparation prior to roll bonding including an only 90s exposure time to air,
ensures a very thin oxide layer and bonding at reductions down at 22.3%. Ductile “stretch
lips” was found on the fracture surface, and run in direction normal to the rolling direction.
The fraction of bonded surface area did not seem to follow the percent of reduction during
roll bonding, which indicates a thinning of the oxide layer.
V|Page
V
VI
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
VI | P a g e
Steinar Lauvdal
Acknowledgement
Acknowledgement
This master thesis is connected to a larger project going under the name; Development of
lightweight structural materials by Accumulated Roll Bonding. This project is fully funded by
the Research Council of Norway.
In addition to the main topic of this thesis a small investigation into the preparations and
circumstances around metal-to-metal gluing has been carried out and have to be presented
in a separate report. The industrial grade glue utilized was delivered by Henkel Norway. All
of the work in this study has been carried out at the Institute of Material Technology (IMT),
NTNU, in the spring 2011.
Throughout work on this thesis several persons have been involved in the different stages.
First I want to thank my main supervisor, Professor Bjørn Holmedal, for giving me the
opportunity to work on this subject. I am grateful for all his support and help in all regarding
matters.
PhD Candidate Nagaraj Vinayagam Govindaraj has been a severe asset in many parts of the
practical work in this project. Along with his ideas and support throughout the work, I greatly
appreciate his help.
Additionally I would like to direct thanks to the following people; Paal Skaret for all help at
the MTS lab, the people at Finmekanisk Verksted in Realfagsbygget, especially Terje Rø and
the workshop at Bergbygget for helping with my customized needs.
VII | P a g e
VII
VIII
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
VIII | P a g e
Steinar Lauvdal
Acknowledgement
Table of Contents
Declaration ................................................................................................................................ III
Abstract ...................................................................................................................................... V
Acknowledgement ................................................................................................................... VII
Abbreviations .......................................................................................................................... XIII
1 Introduction............................................................................................................................. 1
2 Theory...................................................................................................................................... 3
2.1 Deformation ..................................................................................................................... 3
2.1.1 Elastic Deformation.................................................................................................... 3
2.1.2 Plastic Deformation ................................................................................................... 4
2.1.4 The Stress-Strain Curve .............................................................................................. 5
2.1.5 Strain Hardening (Work Hardening) .......................................................................... 5
2.1.5.1 von Mises Strain ................................................................................................. 7
2.2 Annealing .......................................................................................................................... 8
2.2.1 Recovery................................................................................................................... 10
2.2.2 Recrystallization ....................................................................................................... 10
2.3 Bonding ........................................................................................................................... 10
2.3.1 Metallic Bonding Mechanics .................................................................................... 10
2.3.2 Surface Interaction................................................................................................... 11
2.3.3 Oxide Layer and bonding ......................................................................................... 11
2.4 Test Methods .................................................................................................................. 13
2.4.1 Peel Test ................................................................................................................... 13
2.4.2 Shear Bond Strength Test (SBST) ............................................................................. 14
3 Experimental ......................................................................................................................... 17
3.1 The Apparatus................................................................................................................. 17
3.1.1 The Mill .................................................................................................................... 17
3.1.2 The Scratch Brushing Tool........................................................................................ 19
3.1.2 The Roughness Testing Apparatus ........................................................................... 19
3.1.2 The Tensile Testing Apparatus ................................................................................. 20
3.1.2 The Scanning Electron Microscope (SEM) ............................................................... 21
IX | P a g e
IX
X
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
3.2 The Specimens ................................................................................................................ 22
3.2.1 Annealing ................................................................................................................. 23
3.2.2 Specimen Preparation.............................................................................................. 24
3.2.2.1 Shear Bond Strength-test Samples .................................................................. 24
3.2.2.2 Tensile Bond Strength-test Samples ................................................................ 26
3.3 The Roll Bonding Process................................................................................................ 27
3.3.1 Surface Preparation for Roll Bonding ...................................................................... 27
3.4 Testing Methods ............................................................................................................. 29
3.4.1 Tensile Bond Strength Test ...................................................................................... 29
3.4.1.1 Gluing Procedure .............................................................................................. 30
3.4.2 Shear Bond Strength Test ........................................................................................ 32
4 Results ................................................................................................................................... 33
4.1 Peel Test.......................................................................................................................... 34
4.2 Tensile Bond Strength Test (TBST).................................................................................. 35
4.2.1 Strain Rate ................................................................................................................ 38
4.2.2 Gluing Results........................................................................................................... 39
4.3 Shear Bond Strength Test (SBST) .................................................................................... 40
4.3.1 Angular Deflection ................................................................................................... 41
4.4 Fracture Surface Investigation........................................................................................ 43
4.4.1 Bond Types ............................................................................................................... 44
4.4.2 Crack Direction ......................................................................................................... 44
4.5 Surface Roughness.......................................................................................................... 46
4.6 The Effect of Rolling Speed ............................................................................................. 46
4.7 Bond Interface ................................................................................................................ 47
5 Discussion .............................................................................................................................. 49
5.1 Material Selection........................................................................................................... 49
5.2 Surface Preparation ........................................................................................................ 49
5.2.1 Degreasing ............................................................................................................... 49
5.2.2 Scratch Brushing ...................................................................................................... 50
5.2.3 Brush Speed and Force ............................................................................................ 51
5.2.4 Effect of Oxide Layer ................................................................................................ 52
5.3 Tensile Bond Strength Test (TBST).................................................................................. 54
X|Page
Steinar Lauvdal
Acknowledgement
5.3.1 The Machining of the Disc Samples ......................................................................... 54
5.3.2 The Grooves ............................................................................................................. 56
5.3.3 A General Overview ................................................................................................. 57
5.3.4 Strain Rate ................................................................................................................ 59
5.3.5 Glue Limitations ....................................................................................................... 59
5.3.6 TBST vs. SBST ............................................................................................................ 60
5.3.7 Applicability of the TBST Method ............................................................................ 61
5.4 Shear Bond Strength Test (SBST) .................................................................................... 61
5.4.1 Angular Deflection ................................................................................................... 62
5.5 Fracture Surface Investigation........................................................................................ 62
5.5.1 Bond Types ............................................................................................................... 62
5.5.2 Crack Direction ......................................................................................................... 63
5.5.3 Bonded Area ............................................................................................................ 63
5.6 The Effect of Rolling Speed ............................................................................................. 66
5.7 Below Critical Deformation Threshold (CDT) ................................................................. 68
5.7.1 Al 1200NA ................................................................................................................ 68
5.7.2 Al 1200A ................................................................................................................... 68
5.7.3 Al 3103NA ................................................................................................................ 69
5.7.4 Al 3103A ................................................................................................................... 69
5.8 Bond Interface ................................................................................................................ 69
5.8.1 Theory I: Thinning of Oxide ...................................................................................... 71
5.8.2 Theory II: Uneven Oxide Layer Thickness ................................................................ 72
5.8.3 Theory III: Crack Direction........................................................................................ 72
5.8.4 Theory IV: “Pulverized” Oxide.................................................................................. 72
6 Conclusion ............................................................................................................................. 75
The Material Effect ....................................................................................................... 75
Acetone vs. ethanol ...................................................................................................... 75
The Effect of the Oxide Layer Thickness ...................................................................... 75
Effect of rolling speed .................................................................................................. 75
Effect of the General Preparation ................................................................................ 75
The Tensile Bond Strength Test (TBST) ........................................................................ 75
Angular Deflection in the Shear Bond Strength Test (SBST) ........................................ 76
XI | P a g e
XI
XII
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
TBST vs. SBST ................................................................................................................ 76
Fracture Surface ........................................................................................................... 76
7 Suggestion to Future Work ................................................................................................... 77
Scratch Brushing ........................................................................................................... 77
Rolling speed & Adiabatic Heating ............................................................................... 77
Simulated Accumulated Cold Roll Bonding .................................................................. 77
Fraction Bonded Surface Area vs. Reduction and Bond Strength ............................... 77
TBST Samples................................................................................................................ 77
SBST .............................................................................................................................. 78
8 Bibliography........................................................................................................................... 79
Appendix................................................................................................................................... 81
A - Rolling Progression to Sheets .......................................................................................... 82
B - Tables of all Tensile and Shear Samples .......................................................................... 83
C - Adhesion Log ................................................................................................................... 87
D - SEM Pictures: Tensile Samples ........................................................................................ 88
E - SEM Pictures: Shear Samples .......................................................................................... 98
F - SEM Pictures: Bond Interface ........................................................................................ 102
XII | P a g e
Steinar Lauvdal
/Abbreviations
Abbreviations
A(C)RB
Accumulated (Cold) Roll Bonding
TBS(T)
Tensile Bond Strength (Test)
SBS(T)
Shear Bond Strength (Test)
CDT
Critical Deformation Threshold
HAZ
Heat Affected Zone
CSRB
Cross Shear Roll Bonding
RB
Cold Roll Bonding
AT
After Tape
BT
Before Tape
XIII | P a g e
XIII
1 Introduction
1 Introduction
Traditionally joining methods for metals have in large consisted of fusion welding,
where the two metals, with or without the use of filler material, are melted by electric
arc, laser or electron beam. Bonding occurs in the mixed pool of liquid metals. A lot of
the energy used in these processes goes to heating up the metal in and around the
weld zone. This zone is commonly known as the heat affected zone (HAZ). In aluminum
welding, the HAZ leads to huge concerns as it causes changes in the microstructure
that inflict permanent degraded mechanical properties in the base material [19]. The
grain size is increased resulting in a strongly affected strength and ductility properties.
The HAZ is usually the most critical area for any loading-bearing constructions.
Therefore alternative low temperature joining methods are desired. Methods like buttwelding (for wires), roll bonding and accumulated roll bonding (ARB), for plates, can be
performed even at room temperature with high mechanical pressure.
Roll bonding was first applied in the production of compound plates in 1935. ARB [26]
is the natural progression of roll bonding, where the process is simply repeated. The
roll bonding process is described in more details throughout this report, but in general
it is two metal plates pressed together between two rolls at very high pressure,
resulting in permanent metallic bonding. When two plates are welded together, why
not weld together several layers? The roll bonded plates can be folded and roll bonded
again, now containing four layers welded together. In this manner, for each new pass,
the number of layers doubles and the thickness of each individual layer is reduced,
creating a very fine grained structure. This is the definition of accumulated roll
bonding, or accumulated cold roll bonding (ACRB) if performed at low temperature.
When this method is fully mastered it can offer plates with properties tailor made for a
wide specter of uses. Different materials and alloys can be mixed to create the desired
properties, or simply to make a material so hard that it is pushing the limit of the
theoretical maximum strength.
As for today roll bonding is commonly used in a few well known every day products.
Parts of the exhaust system on your car are most likely roll bonded layers of aluminum
and steel; an inner layer of steel to strengthen the tube, and an outer aluminum layer
to shield the steel from corrosion.
The aim of this project is to explore the bonding mechanisms between double layer
roll bonded plates in a selected array of aluminum alloys. This project serves as a
subtopic under the ongoing project on accumulated roll bonding in the department of
physical metallurgy at NTNU.
1|Page
1
2
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
2|Page
Steinar Lauvdal
2 Theory
2 Theory
This chapter presents the most important theories, techniques and methods, upon
which this thesis bases its interpretations of observations and assumptions.
2.1 Deformation
The deformation of a metallic material has large Impact on its properties, altering the
strength, ductility, electrical conductivity and many more. Most of these effects can be
explained by studying the changes in the microstructure. In this section, an
introduction in what deformation of a material involves with regards to roll bonding is
given.
2.1.1 Elastic Deformation
The definition of elastic deformation is this point where the external load applied to a
solid material is no greater than the material will regain its original dimensions when
the load is removed. A common example of elastic deformation is a rubber band.
When the rubber band is stretched, it is deformed. If the shape is restored to original
after stretching, the deformation has been elastic. On a much smaller scale this means
that the atomic structure can be stretched or compressed, but the atoms relative
position to each other is kept unchanged during loading, as illustrated in Figure 1. A
plastic deformation of 0.2% after applied loading is considered the upper limit of
elastic deformation. Comparing this to the analogy of the rubber bad, if the length of
the band is less than 0.2% longer after the stretching cycle, the deformation is still
considered fully elastic.
Figure 1: Elastic deformation with force applied. [2]
Below the elastic limit the behavior of a solid follows Hooks Law:
σ =εE
(1)
3|Page
3
4
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
where σ is the stress, ε is the strain and E is the modulus of elasticity (Young’s
modulus). In reality most forces is not applied normal to any shape and no material has
a 100% perfect structure. This fact introduce the shear stress, and it is defined as
τ=Gγ
(2)
where G is the shear modulus and γ is the shear strain. [1]
2.1.2 Plastic Deformation
Plastic deformation is when an external load leaves a solid material in a permanent
irreversible deformed state after the load is removed. Explained with the rubber band
example, the deformation is plastic from the point where the band is more than 0.2%
longer after a stretch cycle. An even better example, from everyday life, is to use
something that is much less elastic, like a caramel. Just as the rubber band, this too can
be stretched, but when the force is released, the caramel will remain in its stretched
form. This deformation is fully plastic, meaning it is permanent. In Figure 2 below, one
can see that the atomic structure is severely deformed under a plastic deformation in a
metal.
The mechanics of the metal deformation relays upon movement of dislocations and
the slip systems. Deformation of this type tends to occur stepwise, since metals are
crystallographic, and always in the direction of least resistance. When moving through
the material the dislocations can generate new defects, be stopped or annihilated
when they interact with other defects throughout the material.
Figure 2: Progression of plastic deformation with load applied. [2]
Once plastic deformation starts, only a small increase in stress usually causes a
relatively large additional deformation. This process is called yielding, and this
behavior starts to be important at the stress value known as yield strength, σ0 [1].The
yield strength is a practical engineering limits that marks the transition between elastic
and plastic deformation and is show in Figure 3 as the “Yield Strength Point”.
4|Page
Steinar Lauvdal
2 Theory
2.1.4 The Stress-Strain Curve
A useful tool in the engineers’ toolbox is the stress-strain curve. This curve, shown in
Figure 3, describes the strength of a material as it is being strained until fracture. The
first linear part of the curve shows the elastic deformation part which where explained
in the previous section. The “yield strength point” marks the start of the plastic
deformation, which last until fracture at the end of the curve. On the curves highest
point, called the ultimate tensile strength (UTS), marks the highest stress for this
material in regard to its initial dimensions. In reality the specific stress of the material
(called the true stress) continues to increase until the point of fracture. When
calculating the true stress, the measured stress is divided on the true area, which
decreases as the material is stretched thinner. The reason the stress-strain curve
decreases after stretching the UTS is because the stress still is divided on the initial
cross-section area of the sample. This stress, which does not take in account the
reduction in cross-section, is called the engineering stress.
Figure 3: Stress-strain curve.
The engineering stress is defined as
S =P/A0
(3)
where P is the load on the specimen and A0 is the initial cross-sectional area near the
center of the specimen. The engineering strain describes the elongation/compression
of the sample, and is defined as
e=Δl/l=(l-l0)/l0
(4)
where l is the gauge length at a given load and l0 is the original gauge length with no
load. [1]
2.1.5 Strain Hardening (Work Hardening)
Strain hardening, more commonly known as work hardening or even cold working, is
when the strength of a material increases during plastic deformation. This
5|Page
5
6
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
phenomenon is directly related to the interactions between internal dislocations and
boundaries within the material. In more technical terms, strain hardening describes
the rise in the stress-strain curve after yielding, as the material is increasing its shear
stress with increasing strain. A measurement of the degree of strain hardening due to
Hollomon’s equation is
(5)
where σ n is the stress, K is the strength index, εp is the plastic strain and n is the strain
hardening exponent:
n= σu/ σo
(6)
where σu is the ultimate tensile strength and σo is the yield strength. This is illustrated
in Figure 3.
The strain hardening of materials evolves in stages, where the first two stages describe
the different hardening rates between single crystals and polycrystals as slip systems
are activated.
Stage I and II is considered irrelevant for roll bonding, which occur at much higher
deformation. These first two stages will not be explained in further detail, although
more information on the subject can be found in “Mechanical Metallurgy” by Dieter,
listed in the bibliography [1].
Going straight for stage III, the movement of screw dislocations is introduced, which
allows piled up dislocations from stage II to escape and travel longer distances. There
probability for these dislocations to meet another screw dislocation is high, and when
sufficiently close to each other, they will start to annihilate or recover. Annihilation
occur when dislocations with opposite direction to each other intercept and both
dislocations “dissolves”. The recovery phenomenon reduces the dislocation storage
rate, and is explained closer in section 2.2.1. Both phenomena results in a reduction in
dislocation density-increase, and the strain hardening rate is dampened. This stage is
highly temperature-dependent., and both stage II and II is shown in Figure 4.
The fourth and most relevant stage for cold rolling is shown in Figure 4. The hardening
rate at this stage is constant until saturation level is reached and has a dependency on
the deformation temperature as well as the alloy composition. This is a very highly
strained area where the dislocations have very little space to move, along with an
indirectly increase in dislocation-density due to shrinking subgrain size. As some
supplementary information on accumulated roll bonding (ARB), the thickness of each
layer would decrease for every rolling pass, reducing the dislocations ability to move.
In theory, with sufficient roll bonding passes, one could in the end have a structure
6|Page
Steinar Lauvdal
2 Theory
where all the dislocations were entirely immobilized, and the theoretical maximum
strength is obtained.
Figure 4: Graph showing the strain hardening though stage II, II and IV. [15]
2.1.5.1 von Mises Strain
In 1913 Richard Edler von Mises proposed that when the second deviatoric stress
invariant J2 reached a critical value k2 yielding would occur. [1]
(7)
In layman’s terms the von Mises criterion describes the point where the material starts
to “yield plastically” after a certain amount of elastic energy is reached.
When performing cold roll bonding it is more practical to calculate the von Mises strain
directly from the reduction of the plate thickness-reduction
(8)
Von Mises stain does not take in account if the material is anisotropic. Meaning, if such
a material was strained from different directions, each direction would give a different
von Mises value. The advantage by using von Mises is due to this exact same “flaw”, as
it can easily be used to compare strain applied form a wide number of various
methods. In this case it is used to measure the strain by rolling.
By rearranging the nominal strain
(9)
and when implemented it in equation (8), the von Mises Strain expressed by the
nominal strain can be shown as
(10)
7|Page
7
8
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
In equation (8), (9) and (10) t1 is the plate thickness before the deformation, and t2 the
thickness after the deformation has occurred. Below in Figure 5, the relations between
von Mises criterion and Tresca Criterion are illustrated where σ is the yield stress.
Figure 5: Failure criteria for plain stress. Elliptical line showing the von Mises Criterion and dashed line
showing Tresca Criterion. [1]
2.2 Annealing
Cold rolling prior to roll bonding inflicts the material with high stresses that gives high
strength, but may also decrease the ductility severely. To eliminate some or all of the
effects of this work hardening, a heat treatment called annealing may be performed.
However, the strength gained by cold rolling will decrease during this process as well.
There are two main softening reactions occurring when a heavily deformed material is
annealed, and these are called recovery and recrystallization. [9] Before explaining
these two phenomena in more detail, let it be said that recovery and recrystallization
is competing processes driven by the same stored energy, created by the deformation.
Once the stored energy has been “consumed” by one or the other, no further recovery
or recrystallization can occur. Hence they are strongly dependent on each other. [16]
8|Page
Steinar Lauvdal
2 Theory
Figure 6: The effect of work-hardening and annealing temperatures on material properties. Here
shown for a Cu-35% Zn-alloy with an end deformation of 75%. [9]
Note that the effects of work hardening and annealing temperatures in Figure 6 above
does not represent the Aluminum-alloys in this report, but that of a Cu-Zn alloy. The
graph is included here in the absence of a more relevant one. However, the general
idea is the same and in fact almost identical, except for what happens with the
elongation for a temperature of about 400 degrees and up, it continues to increase.
Figure 6 is showing, along with elongation, electrical conductivity and grain size, how
the tensile strength is effected by in the different stages of an annealing process on
work hardened (cold worked) materials.
Figure 7: The effect of annealing temperature on the microstructure of cold work hardened metals. (a)
work hardened, (b) after recovery, (c) after recrystallization, and (d) after grain growth. [9]
9|Page
9
10
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
2.2.1 Recovery
After work hardening the microstructure is deformed containing a large number of
tangled dislocations (Figure 7 (a)). If the material is given a low-temperature heat
treatment, the additional energy allows the dislocations to move and form new
boundaries. This new formation is called a polygonized subgrain structure, and is
illustrated in Figure 7 (b). The term for this stage is recovery and refers to the changes
in a deformed material which partially restore the properties to its pre-deformed
state. Residual stresses due to work hardening are removed, although no change in
dislocation density occurs. At this temperature the mechanical properties is relatively
unchanged. [9][16]
2.2.2 Recrystallization
If the temperature is increased sufficiently, new fine grains will then nucleate on the
cell boundaries of the polygonized subgrain structure formed in the recovery phase,
and at the same time eliminating most of the dislocations. This large decrease in
dislocations reduces the strength of the material, but also increases its ductility. The
temperature at which the dislocation density is rapidly reduced is called the
recrystallization temperature. This process of formation of new grains is called
recrystallization, and is illustrated in Figure 7 (c).
As the amount of work hardening increases, the recrystallization temperature
decreases. There is a minimum amount of work hardening for which recrystallization
will not occur, which applies for deformations below 30 to 40%. This again has a direct
relevance for roll bonding of aluminum, where bonding can occur well below these
values. [9]
Increasing the temperature even further will initiate grain growth shown both in Figure
6 and Figure 7 (d). This stage of annealing is however of little interest in regards of
most cold roll bonding situations, as annealing at these temperatures is not desired
due to the rapid grain growth.
2.3 Bonding
2.3.1 Metallic Bonding Mechanics
The atoms in a metal are generally built up in a regular grid where nearby atoms share
their valence electrons in what is commonly called an “electron sea”, as illustrated in
Figure 8. This way of bonding gives the matrix a strong, but non-directional strength, a
high ductility and Young’s modulus, among other characteristics. The way metals bond,
is by an electromagnetic interaction between delocalized electrons, which is illustrated
by the “electron sea”-parallel. For two metallic atoms to interact in such a bonding the
distance between them needs to be less than one atomic radius. At this point they
spontaneously bond by sharing their valence electrons. This is due to the attractive
10 | P a g e
Steinar Lauvdal
2 Theory
forces decreases with the square of the distance. Put differently, two metallic plates
need to be brought very close for metallic bonding to take place.[5]
Figure 8: Metallic binding where protons are shearing their valence electrons in a so-called "electron
sea".
2.3.2 Surface Interaction
The surface of even a fine rolled plate has a certain roughness and is far from being
perfectly smooth. This roughness results in a major reduction in contact area when the
two surfaces are brought together, as shown in Figure 9. When the normal load is
increased the real contact surface is increasing proportional to the load applied.
Figure 9: Contact area between two surfaces. [5]
Under this load the contact area is experiencing plastic deformation which hardens the
material in contacting region and contributes to restraining further deformation and
growth of contact area. In roll bonding of aluminum such rough surface texture can
however be an advantage when it comes to bonding strength. The reason why will be
explained and discussed later on.
2.3.3 Oxide Layer and bonding
Most metals, when they are exposed to oxygen in the atmosphere, over time develop
an oxide layer on its surface. Such oxides, variously named, are hated as well as loved,
depending on which material of where they are found. On iron based materials it is
called corrosion, and is very porous, allowing for the oxidation to continue “eating” up
11 | P a g e
11
12
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
the metal. For other materials this layer helps protect in different ways depending on
the material.
Figure 10: Oxide layer on a aluminum surface. [3]
On aluminum the natural oxide layer grows fast, thick and hard. In fact it is the second
hardest substance known to man (sapphire), only second to the diamond. This hard
surface shields the base material from any further oxidation (corrosion), but also acts
like a barrier for when two surfaces are attempted joined. The oxide layers must be
fractured to allow the two base materials beneath to interact. This mechanism
elevates the energy required for cold welding of two surfaces. When a force is applied
and the two surfaces are brought together, the asperities are the first to come in
contact with each other, as seen in Figure 9 and Figure 11. Adhesion between the two
oxide layers now makes them act as one. With increasing load the hard and brittle
oxide layer starts to crack, and in between these cracks the base material is extruded
and with sufficient force brought in contact with the base material from the other
plate and metallic bonding is acquired. This exposed virgin material will not form any
new oxide, as the compression forces are sufficient to create an airtight seal around
these openings. Figure 11 illustrates this progression from (a) through (d) with
increasing external load.
The obtained weld strength depends on the area of actual bonded base material which
can be expressed as the exposed surface
Y=(A1-A0)/A1
(11)
where A0 is the initial surface area and A1 is the final surface area after roll bonding. [5]
12 | P a g e
Steinar Lauvdal
2 Theory
Figure 11: The progression of bonding between two aluminum layers under increasing load. [5]
2.4 Test Methods
Prior to this study, two methods for testing he bond strength was considered and also
put to the test. This next part will present these two test methods.
2.4.1 Peel Test
The peel test was initially established for testing adhesive bonding, and there is made a
standard for this peel method, which describes methods and procedures that in lag ear
applicable for peel testing of roll bonded plates as well. [25]
The peel test, as the name reveals, tests the bond strength by peeling the two
layerered sheet apart from one side to the other, as illustrated in Figure 12 b.
The preparations for a peel test are simple and require very simple tools, though the
preparations start ahead of the roll bonding process. Once all the sheets are cut to
desired size for roll bonding, one has to make a decision. To be able to peel test the
sample after it has been bonded, a section of the sheet has to be fairly easy to peel
open, so that the tensile machine have something to fasten its grip onto. To attain
such a section, there are many choices, but the basic rule is that any bonding have to
be prevented or severely reduced in this section. One method is to add some nonmetallic material in between the sheets in this section to prevent bonding. This could
be a piece of tape, some oil, or basically anything that prevents bonding.
When using such methods one should be wary on how this could influence the rest of
the bond. As discovered in a study by Lauvdal [10], when using a piece of tape, a
section stretching several centimeters into the bond was affected by the glue in the
tape. The tape had been placed in the center of the sheets, so that the moisture in the
tape (glue) has been squeezed ahead of the bond front moving the moisture far into
the back side of the sheet. In Figure 32 a sample from this study, with the mentioned
tape-issues can be found. Needless to say this reduced the bonding strength in the
affected area. A secondary problem with using tape was that the material in that area
fractured and cracked to such a degree that the material fractured when any load was
13 | P a g e
13
14
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
applied during testing. One simple solution to both these problems was found from
these mistakes; simply avoid scratch brushing the area intended for grip, not using any
tape, oils or any other material to prevent bonding. This method prevented strong
bonding and the material was not left in a fractured state.
Figure 12: Illustration showing how the peel- test is performed. [6]
It is difficult to compare the results from a peel test carried out in different studies. As
the results is usually measured in the load force applied to part a given area. The peel
test is a continuous process and gives no real cross section to divide the load. The
thickness and bond strength is also affecting the angle at which the sheets are parted,
which directly influences the measured strength. Strong bonds will be peeled in a 90
degree angle apart from each other as in Figure 12 b, while weak bonds and/or thicker
sheets will be peeled at a lower angle, giving more leverage on the crack front,
resulting in a lower measured bond strength. These are some of the weaknesses in the
peel test.
However, the advantage of such a test is that it only requires simple cutting tools and a
tensile testing machine, which should be relatively easily accessed by people that have
interest in performing such tests in the first place.
2.4.2 Shear Bond Strength Test (SBST)
The shear bond strength test, sometimes just called shear test, though it should not be
confused with a traditionally shear test which is usually related to testing the shear
force within a solid material, which has nothing to do with the testing of bond strength
as in this case. The SBST can be referred to as the “state of the art” method for testing
bond strength, as it is very adaptable for testing both weaker and stronger bonded
samples, regardless of the thickness of the sheets.
14 | P a g e
Steinar Lauvdal
2 Theory
Figure 13: This image shows how the shear-test sample is cut and where the tensile force is applied.
The actual dimensions will vary with the individual test and material properties.
Compared to the peel test a SBST requires no additionally preparation prior to roll
bonding. After the bonding is complete however, some more high-tech equipment is
needed. The first step is to cut the outer shape of the sample, like the item to the left
in Figure 13. Further, each layer is dislocated on each side by cutting a groove across,
so that only one section is overlapping in the center. This is illustrated in Figure 13 as
well.
(12)
where l is the overlap length, τw=τ- τ1 is the final thickness of the weaker material.[17]
15 | P a g e
15
16
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
16 | P a g e
Steinar Lauvdal
3 Experimental
3 Experimental
This chapter contains a presentation of the materials and apparatus used in the
experiments. The methods for preparing and testing the samples are explained in
detail, along with limitary factors and special concerns connected to these. Also the full
process of cold roll bonding is described.
3.1 The Apparatus
3.1.1 The Mill
For these experiments a custom built mill where used, therefore there are no
datasheet or any other specified data to find on the mills specifications. For this
reason, any desired specification had to be measured and calculated while operating
the machinery. A separate hydraulic engine delivers the power, with a pressure fixed
at 90 psi for both the pre rolling of the material and the cold roll bonding part of this
experiment. The rolling speed is controlled by software that allows variation of the
power to the engine. When measuring the speed on the roll surface, the velocity
showed to be exponentially proportional with the speed-scale in the software, like the
measured data can indicate in Figure 15. Figure 14 below, shows a picture of the
operational part of the machinery with a close-up picture of the upper roll, which is
205 mm in diameter. To adjust the gap between the rolls, the wheel on top of the mill
is rotated left or right. The solid steel goods helps the mill holding a steady gap-size
during rolling, but still the goods yield slightly when very hard material is rolled.
Figure 14: A picture of the mill with a close-up of the upper roller. The lower roll is below, not visible
on this picture.
17 | P a g e
17
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
As the mill has no digital or even any usable analog scales for adjusting the roll gap, all
rolling was performed on trial and error, mixed with an increasing amount of
experience and a lot of calibration samples.
Table 1: Measured dimensions and velocities of roller.
Roll Circumference
Roll Diameter
645 mm
205 mm
Speed 2
Speed 3
Speed 4
9.11 mm/s
26.62 mm/s
67.32 mm/s
The speed on the surface of the roll was found by measuring the length around the roll
and timing the rotation-time. Three different speeds where chosen for the
measurement, speed 2, 3 and 4 in the software scale, going from 0 to 10. All the
samples in this experiment was roll bonded with a speed of 3 (26.62 mm/sec). The
additional two speeds were measured as a comparison to this one. When the tests
were carried out, there was no resistance on the roll, which is likely to have an
influence to some degree. The energy required to deform the samples will absorb
some of the speed at which the sample passes through the mill. The closer this
resistance is to the maximum capability of the mill, the slower it goes. When pushing
and exceeding this limit, the mill struggles and finally stops with the sample half way
through. This factor was not investigated any further than this. For calculating the
diameter of the roll, the same mentioned measurements were used, and inn Table 1,
all the measured specifications can be found.
Velocity on roll vs Speed scale in the software
80
70
Velocity [mm/s]
18
60
50
40
30
20
10
0
0
1
2
3
4
5
Speed Scale
Figure 15: Graph showing the speed values in Table 1, indicating the start of an exponential increase.
18 | P a g e
Steinar Lauvdal
3 Experimental
3.1.2 The Scratch Brushing Tool
The surface of the connected layers where scratch brushed with a rotating steel brush
with dimensions noted below, and shown in Figure 16. A INOX FLEX LE 14-7 125 IWOX
was used for the scratch brushing, an electrical multi-speed angle grinder.
Table 2: Dimensions of the steel brush (left in Figure 16) and the rotation speed.
Brush diameter
Wire Diameter
Rotation Speed
100 mm
0.3 mm
3800 rpm
The choice of brushing parameters was based upon a previous study by Lauvdal [10],
where a larger variety of both brush types and rotation-speeds were tested.
The reason this particular rotation speed was chosen, relayed upon 2 factors. The first
being the desire for a high speed, as this increases the hardening of the surface
material. A harder and less ductile surface cracks easier and allows for metallic
bonding. All this is explained in further detail later in this report. The second factor is
the one preventing the highest rotation speed to be chosen. It is established that with
higher rotation speed the more sever is the deformation on the surface. On soft
materials, like the annealed 1200 alloy, a high rotation speed removes a lot of material
in a short instant. To make the grinding more controllable for the operator, a lower
speed was therefore chosen.
Figure 16: The brush used in these experiments shown to the left, and the grinder to the right.
3.1.2 The Roughness Testing Apparatus
The surface roughness can play a significant role during roll bonding, as curvature on
the surface focuses compression force on the asperities first, and creates shear
stresses on the oxide layer so that it cracks. A Mitotoyo SJ-201 was used to measure
the roughness of the samples, and is a rather simple testing method which gives one
final roughness value for each test.
First the sample is placed in a track on the instruments table, while a thin needle is
lowered in place on a selected area of the sample. Further the proper mode and
19 | P a g e
19
20
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
parameter is chosen on the instrument controller. For this test the Ra mode is chosen,
as it measures the distance between the highest and lowest measured point in a given
section length. Several such sections are measured and an average of all these is set as
the roughness value. The parameter is chosen for each sample in consolidation with
the NS-EN ISO 4288 standard, which specifies the suggested settings for any
roughness. This means that it is advised to find a coarse measurement of the
roughness so that the test parameters can be calibrated towards this particular
roughness.
For the typical scratch brushed surface used in this experiment, the suggested settings
where found to be Ra 2.5mm x 5 sections. Where Ra is the mode described above,
2.5mm is the length of one section, and 5 is the number of sections that is repeated in
a straight line. Each section needs to be long enough to include enough peaks and
valleys to ensure sufficient accuracy. Although, it has to be as short as possible, not to
be majorly affected by any possible slope or large scale curvature on the surface. This
would be the case if the sample is not perfectly flat. As an example; if the sample was
higher in one end than the other, or just simply bent. This is also the reason the test is
parted up in sections. The needle never stops, but the measured values are reset 4
times during the test. The difference between max and min for each section is
calculated and an average of these 5 sections, is the output value on the instruments
display.
For extra precaution this test is carried out two more times on the same sample and a
final average of the three output values is chosen as the roughness value. The value is
in the unit µm and is the average between the maximum distance between peak and
valley from each of the sections measured.
As the test is using a needle which is in physical contact with the material, there is a
chance that the needle will affect the surface of particular soft samples. This was not
investigated further.
3.1.2 The Tensile Testing Apparatus
All three methods for testing bond strength were carried out on the same type of
testing apparatus. A MTS (Modular Test System), which is a tensile testing apparatus
pulling with a vertical load, while measuring the load and the elongation, and the
option for attaching a gauge to measure elongations over a more specific area. For
these experiments the apparatus was set to measure the load applied while
maintaining a constant strain rate, until fracture.
For the tensile bond strength testing, a 100kN load element was used, while for the
shear bond strength test and the prior peel test this was changed to a 5kN load
element, which is more sensitive at the lower loads. The strain rate was subject to
variation, and the sampling rate was adjusted after this.
20 | P a g e
Steinar Lauvdal
3 Experimental
Figure 17: Two pictures showing the tensile testing apparatus, used for all three tests; peel-test, shear
bond strength test and the tensile bond strength test. This picture was taken during a peel test.
3.1.2 The Scanning Electron Microscope (SEM)
Prior to investigating the fracture surface these samples were soaked in acetone or
ethanol and went through ultrasonic cleaning for two minutes. The samples were
thoroughly dried with a hair dryer before glued to a sample holder by carbon-tape and
mounted in the SEM.
The samples for interface investigations went through the same cleaning before
entering the SEM, but prior to this some additional preparation were performed.
These samples were scratched on Si-carbide paper in steps from grit 80 to 2400,
followed by polishing down to 1 µm. A final electro polish was performed on a Struers
LectroPol-5 with parameters listed in Table 3. The samples were molded when electro
polishing was executed.
Table 3: Showing the settings for electro polishing of samples.
Electrolyte
Exposed Area
Temp. Electrolyte
Voltage
Flow Rate
Time
A2
½ cm2
-36 C
20V
II
10 sec
21 | P a g e
21
22
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
3.2 The Specimens
Figure 18: A piece of 3103 aluminum alloy prepared for rolling down to final sheet thickness.
The material selection is focused exclusively on aluminum alloys, specifically the 1200
and 3103 alloy. Here, the 1200 alloy is representing a very soft aluminum and the 3103
represents the middle of the tree. Both alloys where rolled stepwise, from blocks of
aluminum (Figure 18), down to desirable plate thicknesses giving them a certain work
hardening. This further work hardening and followed annealing created even more
diversity in the material properties prior to roll bonding. Along the stepwise rolling,
samples were picked out at the different thicknesses, so that even at different
reduction the final thickness of the roll bonded sample should be approximately the
same for all samples. In a later batch, a new approach was taken where the initial size
of the sample sheets was the same and the final thickness was left to vary.
Table 4: This table lists the chemical composition of the aluminum alloys utilized in these
experiments. Source: AluMatter.info [12].
1200
Si + Fe
Total Other
Zn
Other Elem
Ti
Mn
Cu
Al
-
3103
<= 1.0
Mn
0.9 - 1.5
<= 0.15
Fe
<= 0.7
<= 0.10
Si
<= 0.50
<= 0.05
Mg
<= 0.30
<= 0.05
Zn
<= 0.20
<= 0.05 Total Other
<= 0.15
<= 0.05
Zr+Ti
<= 0.10
<= 99.00
Cr
<= 0.10
Cu
<= 0.10
Other Elem
<= 0.05
Al
Remainder
All compositions in wt%
The stepwise progression of cold rolling the samples to the desired thickness was
logged and plotted, as shown in Figure 19. In Table A 1 in the appendix the data for this
graph is listed.
22 | P a g e
Steinar Lauvdal
3 Experimental
23
Rolling Log
25,00mm
Thickness
20,00mm
15,00mm
A1 1200
B1 3103
10,00mm
A2 1200
5,00mm
B2 3103
0,00mm
0
5
10
15
20
25
Number of Passes
Figure 19: A graph showing the stepwise rolling of the four sample batches produced.
A1, A2, B1 and B2 in Figure 19 mark the different batch number and are used as a
prefix in sample names in this study. A full explanation of how to read the sample
names is given in the chapter presenting the results.
3.2.1 Annealing
Due to the high strains of the pre-rolling of the sample sheets, one half of each batch
was put through an annealing process prior to roll bonding. The annealing
temperature and time was set to reset all work hardening strain in the material,
returning it to its pre-strained strength.
The sheets were stacked inside a furnace at room temperature and the temperature
set to 450 degrees Celsius. Over a three hour time span the temperature steadily
increased till it reached the set temperature, and were held there for 1.5 hours more.
After being 4.5 hours in the furnace, the sheets were extracted from the furnace and
cooled in air at room temperature. A graphical presentation of this annealing process
is given in Figure 20. No fan or other air-circulation instrument were used, although
the samples were stacked vertically on the side and slightly separated to increase
cooling rate. All the plates from both the 1200 and 3103 batches had geometry similar
to each other. The approximately geometry is given Table 5.
23 | P a g e
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Annealing Procedure
500
Temperature [C°]
24
400
300
200
Annealing
100
Air Cooling
0
0
1
2
3
4
5
6
Time [h]
Figure 20: Graph showing the temperature-curve for the annealed samples used in this study. The
dashed line shows the cooling in room temperature and is only an estimation.
In Figure 20 the cooling curve presented, is an estimation only, as the cooling rate was
not measured.
Table 5: The average sample size when annealed.
Length
Width
Thickness
200 mm
40 mm
1-2 mm
3.2.2 Specimen Preparation
This following section is presenting the different sample preparation methods required
for the two main testing methods utilized in this study. Both tests may use the same
preparations until finished roll bonding. The roll bonded sheets can be seen in Figure
21, however from this point on, some more sophisticated cutting tools is required to
produce the final test samples.
Figure 21: This figure is showing two ends of cold roll bonded strips as they look before they are cut
into either a shear- or tensile- test sample.
3.2.2.1 Shear Bond Strength-test Samples
The cutting of shear-test samples was carried out in cooperation with people at the
workshop. A template for the outer measurements of the shear bond strength (SBS)
sample is the first required item. This template is a thick copper plate where the shape
24 | P a g e
Steinar Lauvdal
3 Experimental
has been drilled into the surface. It allows for a small plate roll bonded plate to be
clamped as a drilling tool is machining the shape, seen in Figure 22.
Figure 22: A shear bond strength sample cut and prepared for testing.
The following step is to machine the grooves seen in both Figure 22 and Figure 23.
Before cutting these samples, the thickness of each sheet was measured and as rule of
thumb the cut was set at a depth of
(12)
where τ1 and τ is explained below with the rest of the dimensions seen in Figure 23.
This cut was made to disconnect one of the layers in the horizontal direction, better
illustrated in Figure 13. When one side was cut through, the sample was released,
flipped and fixed upside down in the exact same position, by hand. The drill was now
programmed to cut through the opposite layer of the sample, again only dislocating
this one layer in the horizontal direction. The two cuts were placed in a pre-calculated
horizontal distance from each other, so that only a specific area of the two layers
overlapped. This overlap, the distance l, was chosen by anticipation the strength of the
bond to fit in between the upper and lower testing limitation of the shear-test. The
lower being strong enough to prevent the sample from breaking during machining and
the upper being the yield strength over the cross-section of the material.
Figure 23: A machined shear-test sample ready for testing, seen from the side.
25 | P a g e
25
26
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
In Figure 23 τ is the thickness of the sample sheet, τ1 is the depth of the groove, l is the
length between the two cut grooves, lg is the width of the groove and b is the width of
the sample. The cross-section area is given by b*l.
3.2.2.2 Tensile Bond Strength-test Samples
This section describes how the sample for the new tensile bond strength test was
performed. In general they were “coins” cut from the sample sheet with a diameter of
15mm, as seen in Figure 24.
Figure 24: Disc samples like these are cut and machined from cold roll bonded sheets as the one in
Figure 21.
A lathe was utilized for the cutting of these samples, after a few attempts on stamping
them out showed to severely distort the sample. Other methods like water-jet cutting
and even laser cutting was thought of, but disregarded due to high cost and heat
concerns with the latter.
Some of the material tested in this study is extremely soft, and in the process of
finding a satisfactory cutting method, many samples were strongly affected and some
flat out ruined. Friction and adiabatic heating could potentially affect the properties
and strength of the bond and for these reasons great care was taken when
manufacturing these samples. The samples were first cut roughly to a circle-like shape.
Then the last finish was machined with much care and plenty of cooling fluids, to the
final diameter of 15mm. After the first set of samples was returned clearly too heavily
deformed by the machining, the samples in the latter sets was machined one by one
with a brass plate on each side to shield the sample. This brass plate prevented the
soft aluminum samples from being welded together as the knife slide sideways over a
full stack of them in the lathe. The new precautions ensured good samples quality
from this point on.
As the samples with large reduction had bond strength overgrowing the upper
limitation of this test-method, a groove was machined in the side of some samples in
an attempt to raise the upper strength limit of the method. The idea behind this
surgical intervention is to reduce the bonded area while leaving the full surface area of
the coins to maintain optimal binding strength to the glue. In Figure 25 there is a
sketch illustrating the position of the groove.
26 | P a g e
Steinar Lauvdal
3 Experimental
Figure 25: A sketch showing where the groove was cut to reduce the bond cross-section.
In cooperation with an employee in the workshop several tools where customized in
an effort to achieve the desired result. This was a particular challenging work, as
cutting a thin groove in the side of a disc that is no thicker then 1mm requires very
small, precise and strong tools. The same type of lathe was used to machine the
grooves. A thin knife with a thickness of 0.2 mm, fixed to a steady sliding part of the
machine was used to cut the groove into the side of the discs.
3.3 The Roll Bonding Process
The procedure of roll bonding from the selection and preparation of the material
through surface preparation and stacking to roll bonding is illustrated below in Figure
26. The surface preparations are explained in greater detail in the following section.
Figure 26: The procedure of roll bonding. [7]
3.3.1 Surface Preparation for Roll Bonding
The first step of the preparation is degreasing, this was carried out with acetone and a
paper cloth. After a thorough surface cleaning, two holes were drilled in each end of
the strips to enable a quick fixating of rivets in the stage just before roll bonding. Both
sample sheets were clamped to the worktable and scratch brushed to remove the
27 | P a g e
27
28
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
oxide layer and create the desirable roughness in the surface. Scratch brushing on an
aluminum surface creates a roughness and certain topography as seen in Figure 27 [4].
A surface such as this is considered well suited for roll bonding.
Figure 27: Scratch-brushed aluminum surface. [4]
The following procedure was to clamp the two plates together with rivets in both ends
before proceeding to the final roll bonding. This was done to keep the two plates
aligned through the mill. As the brushing was carried out by a hand-held machine and
the force and intensity thus depend upon the operator, all surface preparation was
carried out by the same person. This was done to ensure maximum consistency in this
area. Also the time from scratch brushing was initiated to the sample entered the mill,
was kept to an approximate constant. This time was for the majority of samples kept
close to 90 seconds.
28 | P a g e
Steinar Lauvdal
3 Experimental
Figure 28: The preparation table and a close-up of a newly brushed sample. The picture is taken
during preparation of a peel test sample [10].
3.4 Testing Methods
The two considered test methods; the peel test and the shear bond strength test, do
not offer the ideal circumstances for a bond strength test. The peel test being very
dependent on the thickness and strength of the bond, resulting in different tearing
angle which influence the measured strength. It also cannot be compared to other test
methods in a proper manner as the load applied is not distributed to a given area.
When performing a SBS test it is the shear strength that is tested, and not the bond
strength in a tensile direction.
For these reasons there was desire for a new testing method for evaluating the bond
strength between two roll bonded plates. In the following section the preparations and
procedures for this new test method, called the tensile bond strength test (TBST), is
presented. The last part in this section includes the execution of the SBS test as well.
3.4.1 Tensile Bond Strength Test
The preparation and machining of the disc samples were explained in section 3.2.2.2,
and this part will show the reason for these particular shapes. About 7 cm of each roll
bonded strip was spent to cut four disc samples, and the rest was saved for
comparison with another test.
29 | P a g e
29
30
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Also from the workshop, a set of 120 mm long aluminum rods with a diameter of
15mm was ordered. Each of the sample discs was glued with industrial epoxy
adhesives between two of these rods and left to cure, like shown in Figure 29. More
details on the preparation and gluing is described in the following section.
Figure 29: To the left is a stack of the aluminum rods used to test the disc samples. On the right side is
one of the samples fitted in-between two of the rods.
For the strongest bonds the glue alone was too weak and additional measures was
taken in hope of raising the limit of the test. Some of the disc samples had a up to 0.3
mm deep groove machined in from the side. Reducing the boned area but leaving the
full outer surface for gluing, as explained in section 3.2.2.2 and further discussed in
section 5.3.2.
The final TBS testing was performed at different strain rates. While mounting the
samples in the MTS, it was discovered that the upper and lower grip was not perfectly
aligned, putting a small shear stress on the samples. This shear force was however
small, and besides risking fracture while clamping the weakest sample, the small shear
stresses was neglected in regards of the measured bonding strength.
3.4.1.1 Gluing Procedure
Some sense of skepticism was hanging in the air, when the idea of an adhesive that
should be stronger than a metallic bonding. The first few attempts were made with
commercial superglue, bought at a local store. This glue, by the mane Bostik epoxy
Rapid, withheld a load of 20MPa before yielding, and can be found in Table C 1 in the
appendix. While this glue was by no chance strong enough to overgrow the bond
strength of the heavier deformed samples, the hunt for a much stronger adhesive had
started. The search ended with a product from Henkel Technologies Norway; LOCTITE
9466 A&B [18]. This is the 2-component epoxy adhesive in the LOCTITE-repertory that
reports the highest tensile strength between two metals cured at room temperature.
As for the preparation, slightly different advances were tried out. The general
approach was to first pre-clean the surfaces of both rods and the sample with acetone,
removing grease and pen marks. This was followed by gentle scratching of the surface
on Si-carbide paper. The paper was placed on a flat surface and the sample/rod was
scratched onto this, to avoid as much rounding of the edges as possible. Scratching of
the surface is a common method in adhesive bonding to increase the grip surface of
30 | P a g e
Steinar Lauvdal
3 Experimental
the adhesive. After all four surfaces had been scratched the sample and rods were
again cleaned in acetone then sprayed with LOCTITE 7063 cleaning spay and dried with
a hair-dryer for the last time. Below, four types of slightly different preparation
progression are described. Every sample tested, shown in Table C 1, is marked with
one of these types.
Type 1:
Scratched on Si-carbide paper – cleaned with acetone – dried with a hair
dryer.
Type 2:
Same as Type 1 + more thorough cleaning and scratching.
Type 3:
Cleaned in acetone – scratched on Si-carbide paper in water – cleaned in
acetone – cleaned with spray** – dried with a hair dryer.
Type 4:
Cleaned in acetone – scratched on Si-carbide paper in water – cleaned in
acetone – dried with a hair dryer.
Out of these four types, type 3 seemed to give a more steady results, with fewer
samples failing in the glue much bellow the 40MPa region. While time is of the
essence, no further investigation was put into this part once a stable method was
found.
Figure 30: To the left is a picture of the gluing kit shown and to the right a sample clamped between
two rods and a close-up of the glue used.
Due to the small quantities of glue needed per session the mixing tips that followed
the glue was not used. Instead the glue was mixed manually on a piece of paper. A thin
layer of glue was smeared onto each of the four surfaces, ensuring each surface is well
covered and small air bubbles in the glue is minimized. To make sure the disc sample is
aligned dead center between the two rods, a plate with a machined v-shaped groove
was used when assembling. Figure 31 is showing how the sample is placed between
the rods. Once the sample and rods are aligned, a clap, as seen in Figure 30, is gently
used to put a compression load on the glued sample. Any excess glue is wiped off, and
31 | P a g e
31
32
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
for the samples with a machined groove on the side, the groove has to be cleaned
thoroughly as it is easily filled with the adhesive during the process. The glued samples
are now left for curing in compression for a minimum of 24 hours. Most samples are
cured for 3 days or more before the TBS test is performed.
Figure 31: This figure shows how the disc samples are placed between the rods.
3.4.2 Shear Bond Strength Test
As earlier mentioned, a part of the roll bonded strips was saved; this part was used to
compare the TBS test to the well established SBS test and machined as described in
section 3.2.2.1.
The samples were mounted in the same tensile testing apparatus as the TBST samples,
only now with flat clamps and the 5kN load element installed. Great care was taken
when mounting the samples to a vertical position with as little angular deviation as
possible. The small offset was again observed when the samples were mounted in the
tensile machine, but still neglected as a huge concern. A strain rate of 0.2 mm/min
with the sampling rate set at 10Hz was used for these tests.
Some angel deflection was observed as the tests were carried out; these observations
are discussed further in 4.3.1 and 5.4.1.
32 | P a g e
Steinar Lauvdal
4 Results
4 Results
In this chapter the results of the mechanical testing are presented along with the
observations made during a surface analysis in a scanning electron microscope (SEM).
Also some results gathered during a previous study by Lauvdal [10] have been included
for the sake of compensation.
Out of a total 60 pairs of roll bonded aluminum sheets, only samples from 17 of these
sheets gave substantial results during testing. 31 of them where too well bonded for
the TBS test and 12 sheets were ruined in the various machining and testing steps.
The samples cut for the TBS test, but outside the test limits of the test, were too small
to be tested by both the SBS test and the peel test. The amount of excess material that
was left from the roll bonded sheets was too little to create any more shear-test
samples.
In this study the samples are named by a system that contains some information about
the sample. It is build up of 5 elements; the alloy, batch number, whether it is
annealed or not, the number in the batch and finally a test number for cases where
several samples from the same sheet is tested. A breakdown of an example is shown
below.
Example: A2NA07-T2
A2:
The “A” sais that this is an 1200 alloy, while “2” means it is from the second
1200 batch rolled.
NA:
”NA” stands for non-annealed, and the notification for annealed is an “A” in
this position. The term non-annealed have the same meaning as work hardened
in this context.
07:
This number divides the samples with different reduction within one batch.
T2:
“T2” indicates that this is the second TBS test carried out on the set of samples
that has all the prior notification in common.
33 | P a g e
33
34
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
4.1 Peel Test
This section holds some of the results achieved in a prior study by Lauvdal [10], which
served as a precursor to the current study. The data is presented here for its relevance
and direct influence on choices made for this project.
Figure 32: This figure shows how to intrepid the appellations of the peel-test samples. [10]
Figure 32 above, shows a picture of a cold roll bonded strip of aluminum and the
different sections for which the samples in that report was named is illustrated.
Table 6: This table show the results from a peel-test carried out on a Al 1200 alloy, which where work
hardened by cold rolling to a similar strength as the Al 1200 NA material used for both the tensile and
shear test in this thesis. [10]
Initial Properties
#
23BT
23AT
24BT
24AT
25BT
25AT
26BT
27BT
27AT
28BT
28AT
Layer 1
Alloy
vonMises
Al (1200)
2,08
Al (1200)
2,08
Al (1200)
2,08
Al (1200)
2,08
Al (1200)
2,08
Al (1200)
2,08
Al (1200)
2,08
Al (1200)
1,25
Al (1200)
1,25
Al (1200)
1,25
Al (1200)
1,25
Layer 2
Alloy
vonMises Brush
Al (1200)
2,08
SB 1
Al (1200)
2,08
SB 1
Al (1200)
2,08
SB 1
Al (1200)
2,08
SB 1
Al (1200)
2,08
SB 1
Al (1200)
2,08
SB 1
Al (1200)
2,08
SB 1
Al (1200)
1,25
SB 1
Al (1200)
1,25
SB 1
Al (1200)
1,25
SB 1
Al (1200)
1,25
SB 1
End Properties
Brush Speed
3800 rpm
3800 rpm
3800 rpm
3800 rpm
3800 rpm
3800 rpm
3800 rpm
3800 rpm
3800 rpm
3800 rpm
3800 rpm
Reduction
49,0 %
49,0 %
51,5 %
51,5 %
55,1 %
55,1 %
71,9 %
49,8 %
49,8 %
49,5 %
49,0 %
PT
Speed Load
3
16,8 N
3
27,7 N
5
24,6 N
5
19,9 N
3
70,2 N
3
55,3 N
3
254,3 N
3
15,1 N
3
18,4 N
3
13,3 N
3
17,3 N
Bonding
KB
KB
KB+
Bonded
Bonded
failed
weak KB
weak KB
weak KB
weak KB
In Table 6 the results from a peel-test carried out on a Al 1200 alloy, which where work
hardened by cold rolling to a similar strength as the Al 1200 NA material used for both
the TBS and SBS test in this thesis. The brush type SB 1, is the same steel-brush as used
for scratch brushing the entire sample collection in this current study as well.
In Figure 33 the peak load during peel test from Table 6 is plotted against the
reduction. Sample 26BT is removed as it fractured during peeling and is most likely to
show the strength of the material itself rather than the bonding strength. The plots are
also separated in two rolling speeds, where the velocity for speed 5 is estimated to a
value of 150 mm/s, from the measured valued described in section 3.1.1.
34 | P a g e
Steinar Lauvdal
4 Results
Peel Test - Al 1200
80,0 N
70,0 N
Load [N]
60,0 N
50,0 N
40,0 N
Speed 3 (26.62 mm/s)
30,0 N
Speed 5 (*150mm/s)
20,0 N
10,0 N
0,0 N
40,0 %
45,0 %
50,0 %
55,0 %
60,0 %
Deformation [%]
Figure 33: Peel test plots of pre work hardened Al 1200 strips; roll bonded at two different rolling
speeds.
Despite a shortage on tested samples at speed 5, there is reasonable indication on
lower bond strength at higher rolling speeds.
4.2 Tensile Bond Strength Test (TBST)
Due to a maximum strength limit on this test, samples with a bond strength above
40MPa was not possible to measure as all these samples failed in the glued part before
reaching the strength of the bond between the sheets.
Figure 34: This picture is taken while the sample is in the tensile machine. The bonds have just
successfully parted, and the misalignment is clearly visible.
A slight misalignment in a horizontal direction, as shown in Figure 34, caused the
weakest samples to break when being clamped into the machine. To reduce this shear
force, the upper grip was loosened so that it was able to automatically align itself with
the lower grip once tensile force was applied. Any shear forces due to this
misalignment, after testing was initiated, was neglected. However the weakest
35 | P a g e
35
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
samples were, as mentioned, ruined due to a short sudden shear force in the moment
when the second clamp was fastened.
As a precaution to the misalignment, the rods which the samples was glued to, was
clamped on the outer edge with as little grip as necessary. This was to increase the
distance between the grips and reduce the influence of any possible misalignment
during testing. A risk related to this was that the rod would not be fixed in a perfect
vertical position. Any such deviations were kept an eye out for and corrected if
necessary.
Some of the earliest TBST samples were ruined during machining. These samples are
the non-annealed 1200 alloy samples found in Table B 2 in appendix B, sample A1NA01
trough A1NA07.
TBST - All Tested
TBS [MPa]
36
50 MPa
45 MPa
40 MPa
35 MPa
30 MPa
25 MPa
20 MPa
15 MPa
10 MPa
5 MPa
0 MPa
Al 1200NA (0.2)
Al 1200NA (10)
Al 1200A (0.2)
Al 3103NA (0.2)
Al 3103NA (10)
Al 3103A (0.2)
0%
20 %
40 %
60 %
80 %
Glue Limit
Groove-samples
Deformation [%]
Figure 35: Overview of all samples tested using the TBST.
Figure 35 show the gathered data plots for all alloys, annealed and non-annealed, to
give the general trend in increasing bond strength. When plotted together like this the
deviations between the different alloys and strain rates are put in perspective. This is
discussed further in section 5.3.
The samples marked as kissing bond do not show actual measured bond strengths, but
a strength estimated by the operator that executed the test. The samples in question
fractured while being mounted, indicating they were only barely bonded. These values
are considered accurate within reasonable margins by the operator/author.
The term kissing bond is a trivial name on the initial phase of bonding between two roll
bonded plates. It describes a barely bonded state where the sheets is easily separated
and consists of more mechanical binding mechanisms then of metallic bonds.
36 | P a g e
Steinar Lauvdal
4 Results
TBST - Al 1200 A
60 MPa
TBS [MPa]
50 MPa
40 MPa
30 MPa
0,2 mm/min
20 MPa
Kissing Bond
10 MPa
Glue Limit
0 MPa
0%
20 %
40 %
60 %
80 %
Deformation [%]
Figure 36: TBST of the annealed AA1200 samples. The lines in this graph are indicating the
approximate path where the strength curve is expected to go with increased reduction.
Both the annealed sample collections were subject to a lot of ruined samples during
the initial steps of sample preparations. A large portion of the low reduction samples
were ruined in both machining of the shear samples and tensile samples. Hence the
few data plots for this alloy, shown in Figure 36 and Figure 37.
In Figure 36 an estimated trajectory for the bond strength at increasing reduction,
marked by the area between the two dashed lines. The base for this assumption is the
measured data at low reductions and the minimum strength at higher reductions,
made possible by the glue limit.
TBST - Al 3103 A
30 MPa
TBS [MPa]
25 MPa
20 MPa
15 MPa
0,2 mm/min
10 MPa
Glue Limit
Kissing Bond
5 MPa
0 MPa
0%
10 %
20 %
30 %
40 %
50 %
Deformation [%]
Figure 37: TBST of the annealed AA3103 samples suffered big losses in the machining phase and
resulted in very limited results.
37 | P a g e
37
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Some samples were made with a thin groove machined in depth of 1-3 mm in from the
side, reducing the bonding area between the two sheets. The tested strength on these
samples showed a negative deviation to the expected strength. Of all the samples with
a groove machined in the side only three survived long enough to be tested. These
samples can be seen in Figure 35 where they are clearly below the trend of all other
samples.
4.2.1 Strain Rate
A few sample pairs were tested at different strain rates and the result from these
experiments can be seen in Figure 38 and Figure 39. In first figure a trajectory is made
for the combined path of assumed bond strength at increasing reduction. Yet again the
glue limit is guiding the path.
TBST - Al 3103 NA
60 MPa
50 MPa
TBS [MPa]
38
40 MPa
30 MPa
0,2 mm/min
20 MPa
10 mm/min
10 MPa
Glue Limit
0 MPa
0%
10 %
20 %
30 %
40 %
50 %
60 %
Deformation [%]
Figure 38: In this figure the gathered path of the assumed strength curve for both strain rates is
sketched in for the 3103 samples in the TBST.
Similar results was found in the non-annealed 1200 alloy, shown in Figure 39, but here
an individual trend line is added to illustrate the shift in bonding strength between a
stain rate of 10 mm/s and 0.2 mm/s. The few plots, do not allow for a high accuracy on
the trend lines, but the elevated strength found at 10 mm/s in both the 1200 and 3103
alloy seems clear.
Some sample, for instance the sample marked in Figure 39, under Glue limit at 40%
reduction, is tested at a stain rate of 10 mm/min. All samples that failed in the glue are
marked by one mutual symbol, regardless of the strain rate of which it was tested. The
strain rate does however seem to have an influence on the measured value in both the
adhesive bond and the metallic bind, and is discussed more in section 5.5.3.
38 | P a g e
Steinar Lauvdal
4 Results
TBST - Al 1200 NA
60 MPa
TBS [MPa]
50 MPa
40 MPa
30 MPa
0,2 mm/min
20 MPa
10 mm/min
10 MPa
Glue Limit
0 MPa
0%
10 %
20 %
30 %
40 %
50 %
60 %
70 %
Deformation [%]
Figure 39: The graph is showing the plots of the Al 1200 samples in the TBST, with a plotted trend line
for the samples tested at the two different strain rates.
4.2.2 Gluing Results
The fracture point for the samples that yielded in the glue, reaching from bellow
20MPa up to 40MPa. When a better preparation methods for gluing was established
the deviation from what seemed to be the maximum achievable strength was reduced
to barely fall below 35MPa. This is based on the waste majority of samples that where
tested at a stain rate of 0.2 mm/min. For the samples tested at 10 mm/min the
maximum bond strength of the glue was measured up to 48.02MPa.
All glue results can be found in Table C 1 and plotted in Figure 35 through Figure 39.
Except for in a few occasions, all the samples was glued to aluminum rods. The
aluminum rods were chosen to ensure that the bonding between the glue and the rod
did not have any disadvantage in bond strength when compared to the sample-glue
bond strength. Later in the study however, as the datasheet for the new glue was
inspected; stronger bonding properties was reported on steel surfaces. As a result to
these findings some samples were tested glued to a steel rod. These samples did not
distinguish themselves from the aluminum samples in any remarkable way. A further
discussion on this topic is found in section 5.3.5.
Figure 40: The fracture surface of samples which yielded in the glue. From the left: A1A01-T1, A1A02T1, A1A04-T1 and A1A06-T1.
39 | P a g e
39
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
In Figure 40 is four pictures taken of the fracture surface of four samples where the
glued bond was the first to yield. The surfaces of these samples and rods show a
variation cover of epoxy residue. Some seem uneven, and some have a clean outer rim
that could indicate no bonding in the area. This would reduce the strength directly by
the fraction of the non-covered area. Attempts were made to improve and correct this
as the experiments progressed by higher and lower compression during curing and by
scratching with different grit size. The amount of other necessary variable made it
impossible to confirm any improvement with the changes made and any further
experimentation were not pursued. The grit size and preparation type for these
samples can be found in Table C 1 in appendix C.
4.3 Shear Bond Strength Test (SBST)
As the “state of the art” method of testing bonding strength, this method was included
as a comparison to the new TBST.
In the initial phase of machining the SBST samples, a few was ruined. Adjustments
were made as a result of the losses, which lead to a much safer production. The
method was now able to produce samples for testing that was just barely bonded.
SBS - All Tested Samples
45 MPa
40 MPa
Bond Strength [MPa]
40
35 MPa
30 MPa
25 MPa
Al 3103 NA
20 MPa
Al 3103 A
15 MPa
Al 1200 NA
10 MPa
Al 1200 A
5 MPa
0 MPa
0%
10 %
20 %
30 %
40 %
50 %
60 %
Deformation [%]
Figure 41: The graph shows the plot of all successful SBST samples.
Figure 41 show the full plot of all the successful SBS tests performed. Unfortunately in
the majority of cases, either the SBST sample or the TBST sample from each pair was in
some way ruined in an almost perfect overlapping pattern. Only a handful of the
paired up comparison tests made it through all machining and testing phases. These
40 | P a g e
Steinar Lauvdal
4 Results
results are compared in Figure 42. The TBST samples are scattered on both side of the
SBST strength curve. This is discussed further in section 5.5.
Bond Strength [MPa]
TBST vs SBST
45 MPa
40 MPa
35 MPa
30 MPa
25 MPa
20 MPa
15 MPa
10 MPa
5 MPa
0 MPa
0.2 mm/min
10 mm/min
SBST
0%
10 %
20 %
30 %
40 %
50 %
60 %
Deformation [%]
Figure 42: The graph is showing the compared results from an SBS test and a TBS test, where identical
sample pairs have been tested against each other.
Table 7: This table is containing the data which Figure 42 is plotted from.
TBST
Reduction
0.2 mm/min
38,80 %
38,29 MPa
48,50 %
31,60 %
12,70 MPa
SBST
10 mm/min
29,54 MPa
30,85 MPa
39,12 MPa
23,17 MPa
21,92 MPa
4.3.1 Angular Deflection
While testing the SBST samples an observation was made. The angle at which the
sample was split changed during the tensile loading. When mounted and before any
load was applied all the samples were of course standing horizontally, and while the
initial load was applied the stresses was 100% shear. As the load increased the angle
increased as well, as seen in Figure 43 of a sample after fracture.
41 | P a g e
41
42
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Figure 43: Illustrating how the angle was measured on SBST samples.
A set of macro pictures was taken of all the SBST samples and the angle was measured
for both parts, as shown in Figure 43. The average value of the two measured angles
was used as the angle for that sample, and everything is discussed in more detail in
section 5.4.1.
Table 8: This table lists the correlations between the angle Ɵ, the overlap length l, the reduction %
and the bond SBS.
Sample
B2NA05
B2NA03
B2A06
B2A05
B2A04
A2NA04
A2NA03
A2NA02
A2A03
Ө(snitt)
3,1:
1,3:
14,0:
4,0:
18,4:
19,2:
6,5:
3,9:
4,5:
l
2,50mm
2,50mm
2,50mm
2,50mm
2,50mm
2,50mm
2,50mm
5,00mm
10,00mm
%
41,8 %
31,6 %
45,4 %
39,8 %
34,2 %
48,5 %
38,8 %
33,0 %
27,2 %
SBS
29,00 MPa
21,92 MPa
27,88 MPa
23,42 MPa
25,64 MPa
39,12 MPa
29,54 MPa
15,98 MPa
8,36 MPa
To filter out static two of the samples which had a different overlap-length was
discarded, when an attempt on connecting the angle deflection to the bond strength.
These two samples are marked grey in Table 8 and the comparison to the degree of
reduction is plotted in Figure 44.
42 | P a g e
Steinar Lauvdal
4 Results
Angle vs SBS
25⁰
Angle [Ɵ]
20⁰
15⁰
10⁰
5⁰
0⁰
0 MPa
5 MPa 10 MPa 15 MPa 20 MPa 25 MPa 30 MPa 35 MPa 40 MPa 45 MPa
SBS [MPa]
Figure 44: Shows a plot of samples in Table 8 comparing the angle with the SBS.
As the bond strength is directly influenced by the reduction, the angle can be
compared to the reduction as well, seen in Figure 45.
Angle vs Reduction
25⁰
Angle [Ɵ]
20⁰
15⁰
10⁰
5⁰
0⁰
0%
10 %
20 %
30 %
40 %
50 %
60 %
Deformation [%]
Figure 45: Shows a plot of samples in Table 8, comparing the angle directly to the reduction.
4.4 Fracture Surface Investigation
The fracture surface was investigated in a scanning electron microscope (SEM). The
investigation focused on finding abnormalities, crack initiations, bonding patterns etc.
In addition to the pictures presented in this section, a larger selection of SEM-pictures
of the fracture surfaces is listed in the appendix under section E.
43 | P a g e
43
44
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
4.4.1 Bond Types
The primary thing to look for is signs of bonding. Figure 46 shows the characteristics
signs of a ductile fracture. Further details are also seen in Figure 55 in section 5.6.1
where the topic is discussed.
Figure 46: Typical sign of a ductile fracture seen on the fracture surface of sample B2A04-T1 A.
4.4.2 Crack Direction
At higher reduction it is clear, from observations in SEM images, that bonding mainly
occurs in lines stretching normal to the rolling direction. Figure 47 shows the surface of
a shear test sample with these categorist lines. Due to the shape of the sample the
rolling direction, is known. The rolling direction is indicated by arrows in Figure 47.
These characteristic lines were first observed on the fracture surface of a TBST sample,
but with the information on rolling direction lost due to the circular shape of the
samples, the mentioned hypothesis could not be confirmed at that point. More
pictures showing these bonding lines can be seen in the additional pictures found in
the appendix section D and E.
44 | P a g e
Steinar Lauvdal
4 Results
Figure 47: The fracture surface of sample A2NA02 seen from above. Al 1200 Non-annealed, 33%
reduction. Showing the bond-lines stretching normal to the rolling direction at 20x and 100x
magnification.
45 | P a g e
45
46
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
4.5 Surface Roughness
A random sample was chosen to determine the surface roughness of plate strips prior
to roll bonding. As the time from the strips were brushed to they are roll bonded is
crucial to maintain short, roughness testing of each sheet is not practical possible. The
chosen sample is scratch brushed under the same circumstances as all other scratch
brushing performed in these experiments.
The result from the roughness test on the scratch brushed sample was as follows:
Ra 2.5x5: 2.43 µm
4.6 The Effect of Rolling Speed
The most interesting result, seen in Table 9, of the effect of the rolling speed on the
bonding strength was found between the sample sets 21BT/AT and 22BT/AT, which
were rolled at the speeds 1.5 and 5 representatively. As the speed scale is exponential
increasing makes this a pronounced speed difference. Both sample 22BT and 22AT
failed to be peel-tested due to the high bond strength and thin sheets. However by
comparison one could tell these samples contained a far better bond then the slower
rolled samples. The samples rolled at higher speeds gained a much higher temperature
during roll bonding due to adiabatic heating. The temperature in the material during or
after cold roll bonding was not measured by any instrument, but the heat differences
were noticed by the operator.
However the results found in another section of this same study shows the general
accepted idea that higher bonding speed gives weaker bonding, as seen in Figure 33.
These observations are discussed further in section 5.7.
Table 9: Results from rolling speed test. [10]
#
13BT
13AT
14BT
14AT
21BT
21AT
22BT
22AT
Initial Properties
Layer 1
Layer 2
Alloy
vMises
Alloy
vMises Brush
Al (1200) 3,03 Al (1200) 3,03
SB 2
Al (1200) 3,03 Al (1200) 3,03
SB 2
Al (1200) 3,03 Al (1200) 3,03
SB 2
Al (1200) 3,03 Al (1200) 3,03
SB 2
Al (1200) 3,03 Al (1200) 3,03
SB 2
Al (1200) 3,03 Al (1200) 3,03
SB 2
Al (1200) 3,03 Al (1200) 3,03
SB 2
Al (1200) 3,03 Al (1200) 3,03
SB 2
46 | P a g e
Brush Speed
3800 rpm
3800 rpm
3800 rpm
3800 rpm
3800 rpm
3800 rpm
3800 rpm
3800 rpm
Def. %
61,6 %
59,3 %
59,3 %
59,3 %
58,1 %
58,1 %
58,1 %
58,1 %
Speed
3
3
2
2
1,5
1,5
5
5
End Properties
Peel-Test
Load
Bonding
6,2 N
weak KB
weak KB
weak KB
weak KB
16,5 N
KB
33,5 N
KB
too strong
too strong
Steinar Lauvdal
4 Results
4.7 Bond Interface
The interfaces of three samples were investigated in a scanning electron microscope
(SEM). All the samples were first molded in an epoxy to ease the initial grinding; first
on Si-carbide paper (grit 80 to 2400), followed by polishing down to 1µm. The in the
end, two of the samples was finished off with electro polishing, as described in section
3.1.2. Just before the samples was mounted on a sample holder for observation in the
SEM, the sample was broken out of its mold and cleaned for two minutes in a bath of
acetone by ultrasonic vibrations.
Figure 48: Sample A2NA02 observing the bond interface from the side. From the left a 800x and 1000x
magnification image of the interface of roll bonded aluminum. The electro polishing for this sample
was not fully satisfactory, but the interface is yet clearly visible.
The electro polishing was not fully satisfactory, but still the samples were passed on to
the SEM for investigation. Despite the preparation was far from perfect, it was
sufficient to make out what was expected. Figure 48 show a magnified SEM image of
what is the fractured oxide layer between the two roll bonded sheets. More images of
this interface are found in appendix F.
47 | P a g e
47
48
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Figure 49: This figure shows a SEM image of a sample that was stopped mid process in roll bonding.
The interface seen from the side.
Figure 49 shows an image at 200x magnifications of a sample that was stopped mid
process during roll bonding. The sample was cut out, slit in the rolling direction and
prepared in the same manner as the other interface-investigated samples, except for
the electro polishing. In the top right of Figure 49 the transition between bonded and
not-bonded is visible at the end of the crack.
On the second disc shaped sample no interface was found and the surface was not
photographed.
48 | P a g e
Steinar Lauvdal
5 Discussion
5 Discussion
In this chapter the results found is discussed.
5.1 Material Selection
The material selection was composed of a 1200 and a 3103 aluminum alloy, which
were work hardened and annealed to different strains.
From Figure 35 a slightly increased strength can be observed in 1200NA from the
1200A, and the same can be seen for the 3103NA compared to the 3103A. When
comparing the 1200NA to the 3103NA, no notable difference in bond strength can be
found. The same applies for the comparing of 1200A and 3103A.
Even though the strain hardening has an effect on the bond strength, it is very small.
The property differences between the 1200 and 3103 alloy seems to be too small to
give any fluctuations in the bond strength. In comparison the strain rate at which the
samples were tested, has a much higher influence on the measured strength, this is
discussed further in section 5.3.4.
5.2 Surface Preparation
When dealing with roll bonding the most important feature for success is a proper
surface preparation. If nothing is done to the surface of an aluminum sheet, it will be
greasy, have a very thick hard oxide layer and the topography of the surface will not be
ideal for a good bonding. All these factors will prevent all the bonding mechanisms and
a bond might be practically impossible. This section is discussing the effect of these
parameters in more detail.
5.2.1 Degreasing
In a former study on roll bonding by Lauvdal [10], degreasing with ethanol was tested.
The high critical deformation threshold (CDT) observed in that study raised concerns.
One of the reasons for this was believed to be the use of ethanol instead of acetone
for degreasing. Ethanol being less viscous and a poorer solvent then acetone could
support this assumption. In this study acetone was utilized and a comparison to the
prior study was attempted.
The Al 1200 non-annealed samples from each experiment were chosen as they are of
the same material composition and have experienced approximately the same work
hardening. Although the peel test does not give any good comparison of the bond
strength, it does give a good indication on at which reduction bonding begins. At the
peel test carried out with degreasing with ethanol the lowest measured bonding was
at 49.0% reduction and a peel-test strength of 27.7N, as found in Table 6. Based on the
49 | P a g e
49
50
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
trend line in Figure 33 it is reasonable to assume that bonding in this case did not start
much lower then at 45% reduction. When comparing this to the lowest tested TBST
sample at 29.6% reduction, which were tested to a TBS of 6.63MPa (at 0.2 mm/min)
and withstanding 13.02MPa in the SBS test. The SBS test of this sample ended with a
fracture in the material itself while the bond held; hence the real SBS is not shown. If a
similar estimation was to be made here, based on the trend line found in Figure 39, it
would indicate the bonding started around 25% reduction.
Before drawing any conclusion there should be mentioned that the degreasing agent is
one of two factors which could cause this dramatic difference. The peel-tested samples
were exposed for oxygen in the environment for up to 5 minutes before being roll
bonded. This is more than three times longer than The TBST and SBST samples, and
allows for formation of a thicker oxide layer. The ideal thickness of an oxide layer on
aluminum for roll bonding is not known by the author, but is discussed in section 5.2.4.
In section 5.2.4 a comparison to 5 other experimental studies have been carried out
with regard on the thickness of the oxide layer. At two of these studies the exposure
time for oxide growth was similar to the one discussed here, 5 minutes, except the
elevated heat. These two studies report bonding at hence 24% and 47% reduction,
where one strongly supports the effect of acetone vs. ethanol, while the other shows
the same result with acetone. The comparison is inconclusive.
The reduction at were bonding starts going from 45% to 25% reduction is a huge
improvement, and from what have been discussed above two out of three comparing
studies support that acetone is a far better degreasing agent the ethanol, as far as at
which reduction bonding begins.
5.2.2 Scratch Brushing
In a study by Zhang et. al. [17] it was found that semibright- and chemical Ni plating
gave bonding at a lower reduction then scratch brushing. The publication concludes
that Ni plating is the optimum preparations method for al-al cold roll bonding. This
conclusion might be true in some or most cases. However, the thoughts of the author
are that the term “better” or “optimum” is dependent on the desired properties for
the use in mind. In Figure 50, the scratch brushing curve has a steep climb in strength
and passes Ni plating in shear strength already at around 42% reduction. In the case of
ACRB, a reduction of 50% may be desirable as one can maintain the total thickness of
the plate throughout the process. This can have practical industrial advantages such as
the roll gap can be kept constant, saving adjusting time, simpler and cheaper mills,
or/and fewer rolls. For such situations the higher strength from scratch brushing might
be desirable.
50 | P a g e
Steinar Lauvdal
5 Discussion
Figure 50: Graph showing the shear strength of a cold roll bonded aluminum strips for increasing
reduction. AMAM is for semibright matt Ni plating (starts at 0.21), ACAC is for chemical Ni plating
(starts at 0.26) and ABAB is for scratch-brushing (starts at 0.33). The remaining is of less interest for
this report. [17]
In a study made 5 years earlier, also by Zhang et. al. [11] the results is basically
reversed. The scratch brushed samples now shows bonding almost as low as 20%
reduction, much like what was found in this current study with scratch brushing. The
chemical NI plating passes now the scratch brushed samples in bond strength
somewhere above 40% reduction, as seen in Figure 51. The reason for this
contradictory result was not found.
Figure 51: Graph showing the bond strength of a cold roll bonded aluminum strips for increasing
reduction. ACAC is for chemical Ni plating (starts at 0.39), ABAB is for scratch brushing (starts at 0.23)
and AMAM is for electrochemical matt nickel plating (starts at 0.40). The last curve is of less interest
for this report. [11]
5.2.3 Brush Speed and Force
One parameter in question when searching to achieve strong bonds at cold roll
bonding, is the local hardening of the surface while scratch brushing the plate or sheet
surface. In addition to remove the oxide layer, brushing with a steel brush also
deforms and hardens the surface. This straining of the surface makes it harder and less
51 | P a g e
51
52
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
ductile, and under deformation it will fracture more easily and have a desirable effect
on the bonding strength.
In this current study both brushing force and brushing speed was kept at a constant, to
allow more time to be focused on the other parameters involved. The effect of these
parameters is discussing the results found in a study by Lauvdal [10].
The results from those experiments showed no clear indication that variation in
brushing speed or the force had any impact on the bonding strength, beyond a
minimum amount. This minimum amount required is not easily quantified or
explained, but found rather quickly when performing the task. It should leave a distinct
dim finish with a clear shift in reflected light on the surface. When the scratch brushing
was increased slightly more a new pattern could be observed if looking closely. This
pattern could resemble the skin of an orange, only crumpled. Although the surface was
not observed in a light microscope or SEM, this pattern could well be equivalent to the
one in Figure 27.
No attempt was performed in discovering a possible limit where even non-scratch
brushed sheets of aluminum would bond. However, later in that study a non-scratch
brushed area was used to prevent bonding in a small area on the peel-test sample.
These samples were rolled up to over 80% reductions without any observations of
bonding in this non-scratch brushed area. These results were deemed sufficient to
decide a down-prioritizing of these parameters in this current study.
5.2.4 Effect of Oxide Layer
In this experiment the exposure time allowing oxide growth was kept to 90s and SEM
analysis of the interface showed the oxide layer to be 5µm thick. To refresh the key
preparation factors; acetone was used for degreasing, the samples were scratch
brushed and no heating of drying period was performed. Across all the samples
bonding was reported as low as at 22.3% thickness reduction, although a very weak
bond. When these results was compared to a prior study where the degreasing was
carried out with ethanol, found in section 5.2.1, there were indications that the
degreasing agent could be the main cause for the huge improvement. This section is
looking closer into whether or not the oxide layer thickness has any effect on when
bonding starts.
The effect of the oxide layer was compared to five other studies where all included
degreasing in acetone and scratch brushing. Two of the studies dried the sample in a
furnace for 5 minutes while the other three held the exposure time for oxide growth
low and at room temperature.
An experiment carried out by M.Z. Quadir et. al. [6] dried the sheets in an air
circulating furnace at 300°C for 5 minutes in addition to the common preparation
mentioned above. The peel test carried out proved bonding started at around 47%
52 | P a g e
Steinar Lauvdal
5 Discussion
reduction. When drying an aluminum sheet in an air circulating furnace at high
temperature, the oxide growth is accelerated. It can safely be said that there was a
very thick oxide layer on the surface of these sheet prior to roll bonding. From the
given data it is natural to assume the thick layer of oxide might have prevented
bonding at low reductions. The samples were also roll bonded at 300°C.
In another experiment by W. Zhang et. al. [11] cold roll bonding was carried out after
the sheets had gone through the common preparations mentioned and dried in a
furnace at 150°C for 5 minutes. Although these where dried at a lower temperature
the time allows a considerably amount of oxide to form. The odd part with this test is
that bonding is reported as low as at 24% reduction. The bond strength was tested by
a type of tensile testing method unfamiliar to the author. This result could indicate
that the thickness of the oxide layer has less or no influence on at which reduction
bonding begins.
N. Bay et. al. [21] present yet another study where the aluminum sheets where
cleaned in acetone, but this time cross shear roll bonded immediately after scratch
brushing. It should be mentioned that cross shear roll bonding (CSRB) is a method
where upper and lower roll is rotating at different speed, or have different diameter so
that the velocity on the roll surfaces is different from each other. This method is
commonly used bonding two metals with different hardness. It has also showed
improved bond strength by the use of this method.
The short exposure time should leave a rather thin oxide layer, and with bonding
proven at 30% and probably considerably lower when considering the bond strength.
This result will again be in favor of the theory of thinner oxide layer giving bonding
earlier. Of course the use of CSRB is something to take into consideration; in addition
to the fact that the samples were annealed post roll bonding, which has a proven
positive effect on the bond strength [23].
In a paper by R. Jamaati and M.R. Toroghinejad [22] preparations were carried out
with degreasing with acetone, scratch brushing, then cold roll bonded after 120s. A
peel test revealed bonding as low as 30% reduction.
One year later R. Jamaati and M.R. Toroghinejad [23] published a similar study where
the preparations were the same, and found bonding at around 33%, just slightly
higher.
The last two results show bonding at a decent level of reduction. Had it not been for
the fact that bonding where found at around 23% reduction, in [11] where the oxide is
assumed to be much thicker than in the latter examples, the conclusion would have
been easier to draw. Acknowledging parameters outside the ones that have been
taken in account which could be of influence to the presented results, a trend will
judge the conclusion here.
53 | P a g e
53
54
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
All three of the experiments with assumed thin oxide layer, four, if the one in this
current study is added, show bonding at a relatively low reduction, from 30% and
below. On the other scale, the two experiments are divided. One of them is showing
that bonding do not occur until a very high reduction, 47%, and another showing
bonding alongside the best “thin oxide layer” experiments, at 24% reduction. One
could keep in mind the results of the peel test carried out by the author in a prior
study [10], where bonding where found at above 45% with 5 min exposure time before
cold roll bonding. This study used ethanol to degrease the samples however.
The trend however, the deviation in results, indicates that bonding is likely to occur at
a lower reduction when the oxide layer is kept thinner.
5.3 Tensile Bond Strength Test (TBST)
The idea of this test was to be able to measure the tensile bond strength directly, and
in a way that the results easily can be compared cross studies. With little to no
restrictions to the material properties or geometry or such parameters that is most
likely to vary frequently from one test to the next. In addition the measured value will
be directly comparable to the yield strength of the material.
Another advantage with this method is that it required very little material to perform
the test. All that is needed is a 15 mm (or less) circular disc from the bonded area.
5.3.1 The Machining of the Disc Samples
Since the machining of the coins were done by a workshop, and for most of the time
without the authors supervision, the understanding of what forces these sample was
exposed to, is not fully documented. The process was supervised on the machining of
some of the coins and at that stage it seemed ok. A compression force, which could
not be measured, was applied by a screw-clamp tightened by the operator to hold the
samples in place in the lathe during the machining. This compression force was one of
the concerns in this process. If the compression was high enough it could affect the
bond strength, which by obvious reason is very bad for any test sample.
When the first batch of samples were returned from machining this concern became
reality, as each and all samples were severely deformed and had deep circular
scratches in the surface. The explanation from the workshop was that the material was
too soft and the samples had been welded together during the machining. To separate
the disc the operator had to use clamping tools and twist the disc to part them. Since
the operator had stacked a large amount of samples in the lathe at the same time, to
save time, the compression force needed to hold all the samples in place also
increased. This high compression force combined with the adiabatic heating from the
knife cutting, and low yield strength of the soft aluminum was a recipe for disaster.
Needless to say, these samples were trashed.
54 | P a g e
Steinar Lauvdal
5 Discussion
As mentioned in section 3.2.2.2 a solution was found for this problem; greater care
was taken by machining the softest samples one by one and adding a brass plate on
each side of each sample to prevent friction welding between the soft aluminum as the
knife cut through. Even with this method there were occasionally found samples which
were visibly affected. The most common flaw was a small crack opening in the bond
interface, visible the naked eye. There is a possibility that the opening was left behind
by a poorly bonded area of the roll bonded sheet. All the material sent for machining
were carefully chosen from areas of the sheet that should be within the proper
bonded region. However, the cause for these cracks cannot be determined.
One theory and concern to how the cracks mentioned above might have occurred, is
from the knife cutting the samples. This is a knife that is fixed in a holder that can be
moved in 3 axes. The vertical z-axis is fixed at constant height prior to the machining,
while the knife is moved in the y-axis to cut the specific diameter of the disc and the xaxis is moving the knife sideways cutting through the material. When this knife is
cutting through the samples sideways, it is possible that metal shavings or the blade
itself is jammed in between the bonded layers for a short period of time, ripping a
small crack in the interface. This may or may not be visible, but could serve as a crack
initiation during testing, and what is intended as a 100% tensile test, might now
include peel-effects directly influencing the measured strength. No high resolution
investigation as performed on the sample s interface prior to the tests, and could in
hindsight be an interesting factor to investigate in a further study.
Another concern related to the knives path though the material is the friction heating,
which is already mentioned in another relation. If the cutting is done too quick, and
poorly cooled, in such soft material, there is a chance of welding the two layers
together along the outer rim of disc. This would affect tested strength value in a
positive direction, showing a larger strength. For this reason the last part, reducing the
disc diameter down to 15.00mm was performed with a constant feed of cooling fluid.
From observation in the SEM; no sign of weld zones in the outer edge of the sample
was identified, and the concern laid to rest.
The third concern is the cooling fluid used. A possibility of cooling fluid enter between
the bonded layers is clearly there. With poorly bonded plates the risk of contamination
is even greater. The concern for the knife to create initiation cracks is only contributing
to this concern. However, effect on the bond, should cooling fluid “diffuse” in between
the layers, is not fully know. Usually these cooling fluids are a water based solution
with synthetic oils diluted to 3-10%. Any such effect is also expected to be discovered
at a SEM investigation.
55 | P a g e
55
56
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
5.3.2 The Grooves
Cutting grooves in the side of the sample to reduce the bonded area while maintaining
the full glued area, as illustrated in Figure 25, should in theory be a simple way of
raising the limit of the test method. When put into practice it was not that easy.
Figure 52: This figure is showing a picture of a sample which split in the metallic bonding during
testing. Prior to testing this sample had a groove machined in its side.
Irrespective of how careful the machining was carried out, the results of the tensile
test had clear indications that the bond of the sample had been damaged, as seen in
Figure 35. This is most likely due to an initiation of a small crack that allowed for
peeling of the surface under testing.
The current method for machining the groove in the disc samples prove to have a high
risk of affecting the bond in such a way that the following tensile testing would not
reveal the true strength. In fact none of the tested samples gave a strength value that
would satisfy the assumed bonding strength. Two samples even failed when fixating
the samples in the tensile testing machine.
Even though this modification turned out to be a failure in this study, the theory is
simple and only a more reliable way for performing the operation is needed. This is
mentioned in chapter for further work.
56 | P a g e
Steinar Lauvdal
5 Discussion
5.3.3 A General Overview
Al 1200
120,00
100,00
TBS [MPa]
80,00
Result Based
60,00
Estimated - A
40,00
Estimated - NA
Glue Limit
20,00
0,00
0,0 %
-20,00
20,0 %
40,0 %
60,0 %
80,0 %
100,0 %
Reduction [%]
Figure 53: Al 1200. Estimated curve for TBS as a function of reduction.
These two figures (Figure 53 and Figure 54) is meant to give a general impression of
how the bond strength of the two alloys, annealed and non-annealed, is affected by
the reduction. In addition the current limitations with the tensile test are indicated by
the “Glue Limit”, at which above, bonding strength cannot be measured. The green
and blue blots are measured values for which the curves “Glue Limit” and “Result
Based” are based upon.
Al 3103
140,00
120,00
TBS [MPa]
100,00
80,00
Result Based
60,00
Estimated - A
40,00
Estimated - NA
20,00
Glue Limit
0,00
0,0 %
20,0 %
40,0 %
60,0 %
80,0 %
100,0 %
Reduction [%]
Figure 54: Al 3103. Estimated curve for TBS as a function of reduction..
57 | P a g e
57
58
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Figure 53 and Figure 54 show the expectancy on how the bond strength is assumed to
increase with increasing reduction. Each graph is built on four equations and plots. The
“Result Based” line, as the name indicates, is a curve based on the test results in this
thesis. The “Glue Limit” is the limit for which the specific glue used is the weaker of the
two bonds. Some methods have been tried out to increase this limit; this is discussed
in section 5.3.5. The two other dashed lines is an estimation on how the strength will
continue to increase, depending on the work hardening and alloy. These estimations
are based on a publication from W. Zhang et. al.[11] where the shape of the graph is
found to follow a 3rd grade polynomial equation. Further is the strength increase from
work hardening the determining factor for the maximum strength limit at high
reduction.
The information used to make these estimations was found on AluMatter[13]. For the
annealed (marked with an A)samples which only have been exposed for work
hardening during the roll bonding pass after annealing was compared to the Hx4 – half
hardened condition classification. The non-annealed (marked with NA) samples were
compared to the Hx9 – extra hard condition. In Table 10 below the extracted data is
presented. The upper limit of these curves is based on the minimum ultimate tensile
stress. As the bond strength between the two layers is dependent on the fraction of
actual bonded area, the RMN value found in the table will be much higher than the
expected bonding strength. The estimation that about 70% of the area is actually
bonded is more likely an overestimation, but a more plausible value. In addition there
is a small positive strength increase effect, by the extra work hardening that occurs
when base material is extruded between fracturing hard surface layers like oxide. This
effect is however neglected in comparison to the generous estimation already given. In
support of this neglected increase is the very thin oxide layer observed in these
experiments, as seen in Figure 48.The value used in producing the graphs is found
under RMN(70%) in Table 10.
Table 10: This table shows the assumed minimum UTS and min yield stress for two degrees of strain
hardening. The information is taken from AluMatter.info. [13]
1200
Hx4 (A)
Hx9 (NA)
RP02N
100 MPa
140 MPa
RMN
120 MPa
160 MPa
RMN(70%)
84 MPa
112 MPa
3103
Hx4 (A)
Hx9 (NA)
RP02N
110 MPa
165 MPa
RMN
140 MPa
185 MPa
RMN(70%)
98 MPa
130 MPa
RMN:
RP02N
RMN(70%):
58 | P a g e
Minimum Ultimate Tesile Stress
Minimum Yeild Stress at 0.2%
70% of RMN
Steinar Lauvdal
5 Discussion
The margins of error for the estimated lines are considerable large. After all these are
estimations and should be treated as such.
The “Glue Limit”, with a value set at 40MPa, is not an absolute limit, as the
performance of the glue is largely dependent on the preparations of all connecting
surfaces in adhesion process. This is described in section 3.4.1.1. However, the upper
limit at a stain rate of 0.2 mm/min seemed to approach 40MPa, and never above. Only
in two cases did the strength pass 40MPa (44.37MPa and 48.02MPa), but at a strain
rate of 10 mm/min. This effect is discussed further in the following section. With a few
exceptions of samples failing to around the 20MPa region, most of the samples where
the glue was the weakest link, the strength was rather stable within 34 to 40MPa, once
the preparation procedure was stabilized.
5.3.4 Strain Rate
An idea on whether or not the glue that was utilized might have hardening effect when
exposed for a high sudden energy burst. Like some vicious fluids can render low
resistance to an object moving slowly through it, and virtually act like a solid object, for
a short period of time, if struck with a high kinetic energy. Cornstarch mixed with
water is an example on such a fluid, along with many salt solutions and molten
polymers. It forms a fluid with a viscosity like syrup or a thin paste. These are nonNewtonian fluids and change their viscosity depending on the shear stress-rate it is
exposed to [14]. The idea was triggered by these fluids, that the glue, being composed
by lots of polymers, could have a similar effect when going through tensile testing at
higher strain rate. If this was the case, and the aluminum alloy was less influenced by
this effect than the glue, this could be a method for increasing the operational range of
this new TBS testing method.
The results did however clearly show a similar increase in both the strength of the glue
bond and the bonded sheets, canceling out the gain with increasing strain rate. In fact
some results even indicates that the aluminum bond is gaining more of the
strengthening effect by increased strain rate than the epoxy glue, as seen in Sample
A2NA03-T1 and T2 found in Table B 2. Where T1, strained at 0.2 mm/min, yielded in
the metallic bond at 38.29MPa, and T2, strained at 10 mm/min, yielded in the glue
bond at 44.37MPa. These two samples can also be seen plotted in Figure 39.
In conclusion; in the reduction range of 30-40% a strain rate at 10 mm/min compared
to 0.2 mm/min, showed in average a bond strength that was 10MPa higher. Whether
the increase is parallel or increasing as a percentage of the bond strength is not
possible to determine with this data-set.
5.3.5 Glue Limitations
Even as the glue failed to test the strong bonded samples, there is little reason to
complain about its strength. At its strongest, the glue withstood a tensile force up to
59 | P a g e
59
60
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
39.89MPa (at 0.2 mm/min strain rate) and 48.02MPa (at 10 mm/min strain rate) which
over the area of a disc with a diameter of 15mm equals hence 718 kg and 865 kg.
That’s above ¾ the weight of an average small car, hanging from something the size of
your papilla, or a small coin.
As introduced in section 4.2.2; information found on the datasheet for the glue [18],
presented steel as a better material for attaining high adhesive strength, much higher
than aluminum. The sample always would be in aluminum, and therefore the bond
between it and the glue would be the weakest link no matter how strong the bond
from the glue to the rod would be. This fact would never change. However, the reason
for attempting this was because the TBS method had currently reached a limit and this
method had potential to even out fluctuations with this limit.
When testing a TBST sample the outcome can technically end in fracture in one of five
different transitions. The first and most desired outcome, being the metal-to-metal
bond in the sample. While the other four is divided equally on each side of the sample.
Two bonds in between the rod and the glue and the other two between the glue and
the sample. This can be visualized in Figure 31 when imagining that the two open
spaces between the rods and the sample is filled with glue. Having four surfaces to
clean, scratch and apply glue to, includes a reasonable risk for failure. As it always will
be the weakest bond that yields, being able to remove two of the four bond transitions
form the equation, will in theory greatly reduce the chances for failure below the
assumed strength limit for the glue at 40MPa. The use of steel rods did however not
give any overwhelming results and was disregarded in favor of consistency and
reduced cutting time due to the much higher harder material.
5.3.6 TBST vs. SBST
The compared bond strength shown in Figure 42 is showing bond strength of the TBS
test both stronger and weaker then with the SBST method. The low amount of data
makes it hard to draw any hard conclusion. However, the only sensible thing to
conclude with, besides deeming the comparison inconclusive, is that the TBS test
seemed to be very unstable.
In one scenario one could say that the TBST method measures a higher strength than
by the SBST method, but due to various problems with the TBS test, some samples
have their bond damaged more than others, during machining etc.
Another scenario could be that the TBST and SBST method should show the same
strength, but due to the already mentioned reason, the TBST samples is now affected
by both a positive and negative increase in bond strength. The negative being the
same reason as previous, but the positive increase due to perhaps small weld zones in
the outer rim of the disc.
60 | P a g e
Steinar Lauvdal
5 Discussion
The conclusion will be that the tensile bond strength test (TBST) method still suffers
from startup hiccups, mainly with the machining of the disc, and even more when
adding the grooves. This method has shown that it is possible to do 100% tensile bond
strength test of bonds, using industrial type glue to distribute the load evenly over the
grip surface.
5.3.7 Applicability of the TBST Method
The TBST method as much revealed in the conclusion of section 5.3.6, still has factors
to sort out. However the test is currently fully applicable for tensile testing of samples
up to a strength of 40MPA, with a very low requirement of test material
This maximum strength limit of 40MPa is mainly due to two factors. The first being
that this is the limit of the current glue in use, and to the knowledge of the author
there is few other adhesives in the world that is stronger. One type of epoxy adhesive
under the name LOCTITE 9514 has reported tensile bond strength even higher than the
LOCTITE 9466, currently used. This glue however required curing at a temperature
over 120°C or above for reaching this high strength. An attempt was made to get a
sample of this glue, but was never attained in time.
The other factor is the difficulty of changing the ratio between the area of metallic
bonding to the glued area. If one could keep a large area for gluing while having the
strength evenly transferred to a much smaller roll bonded area, the lack in glue
strength could perhaps be overcome. The method of making grooves in the side of the
disc samples was an attempt on achieving this. This failed due to very thin discs and
improper cutting tools and methods.
5.4 Shear Bond Strength Test (SBST)
The SBS test, which was mentioned as the “state of the art” testing method of today,
was carried out on samples taken from the same roll bonding sheets as some of the
samples that was tested with the TBS test. This was done with a hope of observing a
relationship between the two testing methods, as was discussed in section 5.3.6.
In a study [17] where the SBS test method was the main means for testing equation
(12) found in section 2.4.2 was used in calculation of the gap overlap. In contradiction
to that equation, in this current experiments the calculations for the overlapping area
was only based on the yield strength of the material, as kept below an area that would
not lead to necking and failure in the material itself. On the other hand some of the
samples were poorly bonded and an area, as high as possible, was desired to avoid
fracture or damage on the bond during machining or clamping of the sample into the
tensile machine.
61 | P a g e
61
62
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
5.4.1 Angular Deflection
The plots in Figure 44 indicate that the angle deflection increase with increasing bond
strength, which is as expected. This variation in angle during testing of SBS will
obviously have an influence on the tested value. The pulling force is gradually moving
from pulling in a shear direction towards pulling in a partly tensile direction.
To reduce or perhaps avoid the angular deflection, the overlap length l can be reduced.
This will reduce the maximum load required to part the bonds and the shear forces
due to misalignment is reduced as well. The risk with reducing the overlap length is
that the sample is much more fragile, which can result in accidental fracture during
machining or when any other force is applied.
Figure 44 is showing a correlation to the bond strength. The angle is increasing with
increased bond strength in the samples where the overlap length is kept constant.
At this overlap length any notably angle deflection seems to occur over 20MPa. The
yield strength of the individual sample could be interesting to compare to see if there
was a threshold for where this effect started. It could be a direct correlation between
where the angle deflection starts and a specific fraction of the materials yield strength
at a given overlap length. This is mentioned in section 7.
5.5 Fracture Surface Investigation
In the following sections the observations made during the SEM analysis is discussed.
5.5.1 Bond Types
The main bond type found during the SEM investigation is ductile bonds, as seen in the
close-ups in Figure 55 and Figure 56. Figure 56 shows the bonding that occurs when
the oxide layer cracks during deformation and virgin base material is extruded through
the crack. These are characterized by the mountain range looking effects in the
surface. The reason for this long shape and direction is explained in section 5.5.2.
In Figure 55 a cluster of small dimples is observed. Such dimples are strongly indicating
a propper bonding. The dimples are formed as the virgin metal has bonded around
particles, and when the two layers was pulled apart these particle has created voids at
the sport they was encapsulated.
The size of the particles in these alloys is particularly small, and in the case in Figure 55
they are on the scale of 200-400nm.
62 | P a g e
Steinar Lauvdal
5 Discussion
Figure 55: A close-up image the fracture surface of sample B2A05-T1 A showing a proper bonded area,
indicated by the dimples. Seen from above.
5.5.2 Crack Direction
The characteristic ductile fracture lines found at higher reduction is a strong indication
of how the bonding occurs. This feature support the assumption that bonding occurs in
between the cracked up oxide layer. As the material mainly expands only in the rolling
direction during roll bonding the oxide layer and hardened surface cracks up and
exposes the virgin material in cracks stretching normal to the rolling directions. With
higher pressure also allowing metallic interaction and bonding. This tendency is
frequently found in the circular tensile test samples, but due to the shape of the
sample it was not, without further investigations, possible to determine the direction
in relationship to the rolling direction. This was first confirmed when observing the
fracture surface one of the SBST samples, where the rolling direction can be
recognized from the surface appearance. Figure 47 shows these bond lines going
normal to the rolling direction. This effect is also seen in the surface oxide fracture
investigation done by H.R. Lee et. Al [20], as seen in Figure 59 for the case of much
thicker oxide layers.
5.5.3 Bonded Area
The well bonded area of the fracture surface is easily found, but is a tedious job to
measure exactly. The most exact method would be to measure the area of each
“stretch lip” as they were named by A. Lilleby [5]. Since the observed lip is just the tip
of a ductile fracture point, the actual bonded area is larger. The original area can be
63 | P a g e
63
64
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
traced back by following the stretch lines down the slope of the lip. In Figure 56 the
real bonded area of such a stretch lip is marked. Dividing the found bonded area on
the total area will give the fraction of bonded area. Measuring the bond area of each
stretch lip over a sufficiently large area is very time consuming, unless a computer with
analytical software were made for this purpose.
Figure 56: Bonded area around a “stretch lip”. Showing the fracture surface of sample A2Na03-T1 A,
seen from above.
A simplified method can be used when making a few assumptions. As the sheets are
rolled in only one direction the surface expansion is unidirectional and should be
evenly distributed over the sheet. If line X is drawn, a certain distance anywhere on the
fracture surface in the rolling direction can be represented, as marked by the two
examples X1 and X2 in Figure 57. By measuring the sum of the width of every stretch lip
that the line crosses and dividing it by the total length X, the fraction of bonded area
can be found. Since the bonds should be evenly distributed a measurement for line X 1
and X2 in Figure 57 should give similar results. Of course an average of several lines
should be used if determining the bonded area.
(13)
64 | P a g e
Steinar Lauvdal
5 Discussion
Figure 57: A sketch showing a simplified method for measuring the fraction of bonded surface area.
Figure 58: The fracture surface of sample A2NA04-T1 seen from above, showing directional stretch
lips.
65 | P a g e
65
66
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
This method could be performed on a surface as seen in Figure 58. The sample in this
figure was roll bonded at a reduction of 48.5% and even measuring the fraction of
bonded area this image does not seems to be bonded anywhere close to 48.5% of the
surface area. In comparison the sample Figure 59, which is was rolled bonded at 30%
reduction seems to show a larger bonded area. When the bonded area does not follow
the expansion of the surface, hence the reduction, it seems like the oxide is expanding.
In section 5.8 a theory of thinning of oxide is discussed.
Figure 59: A SEM picture showing crack lines in an oxide covered surface, seen from above. [20]
5.6 The Effect of Rolling Speed
A study by Yan, H. et. al [24] suggest that a slower rolling speed will prolong contact
time while the material is compressed between the rolls and hence give time to better
bonding, as shown in Figure 60. These experiments were however performed with preheating of the samples up to 280°C.
Results found in a pre-study by Lauvdal [10] showed in one case the opposite of this, a
strength increase. These data does not disclaim the previous stated suggestion.
However, for cold roll bonding there must be another contributing factor. When the
speed is increased there is an amount of adiabatic heating produced from deforming
the material, that have less time to escape as the temperature builds up. The ways for
the heat to escape is through the air, which have a very low thermal conductivity. In
addition it can escape through the steel rolls and into the structure of the mill. Initially
66 | P a g e
Steinar Lauvdal
5 Discussion
some heat can also be transported to the not yet deformed part of the sample due to
aluminums high thermal conductivity.
When the roll bonding is performed at room temperature this heat contribution might
be the factor increasing the bonding strength by lowering the critical deformation
threshold. This factor might be of larger influence than the increase in rolling speed. In
cold roll bonding the effect of temperature increase seem to have a much higher
impact on the rolling strength than the effect of the contact time between the rolls. At
least it could be the case for a certain specter of the speed scale.
An IR-thermometer was used in an attempt to read the surface temperature of the
sheets as they exited the mill, but with no result. This current IR-thermometer cannot
properly read infrared radiation of an aluminum surfaces.
Figure 60: Graph showing the shear strength as a function of the rolling speed and reduction. [20]
Two results were found on the speed effect, and both are presented in section 4.6.
Without repeating the results in detail, the results gained from the test between a very
low rolling speed of 1.5 (about 5mm/s) and high speed at 5 (about 150mm/s) showed
that the samples rolled at speed 5 was far better bonded.
On the other hand, where speed was not intended tested, a trend was still visible
when plotted, as seen in Figure 33. The speed here was 3 (26.62 mm/s) and 5 (about
150 mm/s).
The conclusion to be drawn from this is that there is a chance that adiabatic heating
from higher rolling speed can have a desired effect on the bond strength when
performing cold roll bonding. This could be an interesting factor to study in a further
study.
67 | P a g e
67
68
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
5.7 Below Critical Deformation Threshold (CDT)
The critical deformation threshold is defined as the point where the bond is sufficiently
strong. Not a very exact description. This section will present the discoveries regarding
the low reduction region where bonding begins, an area also known under the name
kissing bond.
The reduction is obviously the parameter with the greatest impact on the bond
strength after sufficient surface brushing is gained. With increasing strain of the
material a larger reduction was acquired to gain bonding. This is because more energy
is needed to make the two base materials squeeze between the cracks in the oxide
layers and come in contact with the base material when an increased amount of
defects is present.
5.7.1 Al 1200NA
The lowest reduction attempted with the Al 1200NA alloy was 29.6%. At this reduction
two TBS tests were performed, one at a strain rate of 0.2 mm/min and another at 10
mm/min. The strength measured in these two tests was hence 6.63MPa and
19.49MPa. Also a SBS test was performed on this sample, but this failed in the material
rendering the results less valuable. The SBS test peaked at 13.02MPa indicating a
minimum strength value on the bond.
Of the three samples tested at this reduction, the sample tested at the strain rate 0.2
mm/min is chosen when predicting the initiation of kissing bonds. At 29.6% reduction
in thickness during bonding and the measured TBS of 6.63MPa, this is considered a
bond in the high end or above the kissing bond stage. As no lower reduction was
tested for this material, the point where bonding starts can only be estimated. From
the assumed slope of the other plots for the Al 1200NA samples and comparing to the
similar plotline in the Al 1200A graph an assumption is made that bonding could start
at about 25% reduction. To repeat, no data was recorded below 29.6% reduction.
5.7.2 Al 1200A
The samples in the annealed batch of the Al 1200 section have the best result for
determining the minimum reduction at which bonding begins. Two samples, one with
a reduction of 21.4% and another at 22.3% showed to be on opposite sides of this
border. One of these split straight after it came out of the mill and was never bonded.
The slightly higher deformed one was bonded well enough to survive until it
attempted machined into a tensile sample. It then split due to the external forces
applied. This alone indicates that a lower limit for bonding has been found, unless
small variables in uncontrollable parameters caused this. In addition, samples from the
same sheet were sent to machining for the SBS test. This sample on the other hand
survived the machining and was successfully tested to a SBS of 4.40MPa. The lower
limit for adhesion between the two plates is therefore well defined for roll bonding in
this material with the present parameters, and close to 22% reduction.
68 | P a g e
Steinar Lauvdal
5 Discussion
Due to the margins of error in several of the parameters included in the preparations
prior to roll bonding, this limit will assumable be somewhat floating limit in regards of
reproducing the results. As the small differences in exposure time to air and oxide
growth, degree of scratching and such may play a large role to this limit of roll
bonding. Also mechanisms other then the fracture of hard low ductile surface
elements and metallic bonding, may play a role in the first kissing bonds phase of
bonding.
5.7.3 Al 3103NA
At 20.4% reduction no bonding whatsoever was observed. At a reduction at 27.6% a
TBS of 6.17MPa at 0.2 mm/min strain rate was measured. With similar assumptions
and estimations bonding could here seem to first occur at around 25% reduction. Yet
again, it is important to stress that no real data was recorded below 27.6% reduction.
5.7.4 Al 3103A
In the Al 3103A samples no bonding where found at a reduction of 24.0%. At 28.1%
bonding was observed, but the sample fractured while clamping the sample to the
tensile machine and the bond strength was categorized as a kissing bond. The
following higher deformed samples, at 28.1%, 34.2% and 39.8%, were declared ruined
by machining as they both fractured when mounted in the tensile machine.
5.8 Bond Interface
In Figure 61 the interface is clearly visible, and in the center one can see the dark oxide
layer stretched out. Something else that is of interest concerning this is the clearly
visible lines indicating the deformation. Almost like furrows in wood one can see the
lines being bent around harder particles and oxide, and one can see how the oxide
layer has cracked and made an opening so that the virgin material could extrude
through and make contact in a metallic binding.
69 | P a g e
69
70
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Figure 61: The interface of sample A2NA02 seen from the side. At 2000x magnification one can clearly
see the deformation lines. The dark part stretching in the middle is the oxide layer, and in between
metallic bonding can be observed.
The fact that we can see such a distinct opening in the oxide layer could indicate that
the sample was cut along the rolling direction, or close to parallel. The author believes
at this point that it may be possible to determine the rolling direction by looking for
the weak lines left by the roller in the surface of the sample. This, however, the project
had not enough time to confirm. By looking at the surface of sample A2NA02 seen in
Figure 61, it was unofficially concluded that the cut was within a few degrees angle of
the rolling direction.
In the image of the sample that was stopped mid process, in Figure 49, it was not
possible to make out any interface line. This sample did not undergo any electro
polishing and is likely to explain why the interface is not apparent. However, the
interface was visible under low magnification in light microscope during the polishing
process in some of these samples, before it disappeared in the polishing stage.
In the same surface observations as the sample in Figure 61 the rolling direction for
the second disc sample was found to make a close to 45 degrees angle with the line of
the cut. No interface was found during the SEM imaging of this sample either.
70 | P a g e
Steinar Lauvdal
5 Discussion
Figure 62: Oxide layer seen from the side. Extruded aluminum trough cracks in thick oxide layer. [20]
The shape of the oxide layer in Figure 61 is very unlike any other found in most
interface pictures this author have seen. Compared to the shape of the oxide layer in
Figure 62, which has sharp edges, this new seems like it has been deformed and
flattened.
There is a significant difference in the thickness of the oxide in these two figures. While
the oxide thickness in Figure 61 is about 4-5µm thick, the oxide in Figure 62 is 4 times
thicker in its 20µm. Four theories on how this oxide might have been formed are
aerated in the following pages.
5.8.1 Theory I: Thinning of Oxide
One theory trying to explain the effect is thinning of oxide. Figure 63 from a) through
c) tries to illustrate this theory. The oxide is evenly distributed as the two layers are
brought together, just before any deformation has taken place, as seen in a). As the
material is deformed, the oxide layer starts to crack, as seen in Figure 62. The
deformation proceeds and gaps between the oxides grow and allows for more
bonding. At the same time, the oxide starts to deform on some level and creates the
shape seen I c). This theory is considered little plausible, as the aluminum oxide is a
extremely hard substance.
71 | P a g e
71
72
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Figure 63: A sketch of bond interface seen from the side, illustrating the progression in theory I.
5.8.2 Theory II: Uneven Oxide Layer Thickness
When the oxide forms on a roughly scratch brushed surface, the thickness over the
surface is unevenly distributed, as illustrated in Figure 64 a). When the sample is
deformed and stretched in the rolling direction, it is natural that the oxide layer breaks
off in the thinnest and weakest regions, seen in b). As the deformation continues, the
oxide is moved further apart. The thinning in the end of each oxide is simply because
the oxide shattered in the thinnest region.
Figure 64: : A sketch of bond interface seen from the side, illustrating the progression in theory II.
5.8.3 Theory III: Crack Direction
Another theory, with no base in known oxide behavior is the following: The oxide layer
cracks in an angle, in a straight line through, or perhaps changing direction, cutting out
a v-shape. This is illustrated in Figure 65 c) or the v-shaped in b). Due to the thin oxide
layer the crack-tips seems bigger. Under further deformation the v-opening might be
closed together under the pressure, resulting in a slim pointy end. With the straight
line cut, only some slight bending towards the center is needed to align the pointy end
to something similar of what is observed in Figure 61.
Figure 65: : A sketch of bond interface seen from the side, illustrating the progression in theory III.
5.8.4 Theory IV: “Pulverized” Oxide
Theory four is somewhat similar to theory one in the way that it can help explain a
thinning of the oxide. And as the fracture surface investigation could indicate, the
bonded area is not following the increase in reduction when comparing the amount of
72 | P a g e
Steinar Lauvdal
5 Discussion
bond lines to the reduction. The drawings in Figure 66 will help guide this explanation.
Part a) in the figure shows the two aluminum surfaces prior to scratch brushing, an
evenly distributed oxide layer. When the surface is scratched some oxide is ripped off
and thrown away from the sample. However some oxide might just be grinded to
smaller particles and thrown back down at the surface, most likely distributed in a
manner influenced of the scratch brushed surface. These are grooves going normal to
the rolling direction. New oxide is instantly formed on the exposed surface, creating
the thinnest part of the layer seen in b). If oxide is thrown back in a pattern, this will
lead to an uneven distribution of oxide when roll bonded, as illustrated in c). The oxide
layer is now partly composed by perhaps even more brittle oxide, as much of it is
compressed oxide particles. The uneven distribution explains the varying thickness.
Upon further deformation the oxide is compressed and stretched, and as in theory II it
also here dislocates at the weakest points, which is in the thinnest regions. In addition,
as this oxide is much more brittle, perhaps somewhat like sandstone, it is crushed
under the high external load; the whole mass might stretch like stepping on a pile of
sand. Or the shear forces in the transition zone between the oxide and the metal
breaks off and drags loose oxide fragments down the slope and out to the tips.
This theory could seem a bit more plausible than the others, and when looking closely
at the oxide in Figure 61 it does look like alienated particles are dislocated around the
tip.
Figure 66: : A sketch of bond interface seen from the side, illustrating the progression in theory IV.
73 | P a g e
73
74
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
74 | P a g e
Steinar Lauvdal
6 Conclusion
6 Conclusion
The basic conclusions that can be drawn from these discussions are as following:
The Material Effect
The two aluminum alloys used in this experiment, the 1200 and 3103, did not turn out
to have any notable variations on the bond strength. This is according to their close
similarity in strength.
Acetone vs. ethanol
Acetone can with reasonably probability be said to be a better degreasing agent than
ethanol may, due to which reduction bonding starts. The actual influence on the
bonding strength at higher deformation is not confirmed.
The Effect of the Oxide Layer Thickness
The comparisons experiments where the oxide layer was believed to be the main
variable, the results were somewhat uncertain. However the general trend did support
the assumption that bonding does start at a lower reduction when the oxide layer is
thinner.
Effect of rolling speed
Results indicate that the adiabatic heating from increased rolling speed may have a
significant influence on the bond strength when operating with cold roll bonding.
Effect of the General Preparation
In all general preparation method allowed bonding to start at the following reductions
at the given alloy and prior annealing:
Table 11: Table showing at the lowest reduction bonding was acquired.
Material
Reduction
Al 1200 Non-Annealed
29.6%
Al 1200 Annealed
22.3%*
Al 3103 Non-Annealed
27.6%
Al 3103 Annealed
28.1%
*This is the only value where the limit is accurate within less than +/- 1% margin.
The Tensile Bond Strength Test (TBST)
The strain rate at which the samples are tested is an important factor to take into
account when comparing measured bond strengths. At the reduction range 30-40% a
strain rate at 10 mm/min compared to 0.2 mm/min showed in average a bond
strength that was 10MPa higher. Whether the increase is linear or percentage based is
undetermined.
75 | P a g e
75
76
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
The current testing range for the TBS test is from around 4MPa to 40MPa. The lower
limit is based on the machining part of the preparations, where there is a considerable
risk that poorly bonded samples are parted due to the forces involved. This limit can
be lowered dramatically by altering the preparation approach, and should shear the
same potential limit as the SBS test.
The upper limit is confined on two fronts; the current glue has a bond strength of
40MPa and finding a stronger glue would directly improve this limit. The other option
is to increase the area ratio between the glued surface area over the roll bonded
surface area. Machining a groove over the interface of the disc is one approach to
achieve this.
The experimental machining of grooves to increase glue/bond area ratio turned out to
damage the bond in the process. New and improved methods of machining are
needed to achieve positive results using this method.
The TBST method, with its current limitations can serve as a complimentary test
method to the already established ones at the low to medium range of bond strength,
which in these experiments correlates to <40% reduction. As the test required a
minimum of sample material to be performed, it is very suitable for experiments with
limited sample size or testing products in a production line.
Angular Deflection in the Shear Bond Strength Test (SBST)
An angular deflection while performing the SBS testing was observed, and it showed to
increase with the bond strength (and reduction). The angular deflection was found to
start over bond strength of 20MPa.
TBST vs. SBST
Comparing the TBST method to the SBST method indicated that the TBST method is
currently unstable. The TBST method is most likely to report higher or equal bond
strength compared to a SBS test on the same sample material. To determine this
however, the inaccuracy of the TBS test is too big.
Fracture Surface
During the fracture surface investigation in the SEM, clear proof of ductile bonding was
found at higher reductions. The ductile bonds, by the given name “stretch lips”, run in
intervals normal to the rolling direction.
The fraction of area bonded over non-bonded did not seem to follow the percentage
of reduction, which indicate a thinning of the oxide layer. This could also be supported
by the shape of the oxide observed in the interface. Four theories were presented on
how this shape might have occurred. None were confirmed.
76 | P a g e
Steinar Lauvdal
7 Suggestion to Future Work
7 Suggestion to Future Work
Scratch Brushing
In a further study, there would be an interest looking into this parameter with regard
on the brushing direction and how the surface roughness and directions of the
grooves” may impact the bond.
Rolling speed & Adiabatic Heating
The observation that a slower rolling speed have had a positive effect on the bonding
strength and that the same positive effect is found on cold roll bonding only at higher
speeds, could be of interest to look closer into. Further an investigation on how small
temperature increases would affect the bonding strength at cold roll bonding.
Simulated Accumulated Cold Roll Bonding
How is the bonds affected by ACRB? A simple test by starting with a thick enough
sample and cold roll bonding it to for instance 50% reduction. Then, without stacking
or preparing the sample surface, CRB the sample to 25% of the original thickness. The
next to 12.5% and so on. Between each step, a sample is extracted for closer
investigations. Additionally can half the samples from each pass can be annealed
before continuing to the next step. This way gathering information on the annealing
influence in the process. Is annealing needed to reach high levels of ACRB to remove
accumulating stresses that can lead to brittle fracture? How often an to what degree is
annealing needed?
Fraction Bonded Surface Area vs. Reduction and Bond Strength
Bond strength should in theory follow the same % scale as the fraction boned area
multiplied with the yield strength. Investigating a quantitatively analysis of the fraction
bonded surface area on these samples, or samples with the same thinning effect of
oxide layer.
TBST Samples
An investigation of the bond interface on the surface of a test sample prior to TBS
testing could reveal if and how the machining procedure damages the samples. If the
scattering result in TBST data is not due to this is it perhaps just due to poorly bonded
areas?
Further new experimental studies can be carried out on the “groove” method for
improving the test range of the TBST. The first initiative would be to roll bond thicker
plates, to ease the practical circumstances around the machining. Thicker plates would
allow for experimentation of different shapes of groove-cuts. What is better for evenly
distribute the tensile load; v-shapes, circular shapes, squared shapes…? If the surface
area of the bond can be reduced to an half, while the glued surface remains
77 | P a g e
77
78
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
unchanged, the range of the TBST method would instantly be increased from 40MPa to
80MPa.
Some glues report bond strengths higher than the one currently used. Most of these
glues require however curing at elevated temperatures. En example of one of these
glues have been mentioned; the LOCTITE 9514.
SBST
The angular deflection found during the SBS testing could be investigated further, to
determine at which limits it does occur, and calculating a better estimate of the true
strength when it does occur.
78 | P a g e
Steinar Lauvdal
8 Bibliography
8 Bibliography
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
Dieter, George E. Mechanical Metallurgy. London: McGraw-Hill, 1988
Rolling & Drawing, Copyright © Brynmorgen Press 2001,
http://www.ganoksin.com/borisat/nenam/metal-rolling-n-drawing.htm
http://www.atomefabrik.com/pages_tetsuo/finish.htm
Bay, N. Mechanisms Producing Metallic Bonds in Cold Welding, 1983
Lilleby, Anders Experimental and Finite Element Studies of Cold Pressure
Welding of Commercial Purety Aluminium by Divergent Extrusion. Trondheim :
s.n., 2009
Quadir, M.Z. Influence of processing parameters on the bond toughness of rollbonded aluminium strip, 2008
Li, Long Progress in cold roll bonding of metals, 2008
http://www.ndt-ed.org/EducationResources/CommunityCollege/Materials/
Mechanical/Tensile.htm
Askeland, D.R. & Phulé, P.P. The Science and Engineering of Materials.Chap. 86 US: ©Thomson Canada Limit, 2006.
Lauvdal, S. Roll Bonding Properties at Room Temperature. NTNU Trondheim,
2010
Zhang, W. and Bay, N. Influence of Hydrostatic Pressure in Cold-Pressure
Welding. Inst. Of Manuf. Eng., Techn. Univ. Denmark, 1992.
http://www.aluminium.matter.org.uk/aluselect/06_composition_browse.asp
http://www.aluminium.matter.org.uk/aluselect/09_mech_browse.asp?
http://en.wikipedia.org/wiki/Non-Newtonian_fluid
Nes, E. Modelling of work hardening and stress saturation in fcc metals,
Progress in Materials Science, Vol. 41, pp 129-193, 1998
Humphreys, F.J. and Hatherly, M. Recrystallization and Related Annealing
Phenommena 2nd Ed., Chap. 6, 2004
Zhang, W. and Bay, N. Cold Welding – Experimental Investigation of the Surface
Preparation Methods, 1997
LOCTITE Hysol® 9466TM Technical Data Sheet. Henkel Technologies, 2006
Grong, Ø Metallurgical modeling of welding 2nd Ed., Institute of Materials,
London, 1997
H.R. Le, M.P.F. Sutcliffe, P.Z. Wang, G.T. Burstein Surface oxide fraction in cold
aluminum rolling, Acta Mater.52, 2004
Bay, N. Cross shear roll bonding J. Mater., 1994
Jamaati, R. Investigation of the parameters of the cold roll bonding (CRB)
process, 2009
Jamaati, R. Effect of friction, annealing conditions and hardness on the bond
strength of Al/Al strips produced by cold roll bonding, 2010
Yan, H. A study of warm and cold roll-bonding of an aluminum alloy, 2004
Designation: D1876-8 Standard Test Method for Pell Resistance of Adhesives (TPeel Test)1, 2011
Saito, Y., Tsuji, N., Utsunomiva, H., Sakai, T. and Hong, R. G. Ultra-fine grained
bulk aluminum production by accumulative roll-bonding (ARB) process. Acta
Mater. Vol. 39, pp 1221-1227, 1998
79 | P a g e
79
80
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
80 | P a g e
Steinar Lauvdal
Appendix
Appendix
A – Rolling Progression to Sheets
B – Table of all Tensile- and Shear-Test Samples
C – Adhesion Log
D – SEM Pictures: Tensile Samples
E – SEM Pictures: Shear Samples
81 | P a g e
81
82
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
A - Rolling Progression to Sheets
Table A 1: Logged thickness on each pass during cold rolling of material to the desired thickness.
pass
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
82 | P a g e
A1 1200
20,00mm
17,00mm
15,90mm
14,40mm
12,80mm
11,20mm
9,60mm
8,00mm
6,45mm
4,86mm
3,80mm
2,77mm
2,00mm
1,90mm
1,77mm
1,65mm
1,60mm
1,45mm
1,40mm
1,27mm
1,20mm
1,08mm
1,00mm
Batch
B1 3103 A2 1200
20,05mm 20,00mm
17,20mm 18,30mm
16,00mm 16,05mm
14,50mm 13,65mm
12,90mm 11,50mm
11,30mm 9,40mm
9,70mm 7,30mm
8,10mm 5,65mm
6,52mm 4,55mm
4,96mm 3,55mm
3,88mm 2,50mm
2,86mm 1,92mm
2,25mm 1,40mm
1,98mm 1,10mm
1,82mm 1,03mm
1,76mm
1,64mm
1,44mm
1,38mm
1,27mm
1,21mm
1,14mm
1,03mm
B2 3103
20,00mm
18,40mm
16,15mm
13,75mm
11,60mm
9,50mm
7,40mm
5,75mm
4,40mm
3,60mm
2,60mm
2,01mm
1,47mm
1,10mm
0,98mm
Steinar Lauvdal
Appendix
83
B - Tables of all Tensile and Shear Samples
Table B 1: The results both TBST and SBST for
AA1200 Annealed sample.
83 | P a g e
84
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Table B 2: The results both TBST and SBST for
AA1200 Non-annealed sample.
84 | P a g e
Steinar Lauvdal
Appendix
Table B 3: The results both TBST and SBST for
AA3103 Annealed sample.
85 | P a g e
85
86
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Table B 4: The results both TBST and SBST for
AA3103 Non-annealed sample.
86 | P a g e
Steinar Lauvdal
Appendix
87
C - Adhesion Log
Bonding Strength
Sample
Glue
CT Diameter
Load
Strain-rate
test
Bostik Epoxy Rapid days 20,0 mm 6 286 N 0,1 mm/min
test
Bostik Epoxy Rapid days 15,0 mm
0,1 mm/min
test
Hysol 9466 A&B
22 h 15,0 mm 5 494 N 0,1 mm/min
test
Hysol 9466 A&B
22 h 15,0 mm 5 246 N 0,1 mm/min
A1NA06-T1
Hysol 9466 A&B
91 h 15,0 mm 5 008 N 0,1 mm/min
A1NA07-T1
Hysol 9466 A&B
113 h 15,0 mm 6 295 N 0,1 mm/min
none
Hysol 9466 A&B
113 h 15,0 mm 5 696 N 0,1 mm/min
none
Hysol 9466 A&B
93 h 15,0 mm 6 991 N 0,1 mm/min
A1NA02-T1
Hysol 9466 A&B
93 h 15,0 mm 6 893 N 0,1 mm/min
A1A01-T1
Hysol 9466 A&B
138 h 15,0 mm 5 177 N 0,1 mm/min
A1A02-T1
Hysol 9466 A&B
138 h 15,0 mm 6 992 N 0,1 mm/min
A1A04-T1
Hysol 9466 A&B
138 h 15,0 mm 5 759 N 0,1 mm/min
A1A06-T1
Hysol 9466 A&B
138 h 15,0 mm 6 362 N 0,1 mm/min
B1NA01-T1
Hysol 9466 A&B
115 h 15,0 mm 6 732 N 0,2 mm/min
B1NA02-T1
Hysol 9466 A&B
115 h 15,0 mm 4 833 N 0,2 mm/min
B1NA03-T1
Hysol 9466 A&B
115 h 15,0 mm 6 615 N 0,2 mm/min
B1NA07-T1
Hysol 9466 A&B
115 h 15,0 mm 7 046 N 0,2 mm/min
A2NA01-T1
Hysol 9466 A&B
143 h 15,0 mm 1 171 N 0,2 mm/min
A2NA02-T1
Hysol 9466 A&B
143 h 15,0 mm 2 346 N 0,2 mm/min
A2NA03-T1
Hysol 9466 A&B
143 h 15,0 mm 6 763 N 0,2 mm/min
A2NA04-T1
Hysol 9466 A&B
143 h 15,0 mm 5 793 N 10,0 mm/min
A1A07-T1
Hysol 9466 A&B
143 h 13,2 mm 5 448 N 2,0 mm/min
A1A03-T1
Hysol 9466 A&B
143 h 13,2 mm 6 262 N 2,0 mm/min
B2NA03-T1
Hysol 9466 A&B
142 h 15,0 mm 2 244 N 0,2 mm/min
B2NA04-T1
Hysol 9466 A&B
142 h 15,0 mm 2 355 N 0,2 mm/min
B2NA05-T1
Hysol 9466 A&B
142 h 15,0 mm 3 708 N 0,2 mm/min
B2NA06-T1
Hysol 9466 A&B
142 h 15,0 mm 6 914 N 0,2 mm/min
A2A04-T1
Hysol 9466 A&B
76 h 15,0 mm 1 744 N 0,2 mm/min
A2A03-T1
Hysol 9466 A&B
76 h 15,0 mm 100 N
0,2 mm/min
B2NA02-T1
Hysol 9466 A&B
76 h 15,0 mm 1 090 N 0,2 mm/min
B2NA05-T2
Hysol 9466 A&B
76 h 15,0 mm 8 482 N 10,0 mm/min
A2NA01-T2
Hysol 9466 A&B
51 h 15,0 mm 3 443 N 10,0 mm/min
A2NA02-T2
Hysol 9466 A&B
51 h 15,0 mm 3 763 N 10,0 mm/min
A2NA03-T2
Hysol 9466 A&B
51 h 15,0 mm 7 836 N 10,0 mm/min
A2NA05-T1
Hysol 9466 A&B
51 h 15,0 mm 2 912 N 10,0 mm/min
B2NA02-T2
Hysol 9466 A&B
68 h 15,0 mm
8N
10,0 mm/min
B2NA05-T3
Hysol 9466 A&B
68 h 15,0 mm 6 066 N 0,2 mm/min
B2NA03-T2
Hysol 9466 A&B
68 h 15,0 mm 4 093 N 10,0 mm/min
B2NA04-T2
Hysol 9466 A&B
68 h 15,0 mm 3 866 N 10,0 mm/min
B2A06-T1
Hysol 9466 A&B
48 h 15,0 mm 4 400 N 0,2 mm/min
B2A05-T1
Hysol 9466 A&B
48 h 15,0 mm
0N
0,2 mm/min
B2A04-T1
Hysol 9466 A&B
48 h 15,0 mm 208 N
0,2 mm/min
A2A05-T1
Hysol 9466 A&B
48 h 15,0 mm
39 N
0,2 mm/min
A2A05-T2
Hysol 9466 A&B
97 h 12,9 mm 1 621 N 0,2 mm/min
A2A05-T3
Hysol 9466 A&B
97 h
0,2 mm/min
A2NA04-T2
Hysol 9466 A&B
97 h 15,0 mm 6 064 N 0,2 mm/min
B2A06-T2
Hysol 9466 A&B
97 h 12,2 mm 1 261 N 0,2 mm/min
A2NA04-T3
Hysol 9466 A&B
72 h 15,0 mm 5 257 N 0,2 mm/min
B2A06-T3
Hysol 9466 A&B
72 h 12,8 mm 2 492 N 0,2 mm/min
B1NA02-T2
Hysol 9466 A&B
72 h 15,0 mm 3 725 N 0,2 mm/min
B1A01-T1
Hysol 9466 A&B
72 h 15,0 mm 5 210 N 0,2 mm/min
Table C 1: Results log in relation to the TBST.
T-Strength Yielded
20,02 MPa Glue
Glue
31,11 MPa Glue
29,70 MPa Bond
28,35 MPa Glue
35,64 MPa Glue
32,25 MPa Glue
39,58 MPa Glue
39,03 MPa Glue
29,31 MPa Glue
39,59 MPa Glue
32,61 MPa Glue
36,02 MPa Glue
38,11 MPa Glue
27,36 MPa Glue
37,45 MPa Glue
39,89 MPa Glue
6,63 MPa Bond
13,28 MPa Bond
38,29 MPa Bond
32,80 MPa Bond
39,83 MPa Glue*
45,78 MPa Glue*
12,70 MPa Bond
13,33 MPa Bond
20,99 MPa Glue
39,15 MPa Glue
9,87 MPa Bond
0,57 MPa Bond
6,17 MPa Bond
48,02 MPa Glue
19,49 MPa Bond
21,31 MPa Bond
44,37 MPa Glue
16,49 MPa Glue
0,05 MPa Bond
34,34 MPa Glue
23,17 MPa Bond
21,89 MPa Bond
24,91 MPa Glue
0,00 MPa Bond
1,18 MPa Bond
0,22 MPa Bond
12,41 MPa Bond
34,33 MPa Glue
10,88 MPa Bond
29,76 MPa Glue
19,47 MPa Bond
21,09 MPa Glue
29,50 MPa Glue
Surface Preparation
Grit
Method*
Aceton
Aceton
320
1
320
1
120
2
120
2
120
2
120
2
120
2
120
2
120
2
120
2
120
2
120
3
120
3
120
3
120
3
120
3
120
3
120
3
120
3
120
4
120
4
120
3
120
3
120
3
120
3
120
3
120
3
120
3
120
3
120
3
120
3
120
3
120
3
120
3
120
3
120
3
120
3
80
3
80
3
80
3
80
3
80
3
80
3
80
3
80
3
80
3
80
3
80
3
80
3
87 | P a g e
88
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
D - SEM Pictures: Tensile Samples
The following figures show a selection of SEM pictures taken at various magnification
of the fracture surface of the tensile test samples. In most of the pictures a
denomination A and B, following the sample number relates to which of the two
fracture sides that is observed. All images is taken from above the sample.
Figure D 1: Sample A2NA01-T1; 6.63MPa at 29.6% reduction.
Figure D 2: A2NA01-T2 A; 19.49MPa at 29.6% reduction.
88 | P a g e
Steinar Lauvdal
Appendix
Figure D 3: A2NA02-T1 A; 13.28MPa at 33.0% reduction.
Figure D 4: A2NA02-T1 B; 13.28MPa at 33.0% reduction.
89 | P a g e
89
90
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Figure D 5: A2NA02-T2 A; 21.31MPa at 33.0% reduction.
Figure D 6: A2NA03-T1 A; 38.29MPa at 38.8% reduction.
Figure D 7: A2NA03-T1 B; 38.29MPa at 38.8% reduction.
90 | P a g e
Steinar Lauvdal
Appendix
Figure D 8: A2NA04-T1 A; 32.80MPa at 48.5% reduction.
Figure D 9: A2A03-T1 A; ~0.57MPa at 27.2% reduction.
Figure D 10: A2A04-T1 A; 9.87MPa at 34.5% reduction.
91 | P a g e
91
92
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Figure D 11: A2A05-T1 A; ~0.22MPa at 40.8% reduction. Broke when mounted.
Figure D 12: A2A05-T2 A; 12.41MPa at 40.8% reduction.
92 | P a g e
Steinar Lauvdal
Appendix
Figure D 13: B2NA02-T1 A; 6.17MPa at 27.6% reduction.
Figure D 14: B2NA02-T2 A; ~0.05MPa at 27.6% reduction. Broke when mounted.
93 | P a g e
93
94
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Figure D 15: B2NA03-T1 A; 12.70MPa at 31.6% reduction.
Figure D 16: B2NA03-T2 A; 23.17MPa at 31.6% reduction.
Figure D 17: B2NA04-T2 A; 21.89MPa at 35.7% reduction.
94 | P a g e
Steinar Lauvdal
Appendix
Figure D 18: B2NA04-T1 A; 13.33MPa at 35.7% reduction.
Figure D 19: B2A04-T1 A; ~1.18MPa at 34.2% reduction. Weakened /Damaged.
95 | P a g e
95
96
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Figure D 20: B2A05-T1 A; ~0MPa at 39.8% reduction. Broke when mounted.
Figure D 21: B2A06-T2 A; 10.88MPa at 45.4% reduction.
96 | P a g e
Steinar Lauvdal
Appendix
Figure D 22: B2A06-T3 A; 19.47MPa at 45.4% reduction. Forgot to remove glue from groove, hence a
too high strength was measured.
97 | P a g e
97
98
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
E - SEM Pictures: Shear Samples
The following figures show a selection of SEM pictures taken at various magnification
of the fracture surface of the Shear test samples. All images are taken from above the
sample.
Figure E 1: A2NA02; 15.98MPa at 33.0% reduction.
Figure E 2: A2A02; 4.48MPa at 22.3% reduction.
98 | P a g e
Steinar Lauvdal
Appendix
Figure E 3: A2A03; 8.36MPa at 27.2% reduction.
Figure E 4: B2NA03; 21.92MPa at 31.6% reduction.
99 | P a g e
99
100
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
Figure E 5: B2A04; 25.64MPa at 34.2% reduction.
Figure E 6: B2A05; 23.42MPa at 39.8% reduction.
100 | P a g e
Steinar Lauvdal
Appendix
Figure E 7: B2A06; 27.88MPa at 45.4% reduction.
101 | P a g e
101
102
Experimental Studies of Cold Roll Bonding of Aluminum Alloys
F - SEM Pictures: Bond Interface
The following figures show a selection of SEM pictures taken at various magnification
of the bond interface of sample A2NA02. All images are taken from the side of the
sample.
Figure F 1: The interface shown at 50x and 100x magnification.
Figure F 2: The interface shown at 200x magnification.
Figure F 3: The interface shown at 800x and 1´000x magnification.
102 | P a g e
Steinar Lauvdal

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project