Creep and Failure of Lead-Free Solder Alloys

Creep and Failure of
Lead-free Solder Alloys

J.G.A. Theeven
report number MT02.03
Masters thesis
Supervisor: prof.dr.ir. M.G.D. Geers
Coach (TU/e): dr.ir. W.P. Vellinga
Coach (Philips): dr. J.W.C. de Vries
Eindhoven University of Technology
Faculty of Mechanical Engineering
Materials Technology Group
Eindhoven, March 2002
Contents
Abstract iii
1 Introduction 1
2 Thermodynamics 3
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Extremum principles and evolution . . . . . . . . . . . . . . . . . . . . . 4
2.3 Diusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3 Plastic Deformation Mechanisms 11
3.1 Diusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2 Deformation mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4 Phase Diagrams 15
4.1 Thermodynamics of phase equilibria . . . . . . . . . . . . . . . . . . . . 15
4.2 The Sn-Ag-Cu System . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.3 The Sn-Bi-Ag system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.4 The Sn-Zn-Bi system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5 Experiments 22
5.1 Shear tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.2 Digital Image Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . 24
5.3 Contrast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.4 Creep and coarsening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
i
Contents ii
6 Experimental Results 28
6.1 Microstructures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
6.2 Solder A (SnPb36Ag2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
6.3 Solder B (SnAg3.8Cu0.7) . . . . . . . . . . . . . . . . . . . . . . . . . . 34
6.4 Solder C (SnAg3.3Bi3.82) . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6.5 Solder D (SnZn8Bi3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
7 Conclusions and Recommendations 46
Bibliography 48
A Pure element transition data 50
B Lattice stabilities 51
C Thermodynamic data for the Sn-Zn-Bi system 54
D Binary Ag-Cu phase diagram 55
E Binary Sn-Cu phase diagram 56
F Binary Sn-Ag phase diagram 57
G Binary Ag-Bi phase diagram 58
H Binary Sn-Bi phase diagram 59
I Binary Zn-Bi phase diagram 60
J Binary Sn-Zn phase diagram 61
K EDX Analysis 62
L Shear deformation of gold grid 66
M Hall of Fame 67
Samenvatting 70
Acknowledgements 72
Abstract
Increasing environmental awareness regarding lead-based solder alloys has caused major
eorts to develop a lead-free soldering technology. The RIPOSTE project of TU/e and
Philips CFT aims to construct a tool able to predict the reliability of miniature IC
package soldered interconnects, subjected to (thermo-)mechanical stress, taking into
account the relevant microstructure and chemistry of the solder. This thesis is part
of this project and deals mainly with shear loading, creep, failure and deformation
behaviour of three proposed lead-free solder alloys, i.e. SnAg3.8Cu0.7 (also called SAC),
SnAg3.3Bi3.82 and SnZn8Bi3 and SnPb36Ag2 as a reference.
Due to high homologous operating temperatures of solders used in electronics, defor-
mation of solder joints is always a thermo-mechanical process. This requires an under-
standing of equilibrium as well as non-equilibrium thermodynamic principles. During
deformation there is a constant ow of energy toward the system, implying that we are
not at all dealing with a system in thermodynamic equilibrium.
If the temperature of an electronic system rises, the printed circuit board expands more
than the component, due to the dierence in coecients of thermal expansion, resulting
in the solder connection being subjected to shear stress. Shear tests have been performed
with special samples to investigate the shear strength and maximum shear strain of the
solder joint using a miniature tensile stage. With the aid of digital image correlation
(DIC) techniques, deformation behaviour at the micron level could be observed and
compared to the shear test data.
Intermetallic compounds at the solder-substrate interface have shown to play a crucial
role in solder joint deformation for SnAg3.8Cu0.7 and SnZn8Bi3, where inhomogeneous
and highly local strains could be seen at the interface. SnAg3.3Bi3.82 showed a much
more homogeneous deformation leading to a superior shear strength, but brittle fracture
as well. SnAg3.8Cu0.7 showed the largest plastic deformation, whereas SnZn8Bi3 was
subject to peculiar ageing eects.
Shear stress in combination with high homologous temperatures also induces creep and
possibly coarsening. A creep test setup has been designed to test samples inside a stove.
It appeared that SnZn8Bi3 has the highest creep rate of the three lead-free solder alloys,
followed by SnAg3.3Bi3.82. SAC is the most creep-resistant alloy with about the same
creep rate as SnAg3.3Bi3.82 at 100C, but better resistance to creep at 150C.
iii
Abstract iv
ESEM and light microscopy revealed the failure mechanisms of the creep samples. In
SAC solder joints, cracks parallel to the interface and perpendicular to these cracks
(from one interface to the other), along large colonies could be seen and failure occurred
along cracks parallel to the interface throughout the bulk material. SnAg3.3Bi3.82 also
showed this, but failed along the intermetallics at the interface. SnZn8Bi3 showed colony
boundary sliding again, however, colonies in this material are much smaller and dier-
ently shaped, so cracks could be seen throughout the whole solder joint in any direction.
A more capricious failure occurred along these cracks throughout the bulk material and
not just along the interface. Also, EDX-analysis showed large diusion of copper to the
surface of the solder joint, forming brass with zinc.
In this thesis, baseline thermodynamic data from literature is presented, together with
thermo-mechanical data for the three lead-free solder alloys considered in the RIPOSTE
project. Combining these data with data from other research areas within this project
will lead to a better understanding of the properties of these materials and a good
replacement for lead-tin based solder alloys.
Chapter 1
Introduction
Lead-tin based solders have long been the most popular materials for electronic pack-
aging because of their low cost and superior properties required for interconnecting
electronic components. However, the toxic nature of lead and the increasing awareness
of its adverse eect on environment and health have given rise to the need for develop-
ment of lead-free solders in recent years. Several alloy compositions have been proposed,
however there is a general lack of engineering information and there is also signicant
disparity in the information available on these alloys. Nontoxic substitute materials
should satisfy a number of criteria if they are to serve as an eective replacement for
lead. A major factor aecting alloy selection is the melting point of the alloy, since this
has a major impact on the other polymeric materials used in microelectronic assembly
and encapsulation. Other important manufacturing issues are cost, availability and wet-
ting characteristics. Reliability related properties include mechanical strength, fatigue
resistance, coecient of thermal expansion and intermetallic compound formation.
When an electronic device is in operation, the solder connections are subjected to me-
chanical strains. The primary cause of these strains arise from the fact that the electronic
component and the board have dierent coecients of thermal expansion. An example
of how these strains are generated, between a silicon die and the substrate, is shown
in gure 1.1. If the temperature of the system rises, the board expands more than the
component, resulting in the solder connection being subjected to shear strain. As the
system is switched on and o, it is subjected to thermal cycling, resulting in the solder
connection being subjected to cyclic shear stresses.
Microstructure size has a great inuence on the plastic deformation kinetics of solder
joints. Diusion-driven separation of phases, also known as coarsening, has been well
examined for binary alloys, like eutectic Sn-Pb solder. Experiments have shown that
diusion processes can be considerably accelerated by combination of high homologous
temperatures with mechanical stresses [1]. This research will investigate coarsening phe-
nomena in ternary lead-free solder alloys, if present.
Constant thermo-mechanical loading will also induce creep. Creep is the most common
and important micromechanical deformation mechanism in solder joints, thus creep
1
Chapter 1. Introduction 2
Figure 1.1: Solder joints subjected to shear strain during thermal cycling due to CTE mismatch
between die, solder and substrate [2]
resistance is an important mechanical property. Accelerated thermo-mechanical testing
provides a useful insight in the creep resistance of the dierent kinds of solder alloys
investigated.
The three main goals of this research are (1) to provide baseline thermodynamic data
and theory on equilibrium and non-equilibrium thermodynamics regarding solder joints
for further numerical modeling, (2) to develop experimental tools necessary to test
creep, shear and deformation properties and (3) to provide thermo-mechanical data on
the aforementioned properties. By characterizing the deformation behaviour of solder
joints with the use of digital image correlation (DIC) techniques and identifying the
micromechanisms of their degradation, a much better understanding of the deformation
mechanism can be achieved.
This research is part of the RIPOSTE project (Reliability Improvement with Pb-free
solders to Outlive in Shock and high Temperature Environments) of Philips CFT and
TU/e. The aim of RIPOSTE is to construct a tool able to predict the reliability of
miniature IC package soldered interconnects, subjected to (thermo-)mechanical stress,
taking into account the relevant microstructure and chemistry of the solder. Adding up
the results of this research will lead to a better insight in the mechanical properties of
the proposed lead-free solder alloys. Combining this research with the other research
activities of RIPOSTE will hopefully lead to one winning lead-free solder alloy, which
will eventually lead to better performance and a cleaner environment.
Chapter 2
Thermodynamics
2.1 Introduction
Due to the high homologous temperature, in deformation of solder alloys one is always
dealing with a thermo-mechanical process. This implies that there is a coupling between
thermal and mechanical processes. This requires an understanding of equilibrium as well
as non-equilibrium thermodynamic principles, as well as of the specic nature of the
interactions leading to the energy terms involved in this process.
A major conceptual problem is to relate deformation mechanisms operating on several
levels to thermodynamic quantities. This is crucial, because during deformation there is
a continuous ow of energy toward the system, so we are not at all dealing with a system
in thermodynamic equilibrium. In fact, it is not even certain that we are dealing with a
system that evolves towards a thermodynamic equilibrium. In fact, we can envisage the
solder alloy being deformed as a closed system, and the solder alloy plus deformation
modes as an isolated system. For a closed system, that exchanges energy not matter,
the rst and second law read
dU = dQ+dW (2.1)
dS = d
e
S +d
i
S (2.2)
where
d
e
S =
dQ
T
and d
i
S 0 (2.3)
and d
e
S is dened to be the change of the systems entropy due to exchange of energy and
d
i
S the change in entropy production due to irreversible processes within the system.
3
Chapter 2. Thermodynamics 4
2.2 Extremum principles and evolution
In systems that are allowed to evolve spontaneously or isolated systems, the nal,
equilibrium situation is characterized by an extremum value of some thermodynamic
potential, for example the Helmholtz free energy F, the Gibbs free energy G or the
enthalpy H. Processes leading to a lowering of the thermodynamic potential may occur
spontaneously. An equivalent, but more general point of view is that all processes for
which dS
i
0 may occur and drive the system toward equilibrium. This is true regard-
less of the specic thermodynamic restrictions such as constant pressure or volume.
For the aforementioned thermodynamic potentials it follows that for any spontaneously
occurring process:
dF = TdS
i
0 (T, V = constant) (2.4)
dG = TdS
i
0 (T, p = constant) (2.5)
dH = TdS
i
0 (S, p = constant) (2.6)
Taking the Gibbs free energy as an example:
The Gibbs free energy is dened as [3]:
G = U +pV TS = H TS (2.7)
Using equation (2.2), one can write the change in Gibbs free energy as
dG = dU +pdV +V dp TdS SdT (2.8)
dG = dU +pdV +V dp Td
e
S Td
i
S SdT (2.9)
dG = dQpdV +pdV +V dp Td
e
S Td
i
S SdT (2.10)
At constant p and T this reduces to
dG = Td
i
S (2.11)
So at constant p and T, G is the appropriate potential. In thermodynamic equilibrium,
the value of some thermodynamic potential is at a minimum and that of the entropy at
a maximum. Phase diagrams are a way of presenting the various equilibrium phases and
their coexistence, at a certain T and p. Since they represent the ultimate state to which
a system may evolve, they form an important basis. So in the particular case of solders,
we are interested in explicit values for the Gibbs free energy. In conclusion, the system
may evolve toward equilibrium by a sequence of steps that lower G monotonically.
For the Gibbs free energy G we nd for a pure compound a, assuming the simplest
possible pairwise interaction:
G
a
=
Nz
2
E
a
(2.12)
where N is the mole number, z the coordination number and E
a
the interaction energy.
For a mixture of two phases with fractions x
a
and x
b
= 1 x
a
one assumes
G =
Nz
2
(x
a
E
a
+x
b
E
b
) +TS
mix
+G
mix
(2.13)
with
Nz
2
(x
a
E
a
+x
b
E
b
) the interaction energy between like atoms, G
mix
the interaction
energy between unlike atoms and TS
mix
the term due to mixing entropy S
mix
. For the
term S
mix
we nd the following. Consider the example of a crystal with a total of N
sites available for the occupation of atoms or molecules, n of which are occupied by
A atoms/molecules and (N n) are occupied by B atoms/molecules. It is clear that
the number of congurations the system can adopt depends on the number of possible
positions W in which the B atom or molecule can place itself.
S
mix
= k ln W (2.14)
where
W =
N!
n!(N n)!
(2.15)
Using Stirlings approximation this becomes
S
mix
= k[N ln N nln n (N n) ln(N n)] (2.16)
As the mole fractions of A and B are given by x
a
= (Nn)/N and x
b
= n/N, equation
(2.16) reduces to
S
mix
= N
A
k(x
a
ln x
b
+x
b
ln x
b
) = R(x
a
ln x
b
+x
b
ln x
b
) (2.17)
as R = N
A
k. This then denes the ideal entropy change on mixing. If there are no
repulsive or attractive interactions between atoms A and B the solution is called ideal
and the Gibbs energy of mixing is given by
G = RT(x
a
ln x
b
+x
b
ln x
b
) (2.18)
For the energy in the presence of the mixed terms one could dene
G =
Nz
2
x
a
((1 x
b
)E
a
+x
b
E
ab
) +
Nz
2
x
b
((1 x
a
)E
b
+x
a
E
ab
) (2.19)
with E
ab
an interaction energy between a and b atoms.
This leads to
G =
Nz
2
[x
a
E
a
+x
b
E
b
+x
a
x
b
(2E
ab
E
a
E
b
)] (2.20)
So for a binary mixture the term 2E
ab
E
a
E
b
constitutes the mixing term G
mix
and
determines whether the ab-bonds are preferred or not.
(Note that this can be written as
G =
Nz
2
[x
a
E
a
+x
b
E
b
+x
a
x
b
((E
ab
E
a
) (E
b
E
ab
))] (2.21)
preluding a second derivative term in the Gibbs free energy in case atoms a and b
separated by a straight interface.)
For more components, still restricting the interaction to pairs one nds:
G =
Nz
2
m
i
x
i
((1
m
j=i
x
j
)E
i
+
m
j=i
x
j
E
ij
), (2.22)
G =
Nz
2
m
i
(x
i
E
i
x
i
E
i
m
j=i
x
j
+x
i
m
j=i
x
j
E
ij
), (2.23)
G =
Nz
2
i
x
i
E
i
i
(x
i
E
i
m
j=i
x
j
x
i
m
j=i
x
j
E
ij
)
, (2.24)
G =
Nz
2
i
x
i
E
i
+
m
i
(
m
j=i
x
i
x
j
E
ij

m
j=i
x
i
x
j
E
i
)
. (2.25)
So for a ternary system:
G =
Nz
2
(x
a
E
a
+x
b
E
b
+x
c
E
c
+x
a
x
b
(2E
ab
E
a
E
b
)
+ x
a
x
c
(2E
ac
E
a
E
c
) +x
b
x
c
(2E
bc
E
b
E
c
)) (2.26)
In practice the situation is slightly more complex as the mixing terms are found to
be dependent on temperature and composition. One often encounters an approximate
description of these eects proposed by Muggianu as
G
mix
=
j>i
x
i
x
j
v
ij
(x
i
x
j
)
v
with v = 1 or 2 (2.27)
This leads for example, in the case of a ternary system to the somewhat more elaborate
expression
G
mix
= x
1
x
2
(
0
12
+
1
12
(x
1
x
2
))
+ x
2
x
3
(
0
23
+
1
23
(x
2
x
3
))
+ x
1
x
3
(
0
13
+
1
13
(x
1
x
3
)) (2.28)
with the
v
ij
as a +bT.
G =
i
x
i
G
0
i
+RT
i
x
i
ln x
i
+
j>i
x
i
x
j
v
ij
(x
i
x
j
)
v
(2.29)
where
v
ij
is a binary interaction parameter dependent on the value of v. The above
equation for G
mix
becomes regular when v = 0 and sub-regular when v = 1. In practice
the value of v does not rise above 2. Equation (2.29) assumes that ternary interactions
are small in comparison to those which arise from binary terms. This may not always
be the case and where there is evidence for higher-order interactions these can be taken
into account by a further term of the type G
ijk
= x
i
x
j
x
k
L
ijk
, where L
ijk
is an excess
ternary interaction parameter. Equation (2.29) is normally used in metallic systems
for substitutional phases such as liquid, FCC, BCC, etc. However, for phases such as
interstitial solutions, ordered intermetallics, ceramic compounds, slags, ionic liquids and
aqueous solutions, simple substitutional models are generally not adequate.
The integral Gibbs energy, G
0
i
, of a pure species is given simply by the equation
G
0
i
= H TS (2.30)
where H and S are the enthalpy and entropy as a function of temperature and pres-
sure. Thermodynamic information is usually held in databases using some polynomial
function for the Gibbs energy which, for the case of the Scientic Group Thermodata
Europe (SGTE), is of the form
G
m[T]
H
SER
m
= a +bT +cT ln(T) +
n
2
d
n
T
n
(2.31)
The left-hand side of the equation is dened as the Gibbs energy relative to a Standard
Element Reference state (SER) where H
SER
m
is the enthalpy of the element or substance
in its dened reference state at 298.15 K, a, b, c and d
n
are coecients and n represents
a set of integers, typically taking the values 2, 3 and -1. From equation (2.31), further
thermodynamic properties can be obtained. Thermodynamic data for pure elements can
be found in [4], where equation (2.31) is used. Data for pure elements in the form of
equation (2.33) can be found in appendix B.
S = b c c ln(T)
nd
n
T
n1
(2.32)
H = a cT
(n 1)d
n
T
n
(2.33)
C
p
= c
n(n 1)d
n
T
n1
(2.34)
Relation to chemical potential
From the denition of the internal energy for a homogeneous system U = TS pV +
k
N
k
it follows that
G =
k
N
k
(2.35)
for the molar Gibbs energy
G
m
=
k
x
k
(2.36)
and for
(dG
m
)
p,T
=
k
dx
k
(2.37)
Moreover we nd
dG = V dp SdT +
k
dN
k
(2.38)
so
G
p
T,N
k
= V ,
G
T
p,N
k
= S ,
G
N
k
T,p
=
k
(2.39)
2.3 Diusion
Consider a system consisting of two parts of equal T; one with chemical potential
1
and mole number N
1
, the other with chemical potential
2
and mole number N
2
. The
ow of particles from one part to the other can be associated with a single number of
each species involved:
dN
1
k
= dN
2
k
= d
k
(2.40)
For the entropy production we nd, summing over all species
d
i
S =
1
k
2
k
T
d
k
(2.41)
One can generalize to a continuously changing chemical potential along 1 direction by
dening the driving force across an innitesimal distance dx as

x
dx. Now,
d
i
S(x)dx =

k
k
x
dx
d
k
T
and (2.42)
d
i
S(x)dx
dt
=

k
k
x
dx
1
T
d
k
dt
(2.43)
Now introducing J
k
for the ux
d
k
dt
and furthermore concentrating on terms per unit
length, we get
d
i
S(x)
dt
=
k
x
1
T
J
k
(2.44)
At this point the assumption is made that the uxes J
k
are linearly dependent on the
driving force:
J
k
= L
k
1
T
k
x
(2.45)
In terms of concentration gradients:
J
k
= L
k
1
T
k
n
k
n
k
x
(2.46)
and depending on the sign of

k
n
k
n
k
x
we have Fick type diusion or uphill diusion.
Note that the mechanism of diusion represented here by the constant L
k
does not in
any way follow from thermodynamical considerations, and has to be put in, either from
empirical sources or theoretical considerations based on atomistics. For the change in
concentration over time one nds
n(x)
t
=
J
k
x
(2.47)
or
n(x)
t
=
L
k
T
k
x
2
(2.48)
Remembering the relations between G and the chemical potentials, the link between
the equilibrium phase diagram, the driving forces in non-equilibrium situations and the
uxes leading to equilibrium is clear. Input from outside thermodynamics is necessary
for the determination of the reaction/diusion mechanisms and speeds.
Chapter 3
Plastic Deformation Mechanisms
3.1 Diusion
In solids, atomic movements are restricted due to bonding to equilibrium positions.
However, thermal vibrations occurring in solids do allow some atoms to move. Diusion
of atoms in metals and alloys is particularly important since most solid-state reactions
involve atomic movements. There are two main mechanisms of diusion of atoms in
a crystalline lattice, i.e. the vacancy or substitutional mechanism and the interstitial
mechanism [5].
Vacancy or substitutional diusion mechanism
Atoms can move in crystal lattices from one atomic site to another if there is sucient
activation energy present provided by the thermal vibration of the atoms and if there are
vacancies or other crystal defect for atoms to move into (point defects). As the tempera-
ture of the metal increases, more vacancies are present to enable substitutional diusion
of atoms to take place. This process is called self-diusion, which activation energy is
equal to the sum of the activation energy to form a vacancy and the activation energy
to move the vacancy. In general as the melting point of the metal is increased, the ac-
tivation energy is also. This relationship exists because the higher-melting-temperature
metals tend to have stronger bonding energies between their atoms. Diusion can also
occur by the vacancy mechanism in solid solution. Atomic size dierences and bonding
energy dierences between the atoms are factors which aect the diusion rate.
The combination of enthalpy H and entropy S in G explains the fact that the presence
of point defects can be stable inside the metal. Although defects lead to an increase
of the internal energy U (in the surrounding area of the defect, the atoms are out of
position and every deviation from that position will lead to an increase of the poten-
tial energy), this increase of U will be compensated for in the free enthalpy G by the
simultaneous increase of the entropy S (G = HTS). In total, G will decrease because
of these defects, so dG < 0; exactly what nature strives for. The expression for the
free enthalpy also claries why the solubility of alien atoms in a metal increases with
increasing temperature. With equal increase of entropy (caused by the point defects)
11
Chapter 3. Plastic Deformation Mechanisms 12
the contribution of this to the decrease of G at higher temperature is greater, because S
appears in a product with T. The system can therefore allow more increase of internal
energy without losing equilibrium [6].
Interstitial diusion mechanism
The interstitial diusion of atoms in crystal lattices takes place when atoms move from
one interstitial site to another neighboring interstitial site without permanently displac-
ing any of the atoms in the matrix crystal lattice. For this mechanism to take place,
the size of the diusing atoms must be relatively small compared to the matrix atoms.
Small atoms such as hydrogen, oxygen, nitrogen and carbon can diuse interstitially in
some metallic crystal lattices.
3.2 Deformation mechanics
Plastic ow is a kinetic process. In general, the strength of the solid depends on strain,
strain rate and temperature. It is determined by the kinetics of the processes occur-
ring on the atomic scale, i.e. the glide-motion of dislocation lines, their coupled glide
and climb, the diusive ow of individual atoms, the relative displacement of grains
by grain boundary sliding (involving diusion and defect motion in the boundaries),
mechanical twinning (by the motion of twinning dislocations) and so forth [7]. These
are the underlying atomic processes which cause ow. But it is more convenient to
to describe polycrystal plasticity in terms of the mechanisms to which the atomistic
processes contribute. We therefore consider the following deformation mechanisms:
Collapse at the ideal strength
Flow when the ideal shear strength is exceeded.
Low-temperature plasticity by dislocation glide
(a) Limited by a lattice resistance, (b) limited by discrete obstacles, (c) limited
by phonon or other drags and (d) inuenced by adiabatic heating.
Low-temperature plasticity by twinning.
Power-law creep by dislocation glide and climb
(a) Limited by glide processes, (b) limited by lattice-diusion controlled climb
(high-temperature creep), (c) limited by core-diusion controlled climb (low-temperature
creep), (d) power-law breakdown (the transition from climb-plus-glide to glide
alone), (e) Harper-Dorn creep and (f) creep accompanied by dynamic recrystalli-
sation.
Diusional ow
(a) Limited by lattice diusion (Nabarro-Herring creep), (b) limited by grain
boundary diusion (Coble creep) and (c) interface-reaction controlled diusional
ow.
Power-law creep by climb and glide
At high temperatures, dislocations acquire a new degree of freedom; they can climb as
well as glide (gure 3.1).
Figure 3.1: Power-law creep involving cell-formation by climb. Power-law creep limited by glide
processes alone is also possible.
If a gliding dislocation is held up by discrete obstacles, e.g. intermetallic compounds
(IMCs), a little climb may release it, allowing it to glide to the next set of obstacles
where the process is repeated. The glide step is responsible for almost all of the strain,
although its average velocity is determined by the climb step. The important feature
which distinguishes this mechanism from others is that the rate-controlling process, at
an atomic level, is the diusive motion of single ions or vacancies to or from the climbing
dislocation, rather than the activated glide of the dislocation itself.
Creep deformation kinetics
The steady-state creep deformation kinetics of ne-grained metals and alloys such as
cold worked-and-annealed eutectic Sn-Pb alloys, generally exhibit the characteristic
behaviour given in gure 3.2. In such a log-log plot of the shear creep rate vs the
applied stress, four regions can be identied [8].
The rst three consist of straight line segments with slopes of about 3, 2 and 3-7,
respectively, and the fourth with an increasing slope greater than 10. For the eutectic
Pb-Sn alloy, regions I and II are grain size dependent, whereas regions III and IV are
independent of phase size, in keeping with the behavior of metals and alloys in general.
Steady-state creep is generally expressed by the Weertman-Dorn equation of the form
d
s
dt
=
AGb
kT
b
d
n
D
0
exp (Q/kT) (3.1)
where d
s
/dt is the steady-state strain rate, G the shear modulus, b the Burgers vector,
k Bolzmanns constant, T the absolute temperature, d the grain (phase) size, the
applied stress, D
0
the frequency factor, Q the activation energy for the deformation
process, n the stress exponent, p the grain size exponent and A a constant. Their values
Figure 3.2: Schematic of strain rate vs applied
stress showing four stages, each described by
the Weertman-Dorn equation
Figure 3.3: Strain rate vs stress for the stress
relaxation of bulk Sn
and the mechanism(s) which have been proposed for each of the four regions of gure
3.2 are listed in table 3.2 for the eutectic Sn-Pb solder [9].
Table 3.1: Parameters of the Weertman-Dorn equation and the proposed rate-controlling mech-
anism for the high-temperature plastic deformation of eutectic Sn-Pb solder
region n p H (kJ/mole) A proposed mechanism
I 1.7-3 2.3 84 10
15
grain boundary sliding
II 1.6-2.4 1.6-2.3 57 10
3
-10
5
grain boundary sliding
III 3-7 0 80-100 10
15
power-law creep
IV >7 0 80-100 - power-law creep breakdown
The deformation in region III should be attributed to dislocation climb and glide and is
also called matrix creep. Grain boundary sliding results in intergranular failure while the
other mechanisms lead to grain deformation. In the case of multi-phase materials with
large volume fractions of phases, interphase boundary sliding may act similar to grain
boundary sliding. As the strain rate is increased or the temperature is lowered, stress-
strain behaviour becomes increasingly less dependent on thermally activated processes
[10].
The distinct stages with constant slope as indicated in gure 3.2 for worked and annealed
microstructures do not in general occur for reowed solder joints. Rather, such plots
usually exhibit a continuous curvature with increasing slope, such as in gure 3.3.
Chapter 4
Phase Diagrams
Phase diagrams are a useful way of presenting the various equilibrium phases and their
coexistence, at certain compositions and temperatures. In this chapter, the ternary
phase diagrams of solder alloys B, C and D are reviewed (see table 4.1). Solder type
A is not included here, because this lead-containing alloy was merely added to this
research to serve as a reference for shear and creep tests. All of the phase diagrams were
taken from literature and have been calculated using the CALPHAD method, which is
shortly discussed in section 4.1.
Table 4.1: The four types of solder alloys investigated and their melting points
type composition melting point (C)
A SnPb36Ag2 179
B SnAg3.8Cu0.7 217
C SnAg3.3Bi3.82 210-216
D SnZn8Bi3 189-199
4.1 Thermodynamics of phase equilibria
Figure 4.1 shows one of the simplest forms of phase diagrams, a system with a miscibility
gap. It is characterized by a high-temperature, single-phase eld of which separates
into a two-phase eld between
1
and
2
below a critical temperature of 900 K. This
occurs because of the repulsive interactions between A and B (note that these A and
B are not the same A and B as in table 4.1). While a single minimum exists, the Gibbs
energy of the alloy is always at its lowest as a single phase. However, below 900 K the
system has further possibilities to lower its Gibbs energy. Figure 4.2 shows the G vs x
diagram at 600 K. If an alloy of composition x
0
were single phase it would have a Gibbs
energy G
0
. However, if it could form a mixture of two phases, one with composition x
1
and the other x
2
, it could lower its total Gibbs energy to G
, where G
is dened by
15
Chapter 4. Phase Diagrams 16
Figure 4.1: (a) Phase diagram for an AB system showing a miscibility gap and (b) respective
G vs x curves at various temperatures
G
=
x
0
x
1
x
2
x
1
G
2
+
x
2
x
0
x
2
x
1
G
1
(4.1)
and G
1
and G
2
are respectively the Gibbs energies of at composition x
1
and x
2
.
The equation is formed using the lever-rule [5]. A further separation to compositions
x
1
and x
2
sees a further reduction of the Gibbs energy of the two-phase mixture to
G
.
This process can continue but is limited to a critical point where the compositions
correspond to x
E
1
and x
E
2
where any further uctuation in the compositions causes
the Gibbs energy to rise. This point is then a critical point and the phases
1
and
2
with compositions x
E
1
and x
E
2
respectively are dened as being in equilibrium with
each other. At this point it is convenient to dene the fraction of each phase using the
equations
N
1
=
x
E
2
x
0
x
E
2
x
E
1
and N
2
=
x
0
x
E
1
x
E
2
x
E
1
(4.2)
where N
1
and N
2
are the number of moles of
1
and
2
respectively. The critical
position can be dened as follows. The system AB with composition x
0
has reached
an equilibrium where its Gibbs energy is at a minimum. Equation (4.1) can then be
used in combination with a Newton-Raphson technique to perform a Gibbs energy
minimalisation with respect to the composition of either A or B. The calculation of the
phase diagram is then achieved by calculating phase equilibria at various temperatures
below 900 K and plotting the phase boundaries for each temperature.
Figure 4.2: G vs x diagram at 600 K for an A B system shown in gure 4.1(a) showing
separation of a single-phase structure into a mixture of two phases
4.2 The Sn-Ag-Cu System
The thermodynamic assessment of the ternary Sn-Ag-Cu system is based on the binary
systems Ag-Cu, Sn-Cu and Sn-Ag as shown in appendices D, E and F, respectively [11].
The liquidus projection of the ternary system is shown in gure 4.3. Concerning the
substitute material for the Pb-Sn eutectic solder alloy whose melting temperature is
183C, one should pay attention to the Sn-rich corner of the liquidus surface diagram
[12]. Figure 4.4 shows the partial liquidus surface projection focused on the ternary
eutectic reaction in table 4.2
1
.
Table 4.2: Eutectic equilibrium of the Sn-Ag-Cu system
reaction phase mass % Ag mass % Cu mass % Sn
L 3.73 0.83 95.42
L

Ag
3
Sn + Cu
6
Sn
5
+ (Sn) Ag
3
Sn 73.17 0 26.83
(216.9C) Cu
6
Sn
5
0 39.07 60.93
(Sn) 0.07 0 99.93
4.3 The Sn-Bi-Ag system
Another group of candidate alloys for lead-free solder alternatives contains Sn, Bi and
Ag. The phase diagram provides a basic road map for the evaluation of properties
of alloys in this system. Unfortunately, an experimentally established ternary phase
1
values are taken from http://www.metallurgy.nist.gov/phase/solder/solder.html
Figure 4.3: Liquidus projection of the Sn-Ag-
Cu system
Figure 4.4: Liquidus surface in the Sn-rich
corner
diagram appears to be unknown. Nevertheless, the three binary systems Sn-Ag, Ag-Bi
and Sn-Bi are well documented and can be found in appendices F, G and H, respectively.
The Sn-Bi and Ag-Bi systems are simple eutectic systems with limited solubilities in
their terminal solid solutions [13]. The Ag-Sn system has two intermetallic phases,
both of which form peritectically, and also a eutectic equilibrium. When the binary
systems are known, thermodynamic calculation provides an extremely useful tool to
obtain quantitative information about the ternary system. From the obtained ternary
phase diagram, the melting temperature range for solders and the solidication path,
as well as the susceptibility to intermetallic formation with various substrates can be
found.
Extrapolation to the ternary system
In order to extrapolate the binary systems to a ternary system, some assumptions were
made. The phases L, (Sn) and (Ag) were treated in all three binary systems as solution
phases. For the ternary extrapolation of the Gibbs energies of these phases, no ternary
interactions are used for the solution phases. For the (Bi) phase, solubility was only
considered for Sn and, therefore, (Bi) is treated as a semi-stoichiometric compound with
no solubility for Ag. Negligible solubility of Bi in (Ag) and Ag
3
Sn was assumed. Thus,
(Ag) is treated as semi-stoichiometric compound and Ag
3
Sn is treated as an ordinary
stoichiometric compound in the calculation of the ternary system. The temperatures
and compositions of the phases for the invariant reaction are given in Table 4.3.
The predicted liquidus projection is shown in gure 4.5. Figure 4.6 shows the liquidus
surface in the Sn-rich corner.
Table 4.3: Eutectic equilibrium of the Sn-Ag-Bi system
reaction phase mass % Ag mass % Bi mass % Sn
L 0.68 55.85 43.47
L

Ag
3
Sn + (Sn) + (Bi) Ag
3
Sn 73.17 0 26.83
(137.1C) (Bi) 0 99.89 0.11
(Sn) 0.03 20.60 79.37
Figure 4.5: Liquidus projection of the Sn-Ag-
Bi system
Figure 4.6: Liquidus surface in the Sn-rich
corner
4.4 The Sn-Zn-Bi system
Thermodynamics
Due to a lack of data on thermodynamic properties of terminal solid solutions as well
as on their phase boundaries, only binary contributions were utilized, and the ternary
terms were ignored. In the case of the liquid phase, the availability of experimental data
allows the inclusion of in principle a ternary term. The Gibbs free energy of the liquid
phase is described using the following equation:
G
L
m
=

i=Bi,Sn,Zn
0
G
L
i
x
L
i
+RT
i=Bi,Sn,Zn
x
L
i
ln x
L
i
+ x
L
Bi
x
L
Sn
L
L
Bi,Sn
+x
L
Bi
x
L
Zn
L
L
Bi,Zn
+ x
L
Sn
x
L
Zn
L
L
Sn,Zn
+
ex
G
tern
(4.3)
where
L
L
i,j
=
n
m=0
m
L
L
i,j
(x
i
x
j
)
m
(4.4)
ex
G
tern
= x
L
Bi
x
L
Sn
x
L
Zn
(x
L
Bi
0
L
L
BiSnZn
+x
L
Sn
1
L
L
BiSnZn
+x
L
Zn
2
L
L
BiSnZn
) (4.5)
and where the coecient
m
L
L
i,j
is the parameter in the sub-binary system and
n
L
L
BiSnZn
may be temperature dependent. SGTE data for a selection of elements can be found in
appendix A. Additional SGTE data for Bi, Sn and Zn can be found in appendices B
and C. Binary phase diagrams of the Sn-Bi, Zn-Bi and Sn-Zn systems can be found in
appendices H, I and J, respectively.
Figure 4.8 shows calculated isothermal sections at 135, 170 and 250C, respectively.
The characteristic features of the phase diagram in this ternary system are (1) the
solubilities of Bi and Sn in (Zn) are negligibly small, (2) the solubility of Zn in (Bi) also
small, and (3) (Zn) directly equilibrates with liquid or (Bi) and (Sn). Figure 4.7 shows
the projection of the liquidus surface in the Sn-rich corner.
Figure 4.7: Liquidus projection in the Sn-rich corner
In the Sn-Zn-Bi system, no experimental data for the isothermal sections and very
limited information for the liquidus surface are available. Only a calculated liquidus
surface in the entire composition range was presented by Pelton et al. (1977), where the
calculated eutectic reaction occurs at about 137C and 2.5 at.% Zn and 54 at.% Sn. In
the calculation of Malakhov et al. [14], the eutectic reaction takes place at T = 130C
with compositions of four equilibrium phases given in table 4.4.
Figure 4.8: Calculated isothermal sections at (a) 135C, (b) 170C and (c) 250C
Table 4.4: Eutectic equilibrium of the Sn-Zn-Bi system
reaction phase Bi (wt.%) Sn (wt.%) Zn (wt.%)
liquid 54.54 42.75 2.71
L

(Bi) + (Sn) + (Zn) (Bi) 97.75 1.99 0.26
130C (Sn) 22.37 77.35 0.28
(Zn) 1.93 10
3
7.68 10
4
99.997
Chapter 5
Experiments
Dierent experiments have been done with the four solder alloys. This chapter will
rst describe what kind of sample has been used and how it should be polished in
order to expose the solders microstructure. Next, a concise description of the digital
image correlation technique used to measure strain elds will be presented, followed by
a description of the coarsening and creep test setup.
5.1 Shear tests
When an electronic device is in operation, the solder connections are subjected to me-
chanical stresses and strains. The primary cause of these stresses and strains arise from
the fact that the electronic component and the board have dierent coecients of ther-
mal expansion, as shown in gure 1.1. If the temperature of the system rises, the board
expands more than the component, resulting in the solder connection being subjected
to shear stress. In order to investigate the shear strains caused by these shear stresses
in more detail, it is necessary to perform shear tests on the four dierent kinds of solder
alloys. A special shear sample has been designed for this purpose, which is shown is
gure 5.1.
Figure 5.1: Shear sample
The copper plates are 25 mm long, 9 mm wide and 1 mm thick and are soldered together
by a 10 mm long and 0.5 mm wide solder strip. These measurements were chosen to make
22
Chapter 5. Experiments 23
it suitable for the miniature tensile stage. The tensile stage can be mounted under the
optical microscope as well as in the ESEM, so pictures can be taken during deformation.
These pictures can then be stored and analyzed later.
The samples were used for three important reasons. First of all, important data about
the shear properties of the dierent kinds of alloys can be obtained. Stress-strain curves
can be generated, which give valuable information on the maximum elastic and plastic
strains. Another reason is the investigation of the deformation behaviour of the solder
alloys; using DIC techniques (see section 5.2), strain elds inside the solder strip can be
visualized. And nally, microstructure coarsening under thermo-mechanical loading, as
well as creep, can be observed using a special test setup which is described in the next
section.
Polishing
When trying to expose and visualize a metals microstructure, grinding and polishing
is needed. For this purpose, a grinding table and a polishing machine (Struers DAP-
U with a Struers Pedemin 2 specimen mover) have been used. Because of the small
thickness of the samples (1 mm), it was impossible to press them onto the grinding
paper or polishing cloth manually. Therefore, the samples had to be glued onto small
steel cylinders which t in the three holes of the polishing machines specimen holder
plate. The glue used is based on cyanoacrylate and can be dissolved by immersing it in
acetone during approximately half a day (depending on the amount of glue used).
The polishing method is based on the Metalog Guide from Struers [15] for a type A
metal (very soft). Some adjustments have been made to optimize the polishing results
and at the same time reduce the total preparation time and reduce wear of the polishing
cloths. An overview of the correct preparation method for solder alloys is given below
in table 5.1. All grinding and polishing steps are done with a rotation speed of 150 rpm
and the given force is the force per sample. After each step, the samples must be cleaned
in an ultrasonic bath for 1-2 minutes to remove any remaining abrasive particles.
Table 5.1: Sample preparation method
cloth force (N) time (min.) lubricant abrasive size
plane grinding SiC paper 25 until plane water grit 320
ne grinding MD Largo 30 4 blue 9 m
MD Dur 25 3 blue 6 m
polishing MD Mol 20 2 red 3 m
MD Chem 10 1 OP-S -
5.2 Digital Image Correlation
Introduction
Digital Image Correlation (DIC) is a non-contact technique, used to measure strain
elds on specimens being subjected to deformation. The technique compares two digital
images, one taken before deformation and one taken after. The digital images are made
up of a rectangular array of grey pixels. These pixels are assigned an 8 bit value, so
that they can assume a grey level from 0 (white) to 255 (black). The grey level of the
pixel represents the light intensity, usually received from a camera. Computer software
(Aramis) uses correlation algorithms and displacement elds to match regions of the two
images to each other. From the dierences in positions of the features on the specimen
the strain can then be calculated. This technique will be particularly useful in analyzing
the inhomogeneous strain that can occur at or below the micron level [16].
Aramis
Mapping of the facets
The software initially places a square grid of equally spaced points over the source
image. Centered on each of these points is one square facet (gure 5.2).
Figure 5.2: Source grid and facets mapped to destination grid and facets
The software tries to match the facet from its position on the source image to its position
on the destination image. The software must nd the source facets deformation before
it can nd the best position of the match on the destination image. Further analysis of
the algorithm is beyond the scope of this report and can be found in [16].
Ideal images
The best images for the software to work with contain small, nely distributed features
with a high contrast. The software nds it easier to match this type of image as the
surfaces produced by the grey levels are identiable. A speckled image can also be
provided articially on low contrast samples, as described in section 5.3. The software
can then more easily map corresponding points from source to destination image.
5.3 Contrast
In order to gain more insight in the deformation behaviour of the solder alloys, DIC tech-
niques can be used to visualize the strain eld. However, the DIC software Aramis needs
high-contrast images to be able to calculate the strain eld accurately. Unfortunately,
the solder alloys didnt show sucient contrast, so this had to be created articially.
To achieve this on the shear samples, microsieves from Aquamarijn (a spin-o company
from the University of Twente
1
) have been used. A sketch of these microsieves is shown
in gure 5.3.
Figure 5.3: Microsieve
The black area is the silicon support for the nine membranes with a thickness of 1 m
(grey). Inside these membranes are areas of 500500 m (white) with circular holes
of 5 m diameter and 10 m apart. If this microsieve is placed on the solder joint of
the shear sample and gold is sputtered on the sample, the gold will pass through the
small holes onto the solder. Before doing this, the sample was painted dark grey, thus
creating a good contrast between the dark paint and light gold speckles. The microsieve
had to be pressed onto the microsieve, without braking it (the silicon support is very
1
http://www.microsieve.com
brittle). The design for a clamping device is shown in gure 5.4 and a close-up of the
bottom side of the microsieve holder is shown in gure 5.5. The rectangular pocket in
the microsieve holder is 0.3 mm deep, while the microsieve itself is 0.5 mm thick. This
means that the microsieve will stick out 0.2 mm, thus ensuring a good contact with the
shear sample. To make the handling a bit more easy, the microsieves were glued in the
microsieve holder, using a small amount of silicon gel on the edges.
5.4 Creep and coarsening
Microstructure size has a great inuence on the plastic deformation kinetics of solder
joints. Since room temperature for eutectic Sn-Pb alloys is already 0.65 T
m
(T
m
is the
melting temperature in K), phase coarsening by diusion can be expected at higher
temperatures. When subjected to additional mechanical loading, this process will be
accelerated [1][17]. Although coarsening has only been investigated in binary solders
like Sn-Pb, this research will also examine coarsening in ternary alloys, if present.
Creep can be expected as well, which will ultimately lead to failure of the solder joint.
To investigate these two phenomena, the test setup shown in gure 5.6 was used. The
device can be hung on a special rack and placed in a stove. Weights can be attached to
the bottom hook and because of the symmetrical design, the force will always be vertical
and no bending moment will act on the samples. This also speeds up the testing, because
two samples can be used at the same time.
The samples were taken out of the furnace briey every day and pictures were taken.
Because just half of the solder strip was painted grey, the evolution of the microstructure
could still be examined. The samples used for this experiment were also provided with
a gold grid, this time to measure creep. The coarsening/creep tests were performed at
two temperatures and two loads.
The results of the experiments described in this chapter will be presented in Chapter 6.
Figure 5.4: Microsieve clamping device Figure 5.5: Microsieve holder
Figure 5.6: Creep test setup
Chapter 6
Experimental Results
In order to gain better insight in the deformation behaviour of the proposed lead-free
solder alloys, one must look at the microstructural evolution of the solder joint at dif-
ferent scales and dierent modes. First of all, shear tests were performed to determine
the solders shear strength, maximum shear strain and crack formation. One level down,
DIC-measurements were done on all types of alloys to produce strain eld images of
the deforming solder joint. Furthermore, creep tests have been performed at two tem-
peratures and two loads, in order to determine creep rates for all solder types (table
6.1). Creep experiments have started with solders B and C at 100C. Due to a very
low creep rate, experiments with solder D have been done using higher loads. Solder
A showed a very high creep rate, so the load for these experiments was lowered. Shear
strain was measured with the gold grids as well, so the samples had to be taken out of
the stove briey every day to take pictures. These creep samples were then analyzed in
the ESEM to determine the failure mechanism, so this could be compared to the results
of the normal shear tests and DIC-measurements.
Table 6.1: Creep schedule
100C 150C
1 MPa AJ1 AJ2 -
1.5 MPA - A10 A13
2 MPa AJ3 AJ4 -
3 MPa - A11 A14
1.5 MPa C5 C7 C14 C16
2.5 MPa - C12 C13
3 MPa C8 C10 -
1.5 MPa B4 B9 BJ1 BJ2
2.5 MPa - B18 B19
3 MPa B7 B14 -
1.5 MPa - D9 D16
2 MPa D10 D13 -
3 MPa D6 D11 D14 D15
28
Chapter 6. Experimental Results 29
6.1 Microstructures
The microstructures of the four types of solder alloys in undeformed state have been ex-
amined in order to better understand dierences that occur in these microstructures in
the case of thermo-mechanical loading. Photographs taken with an optical microscope
are shown in gures 6.1 through 6.4, respectively.
Solder A shows a eutectic structure; the dark areas are Pb, the light phase is Sn. Silver
in the form of Ag
3
Sn can also be observed, only not clearly in gure 6.1. Figure K.15
in appendix K shows this more clearly. The needle shaped structure on the left is the
intermetallic compound (IMC) Cu
6
Sn
5
.
Solder B clearly shows two hexagonal Cu
6
Sn
5
particles. Here, the black areas are Sn
with Ag
3
Sn and Sn (eutectic) in between.
Figure 6.1: Microstructure of solder A (bright
eld)
Figure 6.2: Microstructure of solder B (dark
eld)
Solder C again shows Cu
6
Sn
5
particles (H-shaped). The black phase is Sn with a eutectic
mixture of Ag
3
Sn and Sn in between.
Solder D clearly shows the bright lamellar structures of Zn in a black Sn matrix with
Bi dissolved in the Sn.
EDX-analysis of all solder types has been done as well to determine the presence and
location of detected elements. Pictures can be seen in appendix K.
Figure 6.3: Microstructure of solder C (dark
eld)
Figure 6.4: Microstructure of solder D (dark
eld)
During soldering, reactions occur between solders and conductors and IMCs may nu-
cleate and grow at the solder/conductor interface. The presence of these IMCs is an
indication of good metallurgical bonding and a thin and continuous layer is an essential
requirement for good wetting and bonding. It also produces distinct improvements in
mechanical properties of joint. However, due to their brittle nature, too thick of an IMC
layer at the solder/conductor interface may degrade the reliability of the solder joint
[18]. In the case of alloys A, B and C, Cu
6
Sn
5
will form at the interface and in the case
of alloy D, CuZn ( brass) will form. This can also be seen in appendix K.
6.2 Solder A (SnPb36Ag2)
Shear data for solder joints made of this classic lead-containing alloy is shown in gure
6.5 and is corrected for the stretching of the 690N loadcell [19]. It is obvious that this
material shows a low maximum shear strength of about 33 MPa and has a relatively
low maximum shear strain. It should be noted that this data represents the solder joint
and not the material itself. As can also be seen from the DIC-measurements, strain
concentrates at the intermetallics at the interfaces, so its the combination of solder
alloy and interface that determines the mechanical properties of the joint.
0 0.5 1 1.5
0
5
10
15
20
25
30
35
40
Shear strain
S
h
e
a
r

s
t
r
e
s
s

[
M
P
a
]
Shear strain vs shear stress for solder A
Figure 6.5: Shear strain vs shear stress curves for type A solders
Post failure analysis of the shear samples revealed that almost all samples showed crack
initiation along both interfaces (gure 6.6).
Figure 6.6: Crack initiation along both interfaces
Strain eld images (gures 6.7 through 6.10) conrm crack formation along the interface.
High shear strains can be seen in this region, whereas deformation in the rest of the
solder joint is relatively low. The four markers on the x-axis show the macroscopic
strains where the four DIC pictures were taken.
Figure 6.7: Deformation at 19% shear strain Figure 6.8: Deformation at 26% shear strain
Creep tests for this solder were performed to serve as a reference for the other solder
types. However, this material showed very poor resistance to creep, so many samples
failed before pictures could be taken for strain measurements. To provide an indication
of the creep rates, table 6.2 shows the time-to-failure for a number of samples. An
interesting phenomenon, is that failure of the creep samples occurred completely along
one interface, so no crack initiation along both interfaces. So, the cracks along the grain
or colony boundaries inside the solder joint didnt seem to have a major inuence on
nal failure.
Table 6.2: Time-to-failure for solder A samples (hours)
100C (0.83T
m
) 150C (0.93T
m
)
1.5 MPA - A10 A13 16h
2 MPa AJ3 AJ4 48h -
3 MPa A16 A19 <24h A11 A14 <16h
ESEM-analysis has been performed on the creep samples to check the deformation
mechanism. Figure 6.11 shows a BSE-image of a deformed solder joint. Crack formation
can be seen clearly, as well as large deformations along the interface. Figure 6.12 shows a
close-up of the same sample, where can be seen that cracks form along colony boundaries.
Failure however occurred along the interface for every creep sample.
Figure 6.11: Sample AJ1, crack formation
along colony boundaries and interface (BSE
diode A-B image)
Figure 6.12: Sample AJ1, close-up of crack
formation along colony boundaries (normal
BSE image)
6.3 Solder B (SnAg3.8Cu0.7)
Shear data for solder B joints (also called SAC) is presented in gure 6.13 and the data
has been corrected for the stretching of the loadcell. As can be seen, the shear strength
of this alloy is comparable to that of solder A. There is one big dierence though, and
that is the relatively high maximum shear strain. So the rather low maximum shear
strength of this alloy is compensated by its enormous plastic shear deformation, which
can take on values of more than 2.
0 0.5 1 1.5 2 2.5 3
0
5
10
15
20
25
30
35
40
Shear strain
S
h
e
a
r

s
t
r
e
s
s

[
M
P
a
]
Shear strain vs shear stress for solder B
Figure 6.13: Shear strain vs shear stress curves for type B solders
When looking at the failure mode and crack formation during normal shear testing, we
can observe crack formation at the interface region, like weve also seen in solder A. This
can also be illustrated by looking at the strain eld images of the deforming solder joint
(gures 6.14 through 6.17, respectively). It is obvious that shear deformation localizes
at the interface(s), ultimately leading to failure in this region. Like with solder A, cracks
also initiated from both sides interface which could be observed for all samples. The
four markers on the x-axis of gure 6.13 show the macroscopic strains where the four
DIC pictures were taken.
Creep tests were performed according to table 6.1 and show a very high resistance to
creep, as displayed in gure 6.18. Figure 6.19 shows a log creep rate vs log shear stress
plot for SAC, taken from literature to serve as a reference [20]. It can be seen that
the creep rate from own experiments is higher than which can be seen in gure 6.19.
ESEM-pictures have been taken to investigate the microstructural evolution of the alloy
when undergoing constant thermo-mechanical shear loading.
10
0
10
1
10
10
10
9
10
8
10
7
10
6
10
5
10
4
log shear stress [MPa]
l
o
g

s
h
e
a
r

s
t
r
a
i
n

r
a
t
e

[
1
/
s
]
Average creep rate vs shear stress (solder B)
100 C (0.76Tm)
150 C (0.86Tm)
Figure 6.18: Log creep rate vs log shear stress
for solder B (own results)
Figure 6.19: Log creep rate vs log stress for
solder B, taken from literature
Figures 6.20 and 6.21 show an optical microscope photograph and an ESEM-photograph
of creep-induced crack formation along colony boundaries. Note the cracks forming
parallel to the interface as well as perpedicular to the interface and the deformation
along the bottom interface in gure 6.21. The cracks parallel to the interface ultimately
lead to failure of the solder joint, whereas failure during normal shear tests occurs along
the interfaces.
Figure 6.20: Sample B19, crack formation
along colony boundaries (dark eld)
Figure 6.21: Sample B19, crack formation
along colony boundaries (BSE diode A-B)
To complete the microstructural analysis, an EDX-map has been made to determine
the presence and location of the alloys elements. Figures 6.22 through 6.25 show the
diusion of copper into the bulk material and the nely dispersed silver particles, most
of which is actually Ag
3
Sn. Copper and tin can be found in the interface; this is the
intermetallic compound Cu
6
Sn
5
.
Figure 6.22: SE-image Figure 6.23: Silver
Figure 6.24: Copper Figure 6.25: Tin
6.4 Solder C (SnAg3.3Bi3.82)
Shear data for solder C joints is plotted in gure 6.26, where the data has been corrected
for the stretching of the loadcell. The rst feature that can be noticed is the very high
shear strength, unlike any other solder type investigated. A disadvantage however, is
that this material fractures in a brittle way. This could be observed for all samples,
except for sample C4 (curve with the highest maximum shear strain). Not all samples
that were investigated are shown in this plot, because of a problem with the displace-
ment data. These samples did however show brittle fracture and about the same shear
strength. Failure occurred along one interface as well as both interfaces (crossing over
of the crack, like solders A and B) for some samples.
0 0.5 1 1.5
0
10
20
30
40
50
60
70
Shear strain
S
h
e
a
r

s
t
r
e
s
s

[
M
P
a
]
Shear strain vs shear stress for solder C
Figure 6.26: Shear strain vs shear stress curves for type C solders
The strain eld images clearly show a dierent pattern in the shear strain than those
for solders A and B. As can be seen in gures 6.27 through 6.30, deformation is much
more homogeneous and no large shear strains along the interface can be seen, except
for at the very last moment (gure 6.30 was taken 4 seconds prior to fracture). The four
markers on the x-axis show the macroscopic strains where the four DIC pictures were
taken.
As for solder B, this solder also shows good resistance to creep, as can be seen in gure
6.31. Creep tests were performed according to table 6.1. Closer examination of the
samples using light microscopy and the ESEM reveals that this material is also subject
to colony boundary sliding during constant thermo-mechanical loading.
10
0
10
1
10
10
10
9
10
8
10
7
10
6
10
5
10
4
l
o
g

s
h
e
a
r

s
t
r
a
i
n

r
a
t
e

[
1
/
s
]
Average creep rate vs shear stress (solder C)
100 C (0.77Tm)
150 C (0.87Tm)
Figure 6.31: log creep rate vs log shear stress for solder C
Figure 6.32 shows cracks forming along colony boundaries. A remarkable feature of this
alloy is, that failure occurred along the interface and not along the colony boundary
cracks. Figure 6.33 shows a close-up of the interface, where sliding along the intermetallic
layer (Cu
6
Sn
5
) can be seen.
Figure 6.32: Sample C10, crack formation
Figure 6.33: Sample C10, sliding of grains
along intermetallics at the interface
Again, an EDX-map has been made to show the colonies in a dierent way. Figures
6.34 through 6.37 show the distribution of copper, silver and tin. Bismuth was left
out, because this was nely dispersed throughout the whole region and no particular
concentrations could be found anywhere. It can be seen from the silver map that the
orientation of the grains is dierent for both sides of the small crack, thus indicating a
colony boundary.
Figure 6.34: SE-image Figure 6.35: Silver
Figure 6.36: Copper Figure 6.37: Tin
6.5 Solder D (SnZn8Bi3)
Shear data for solder D joints can be seen in gure 6.38. Again, data has been cor-
rected for the stretching of the loadcell. As can be observed, this material shows great
inconsistency in the shear strength. This can however be explained by looking at the
microstructure in the course of time. The appearance of air bubbles could be detected
after a period of time. Immediately after polishing, no air bubbles could be detected,
but the older the samples got, the greater the amount of holes became. The two shear
stress-shear strain curves with the lowest shear strength were the oldest and showed a
large amount of holes, leading to weakening of the solder joint. The two curves with the
highest shear strength were the youngest. This probably has something to do with the
easy oxidation of zinc.
0 0.5 1 1.5
0
10
20
30
40
50
60
Shear strain
S
h
e
a
r

s
t
r
e
s
s

[
M
P
a
]
Shear strain vs shear stress for solder D
Figure 6.38: Shear strain vs shear stress curves for type D solders
When we look at the shear deformation of this alloy, gures 6.39 through 6.42 show
crack formation along the interface, but also through the bulk material. However, DIC
measurements of old samples (with air bubbles) showed crack formation along the in-
terface, where most of the holes appeared. The four markers on the x-axis show the
macroscopic strains where the four DIC pictures were taken.
Creep data was collected again according to table 6.1. It can be seen that this alloy has
a higher creep rate than solder types B and C, which makes this alloy, in combination
with the formation of air bubbles, more unreliable.
10
0
10
1
10
10
10
9
10
8
10
7
10
6
10
5
10
4
l
o
g

s
h
e
a
r

s
t
r
a
i
n

r
a
t
e

[
1
/
s
]
Average creep rate vs shear stress (solder D)
100 degrees
150 degrees
Figure 6.43: log creep rate vs log shear stress for solder D
Again, light microscopy and ESEM microscopy were utilized to examine the microstruc-
tural evolution of this material. Figures 6.44 and 6.45 very clearly show crack formation
along colony boundaries. It can be seen that colonies are dierently shaped and smaller
than the colonies that could be seen in solder types B and C. This ultimately leads to
intergranular failure, rather than failure along the interface. This could also be observed
from the normal shear experiments and DIC measurements.
Figure 6.44: Sample D11, crack formation
along colony boundaries, (bright eld)
Figure 6.45: Sample D11, crack formation
An interesting feature of the thermal aging of this solder type is the vast diusion of
copper to the surface of the joint, forming brass with zinc. EDX-analysis of sample D16
clearly shows this phenomenon (gures 6.46 through 6.49).
Figure 6.46: SE-image Figure 6.47: Tin
Figure 6.48: Zinc Figure 6.49: Copper
Chapter 7
Conclusions and
Recommendations
All three lead-free solders have shown to have advantages over SnPb36Ag2, the most
common solder alloy used in the past. DIC showed that intermetallic compounds at the
solder-substrate interface play a crucial role in solder joint deformation for SnAg3.8Cu0.7
(SAC) and SnZn8Bi3, where inhomogeneous and highly local strains could be seen at
the interface. SnAg3.3Bi3.82 showed a much more homogeneous deformation leading to
superior shear strength, but brittle fracture as well. SnAg3.8Cu0.7 showed the largest
plastic deformation, whereas SnZn8Bi3 was subject to aging eects. The older the sam-
ples got, the more air bubbles could be detected inside the solder joint, which have
detrimental eects on shear strength. A summary of the shear test results can be seen
in table 7.1.
Table 7.1: Summary of shear test results
Solder type A B C D
max
[MPa] 32-35 32-37 66-67 5842
max
1-1.3 2-2.5 0.7 0.61
failure ductile ductile brittle ductile at rst,
fracture fracture fracture brittle after ageing
The DIC measurements done with the gold grid sputtered onto the solder joints gives
very good results. Aramis had no problems to correlate subsequent facets with each
other. This method will therefore be suitable for other applications, where strain elds
have to be calculated as well.
It appeared that SnZn8Bi3 has the highest creep rate of the three lead-free solder alloys,
followed by SnAg3.3Bi3.82 as can be seen in gure 7.1. SAC is the most creep-resistant
alloy with about the same creep rate as SnAg3.3Bi3.82 at 100C, but better resistance
to creep at 150C. SnPb36Ag2 showed such poor creep resistance that most of the
samples failed before creep could be measured, so the superiority of the lead-free solders
46
Chapter 7. Conclusions and Recommendations 47
10
0
10
1
10
10
10
9
10
8
10
7
10
6
10
5
10
4
Average creep rate vs shear stress for all solder types
l
o
g

s
h
e
a
r

s
t
r
a
i
n

r
a
t
e

[
1
/
s
]
solder B, 0.76Tm
solder B, 0.86Tm
solder C, 0.77Tm
solder C, 0.87Tm
solder D, 0.8Tm
solder D, 0.9Tm
reference line n=3
Figure 7.1: Creep data for all solder alloys
with regard to creep is evident. Unfortunately, coarsening could not be observed, so no
data on this subject could be collected.
Furthermore, one important general conclusion that can be drawn from the pictures
taken with the optical microscope and ESEM, is that failure occurs along colony bound-
aries. Each type of solder alloy investigated showed this behaviour, as can be seen by
the pictures presented in chapter 6. If possible, formation of colonies should be avoided
in the reow process, as these will have a detrimental eect on solder joint reliability
at constant thermo-mechanical loading. In SAC solder joints, cracks parallel to the in-
terface and perpendicular to these cracks (from one interface to the other), along large
colonies could be seen and failure occurred along cracks parallel to to interface through-
out the bulk material. SnAg3.3Bi3.82 also showed horizontal and vertical cracks along
large colony boundaries, but didnt fail along these cracks. Failure did occur along the
intermetallics at the interface. SnZn8Bi3 showed the same phenomenon as the other
two lead-free solder alloys: colony boundary sliding. However, colonies in this mate-
rial are much smaller and dierently shaped, so cracks could be seen throughout the
whole solder joint in any direction. A more capricious failure occurred along these cracks
throughout the bulk material and not along the interface. Also, EDX-analysis showed
large diusion of copper to the surface of the solder joint, forming brass with zinc.
The three main objectives, as described in the introduction, were all met. Baseline
thermodynamic data from literature for further numerical modeling has been given,
as well as basic theory on equilibrium and non-equilibrium thermodynamic principles.
Futhermore, tools have been developed to measure the shear, creep and deformation
properties of the dierent kinds of solder alloys and nally, data on these properties has
been collected to be used for the RIPOSTE project. Combining the results presented in
this thesis with results from other research areas within this project will lead to a good
replacement for current lead-containing solders.
Bibliography
[1] W. Dreyer and W.H. M uller. Modeling Diusional Coarsening in Eutectic Tin/Lead
Solders: a Quantitative Approach. International Journal of Solids and Structures,
38, 2001.
[2] M. Abtew and G. Selvaduray. Lead-Free Solders in Microelectronics. Materials
Science and Engineering, 27, 2000.
[3] N. Saunders and A.P. Miodownik. CALPHAD, A Comprehensive Guide. Permagon
Materials Series. Elsevier Science Ltd., 1998.
[4] A.T. Dinsdale. SGTE Data for Pure Elements. CALPHAD, 15, 1991.
[5] W.F. Smith. Foundations of Materials Science and Engineering. Engineering Me-
chanics Series. McGraw-Hill, Inc., 1993.
[6] D. Landheer. Metaalkunde. Lecture notes, august 2000.
[7] H.J. Frost and M.F. Ashby. Deformation Mechanism Maps. Permagon Press, 1982.
[8] H. Conrad, Z. Guo, Y. Fahmy, and D. Yang. Inuence of Microstructure Size on the
Plastic Deformation Kinetics, Fatigue Crack Growth Rate and Low-Cycle Fatigue
of Solder Joints. Journal of Electronic Materials, 28(9), 1999.
[9] P.L. Hacke, A.F. Sprecher, and H. Conrad. Microstructure Coarsening During
Thermo-Mechanical Fatigue of Pb-Sn Solder Joints. Journal of Electronic Materi-
als, 26(7), 1997.
[10] H. Mavoori, J. Chin, S. Vaynman, B. Moran, L. Keer, and M. Fine. Creep, Stress
Relaxation and Plastic Deformation in Sn-Ag and Sn-Zn Eutectic Solders. Journal
of Electronic Materials, 26(7), 1997.
[11] K.W. Moon, W.J. Boettinger, U.R. Kattner, F.S. Biancaniello, and C.A. Handw-
erker. Experimental and Thermodynamic Assessment of Sn-Ag-Cu Solder Joints.
Journal of Electronic Materials, 29(10), 2000.
[12] I. Ohnuma, M. Miyashita, K. Anzai, X.J. Liu, H. Ohtani, R. Kainuma, and
K. Ishida. Phase Equilibria and the Related Properties of Sn-Ag-Cu Based Pb-
Free Solder Alloys. Journal of Electronic Materials, 29(10), 2000.
48
BIBLIOGRAPHY 49
[13] U.R. Kattner and W.J. Boettinger. On the Sn-Bi-Ag Ternary Phase Diagram.
Journal of Electronic Materials, 23(7), 1994.
[14] D.V. Malakhov, X.J. Liu, I. Ohnuma, and K. Ishida. Thermodynamic Calculation
of Phase Equilibria of the Bi-Sn-Zn System. Journal of Phase Equilibria, 21(6),
2000.
[15] Struers. Metalog Guide, 2000.
[16] D. Lu. An Analysis of Digital Image Correlation Applied to Scanning Electron
Microscope Images. Technical Report MT00.020, TU/e, 2000.
[17] L. Li and W.H. M uller. Computer Modeling of the Coarsening Process in Tin-Lead
Solders. Computational Materials Science, 21, 2001.
[18] K. Zeng and J.K. Kivilahti. Use of Multicomponent Phase Diagrams for Predicting
Phase Evolution in Solder/Conductor Systems. Journal of Electronic Materials,
30(1), 2001.
[19] S.H.A. Boers. Private communication.
[20] A. Schubert, H. Walter, A. Gollhardt, and B. Michel. Materials Mechanics and
Reliability Issues of Lead-free Solder Interconnects. 2001.
Appendix A
Pure element transition data
Values below are taken from [4] and are in J/mole.
H
298
H
0
S
298
T
trans

trans
H
trans
S
trans
C
p
Ag FCC A1 5745 42.55 1234.93 11296.8 9.1477 1.5402
Bi RHO A7 6426.624 56.735 544.55 11296.8 20.7479 0.6521
Cu FCC A1 5004 33.15 1357.77 13263.28 9.7684 1.5391
Pb FCC A1 6870 64.8 600.612 4773.94 7.9485 1.1867
Sn BCT A5 6323 51.18 505.078 7029.12 13.9169 -1.0252
Zn HCP A3 5657 41.63 692.68 7322 10.5706 1.6653
50
Appendix B
Lattice stabilities
Data for Ag relative to FCC A1
(Dierence in Gibbs energy between each phase and this reference phase, according to
equation (2.33).)
LIQUID
11025.076 8.891021 T 1.034 10
20
T
7
(298.15 < T < 1234.93)
11508.141 9.301747 T 1.412 10
29
T
9
(1234.93 < T < 3000)
BCC A2
3400 1.05 T (298.15 < T < 3000)
HCP A3
300 + 0.3 T (298.15 < T < 3000)
Data for Cu relative to FCC A1
LIQUID
12964.736 9.511904 T 5.849 10
21
T
7
(298.15 < T < 1357.77)
13495.481 9.922344 T 3.642 10
29
T
9
(1357.77 < T < 3200)
BCC A2
4017 1.255 T (298.15 < T < 3200)
HCP A3
600 + 0.2 T (298.15 < T < 3200)
51
Appendix B. Lattice stabilities 52
Data for Bi relative to RHOMBO A7
LIQUID
11246.067 20.63651 T 5.955 10
19
T
7
(298.15 < T < 544.55)
11336.259 20.810418 T 1.661 10
25
T
9
(544.55 < T < 3000)
BCC A2
11297 13.9 T (544.55 < T < 3000)
BCT A5
4148.07 (544.55 < T < 3000)
FCC A1
9900 12.5 T (544.55 < T < 3000)
HCP A3
9900 11.8 T (544.55 < T < 3000)
TETRAGONAL A6
4148.07 (544.55 < T < 3000)
TET 1
4234 (544.55 < T < 3000)
Data for Sn relative to BCT A5
LIQUID
7103.092 14.087767 T + 1.47031 10
18
T
7
(100 < T < 505.08)
6971.587 13.814382 T + 1.2307 10
25
T
9
(505.08 < T < 3000)
FCC A1
4150 5.2 T (298.15 < T < 3000)
HCP A3
3900 4.4 T (298.15 < T < 3000)
RHOMBO A7
2035 (298.15 < T < 3000)
BCC A2
4400 6.0 T (298.15 < T < 3000)
Appendix B. Lattice stabilities 53
Data for Zn relative to HCP A3
LIQUID
7157.213 10.29299 T 3.5896 10
19
T
7
(298.15 < T < 692.68)
7450.168 10.737066 T 4.7051 10
26
T
9
(692.68 < T < 1700)
BCC A2
2886.96 2.5104 T (298.15 < T < 1700)
FCC A1
2969.82 1.56968 T (298.15 < T < 1700)
Appendix C
Thermodynamic data for the
Sn-Zn-Bi system
Bi-Sn
0
L
L
BiSn
= 490 + 0.97 T
1
L
L
BiSn
= 30 0.235 T
0
L
bct
BiSn
= 2120 1.44 T
1
L
bct
BiSn
= 3710.0
0
L
rho
BiSn
= 5760 + 11.834 T
Bi-Zn
0
L
L
BiZn
= 18265.09 8.6763 T
1
L
L
BiZn
= 6061.21 + 0.79581 T
2
L
L
BiZn
= 6422.6 + 11.72 T
3
L
L
BiZn
= 7227.44 9.2905 T
4
L
L
BiZn
= 21123.07 27.147 T
5
L
L
BiZn
= 20747.56 + 22.0176 T
6
L
L
BiZn
= 7600.36 + 13.16 T
0
L
hcp
BiZn
= 35000
0
L
rho
BiZn
= 10000
Sn-Zn
0
L
L
SnZn
= 12710 9.162 T
1
L
L
SnZn
= 5360 + 3.45 T
2
L
L
SnZn
= 835.0
0
L
bct
SnZn
= 9260
0
L
hcp
SnZn
= 40000
Bi-Sn-Zn
0
L
L
BiSnZn
= 76485.59 + 98.4963 T
1
L
L
BiSnZn
= 7048.7 9.25285 T
2
L
L
BiSnZn
= 265.89 11.10087 T
54
Appendix D
Binary Ag-Cu phase diagram
55
Appendix E
Binary Sn-Cu phase diagram
56
Appendix F
Binary Sn-Ag phase diagram
57
Appendix G
Binary Ag-Bi phase diagram
58
Appendix H
Binary Sn-Bi phase diagram
59
Appendix I
Binary Zn-Bi phase diagram
60
Appendix J
Binary Sn-Zn phase diagram
61
Appendix K
EDX Analysis
Solder type B (SnAg3.8Cu0.7)
Figure K.1: SE image Figure K.2: Tin
Figure K.3: Copper Figure K.4: Silver
62
Appendix K. EDX Analysis 63
Solder type C (SnAg3.3Bi3.82)
Figure K.5: SE image Figure K.6: Tin
Figure K.7: Silver Figure K.8: Bismuth
Solder type D (SnZn8Bi3)
Figure K.9: SE image Figure K.10: Bismuth
Figure K.11: Tin Figure K.12: Zinc
Solder type A (SnPb36Ag2)
Figure K.13: SE image Figure K.14: Lead
Figure K.15: Silver Figure K.16: Tin
Figure K.17: Copper
Appendix L
Shear deformation of gold grid
Figure L.1: Shear deformation at step 18 Figure L.2: Shear deformation at step 22
Figure L.3: Shear deformation at step 26 Figure L.4: Shear deformation at step 30
66
Appendix M
Hall of Fame
Figure M.1: Sample B7, fracture area in a failed solder joint after shear testing (BSE image)
67
Appendix M. Hall of Fame 68
Figure M.2: Sample D16, fracture area in a failed solder joint, caused by constant thermo-
mechanical loading (SE image)
Appendix M. Hall of Fame 69
Figure M.5: Sample AJ1, fracture area near interface, caused by constant thermo-mechanical
loading (BSE image)
Samenvatting
Een toenemend milieubewustzijn met betrekking tot loodhoudende soldeerlegeringen
heeft tot grote inspanningen geleid om een loodvrije soldeertechniek te ontwikkelen. Het
RIPOSTE project van de TU/e en Philips CFT heeft als doel een tool te ontwikkelen, die
de betrouwbaarheid van gesoldeerde miniatuurverbindingen in IC-packages voorspelt,
onderworpen aan (thermo-)mechanische belasting, waarbij rekening wordt gehouden
met de relevante microstructuur en chemische eigenschappen van het soldeer. Deze
scriptie maakt onderdeel uit van dit project en behandelt hoofdzakelijk afschuifbelasting,
kruip, faalgedrag en vervormingsgedrag van drie loodvrije soldeerlegeringen, namelijk
SnAg3.8Cu0.7 (SAC), SnAg3.3Bi3.82 en SnZn8Bi3 en SnPb36Ag2 als referentie.
Als gevolg van hoge homologe bedrijfstemperaturen van soldeerverbindingen in elektron-
ica, is de deformatie van deze verbindingen altijd een thermo-mechanisch proces. Dit
vereist kennis van zowel evenwicht- als niet-evenwicht thermodynamica, omdat tijdens
vervorming een constante ow van energie naar het systeem aanwezig is. Dit impliceert
dat er geenszins sprake is van een systeem in thermodynamisch evenwicht.
Als de temperatuur van een elektronisch systeem stijgt, dan zet de PCB (printed circuit
board) meer uit dan de component als gevolg van een verschil in uitzettingscoecient;
dit resulteert in afschuifspanningen in de soldeerverbinding. Teneinde de afschuifsterkte
en maximale afschuifrek te kunnen onderzoeken, is er een speciaal sample ontworpen,
zodat proeven in een miniatuur-trekbank uitgevoerd konden worden. Met behulp van
digitale beeldcorrelatie (DIC) kan het vervormingsgedrag op micronschaal worden geob-
serveerd en vergeleken worden met de data van de afschuifproeven.
Intermetallische verbindingen aan de soldeer-substraat interface blijken een belangrijke
rol te spelen in de deformatie van de soldeerverbinding bij SnAg3.8Cu0.7 en SnZn8Bi3.
Inhomogene en hoge locale rekken kunnen worden waargenomen aan de interface. De
deformatie van SnAg3.3Bi3.82 is echter veel homogener, wat leidt tot een superieure
afschuifsterkte, maar ook brosse breuk. SnAg3.8Cu0.7 heeft de grootste plastische de-
formatie, terwijl SnZn8Bi3 onderhevig is aan vreemde verouderingseecten.
Afschuifspanning in combinatie met hoge homologe temperaturen brengt ook kruip en
mogelijk coarsening teweeg. Een proefopstelling om kruip te meten is ontworpen om
samples in een droogstoof te testen. Het blijkt dat SnZn8Bi3 de hoogste kruipsnel-
heid heeft van de drie loodvrije soldeerverbindingen, gevolgd door SnAg3.3Bi3.82. SAC
is de meest kruip-bestendige soldeerlegering met ongeveer dezelfde kruipsnelheid als
70
Samenvatting 71
SnAg3.3Bi3.82 bij 100C, maar een betere weerstand tegen kruip bij 150C. Coarsen-
ing kon niet worden gedetecteerd, waardoor het niet mogelijk was om data hierover te
verzamelen.
ESEM- en lichtmicroscopie onthulden de faalmechanismen van de kruipsamples. In SAC-
soldeerverbindingen konden parallel aan de interface en loodrecht hierop (van de ene
interface naar de andere), langs grote kolonies worden waargenomen. Het uiteindelijke
falen van de verbinding werd veroorzaakt door scheuren parallel aan de interface door
het materiaal heen. Bij SnAg3.3Bi3.82 kon dit eveneens worden waargenomen, maar
falen gebeurde langs de intermetallische verbindingen aan de interface. Dit was ook in
SnZn8Bi3 te zien, alleen zijn de kolonies in dit materiaal kleiner en hebben een andere
vorm. Scheurvorming kon door de hele soldeerverbinding gezien worden in alle richtin-
gen. Een wat grilliger falen langs de koloniegrenzen en dwars door het materiaal heen
kon hier worden waargenomen en niet alleen langs de interface. EDX-analyse wees tevens
uit dat grote hoeveelheden koper naar het oppervlak van de verbinding diundeert en
daar brons vormt met zink.
Dit onderzoek heeft thermodynamische gegevens uit de literatuur opgeleverd, alsmede
thermo-mechanische gegevens voor het RIPOSTE project voor drie loodvrije soldeer-
verbindingen. De combinatie van deze data met data van andere onderzoeksgebieden
binnen dit project zal leiden tot een beter begrip van deze materialen en een goede
vervanger voor loodhoudende soldeerlegeringen.
Acknowledgements
Writing this thesis has been a job I couldnt have done without the help of many people.
Their help and support is greatly appreciated and will not be forgotten. I would like to
thank in particular:
My supervisor, Marc Geers, who made it possible to do this project and who has always
been very supportive and understanding during dicult personal circumstances.
My coach, Willem-Pier Vellinga, whos ideas and enthusiasm have been a great contri-
bution to my research.
Hans de Vries, whos expertise in the eld of lead-free soldering has been very useful for
my research.
Jo van Gerven, who has provided me with the shear samples.
Jo Sijben, who has made additional samples for me.
Toon Hoeben, who has often helped me with his knowledge of equipment and materials
and who has always made sure I got the materials and consumables I needed.
Toon van Gils, who has very accurately realized all the designs I made for my experi-
ments.
Pauline Schmit, who made it possible for me to use the sputter coater at the faculty of
Chemical Technology.
Jeroen Visser, who helped me with EDX-analysis and who provided me with a number
of EDX-map images.
Dick Landheer, who was always prepared to help me with phase diagrams or other
questions related to alloys and metals.
My colleagues and friends from student fellowship 3004, who always made sure that
the atmosphere in our room was good and who helped me with advice and creativity
when I needed it, especially Sebastiaan Boers.
My friends and family, who have stood by me and supported me throughout this project.
And last but certainly not least I would like to thank my girlfriend Manon, whos
patience and love helped me get through dicult periods and who supported me un-
conditionally.
72

Creep and Failure of Lead-Free Solder Alloys

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Creep and Failure of Lead-Free Solder Alloys

Загружено:

Авторское право:

Доступные форматы

Creep and Failure of

Lead-free Solder Alloys

Вам также может понравиться