Академический Документы
Профессиональный Документы
Культура Документы
The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
Journal of Materials Processing Technology 212 (2012) 471483
Contents lists available at SciVerse ScienceDirect
Journal of Materials Processing Technology
j our nal homepage: www. el sevi er . com/ l ocat e/ j mat pr ot ec
High temperature reliability of lead-free solder joints in a ip chip assembly
Emeka H. Amalu
, Ndy N. Ekere
Electronics Manufacturing Engineering Research Group, School of Engineering at Medway, University of Greenwich, ChathamMaritime, Kent ME4 4TB, UK
a r t i c l e i n f o
Article history:
Received 18 August 2011
Received in revised form10 October 2011
Accepted 13 October 2011
Available online 19 October 2011
Keywords:
Power electronics
High temperature electronics
Flip chip
Pb-free solder
Intermetallic compound
Microelectronics reliability
Fatigue failures
Solder joint
a b s t r a c t
The visco-plastic behaviour of solder joints of two models of a ip chip FC48D6.3C457DC mounted on a
printed circuit board (PCB) via SnAgCu solder is investigated using Anands model. While the bumps of one
of the models are realistic with 6 m thickness of intermetallic compound (IMC) at interconnects of solder
and bond pads, the other are made up of conventional bumps without IMC at these interconnects. The
solder bump proles were created using a combination of analytical method and construction geometry.
The assembled package on PCB was accelerated thermally cycled (ATC) using IEC standard 60749-25. It
was found in the result of the simulation that IMC does not only impact solder joint reliability but also
is a key factor of fatigue failure of solder joints. The IMC sandwiched between bond pad at chip side and
solder bulk is the most critical and its interface with solder bulk is the most vulnerable site of damage.
With reference to our results, it is proposed that non inclusion of IMC in solder joint models composed of
Sn-based solder and metalized copper substrate is one of the major causes of the discrepancy on solder
joint fatigue life predicted using nite element modelling and the one obtained through experimental
investigation.
2011 Elsevier B.V. All rights reserved.
1. Introduction
The packaging of high-reliability and high-performance, solid-
state power electronic modules remains a key challenge for the
micro-electronics packaging industry owing to frequent failure
of solder joints in these modules. The reliability of high tem-
perature electronic devices in operation is directly affected by
the static structural integrity of the joints of components used
in their assembly. Amalu et al. (2009) reported that miniaturisa-
tion is still a key design trend in electronics industry and Johnson
et al. (2004) also reported that high-temperature electronics use
in automotive systems will continue to grow. As electronic mod-
ules increasingly get smaller and nd applications in sectors where
operating ambient temperatures are high, assembly and packaging
engineers have come under pressure to mount components with
high solder joint integrity to meet customer expectation of reli-
able package performance at extreme high temperatures in the
eld. Power modules which operate at temperatures above tra-
ditional electronics limit of 125
C) are
termed high temperature electronics (HTE) and these assemblies
susceptibility to failures in the eld is indicated by their high
mortality rate. In Sandia report on rst high-temperature elec-
tronics products survey, Normann (2005) identied reliability as
FC48D6.3C457-DC FBT250
Fig. 1. Effect of CTE mismatch in a typical ip chip assembly showing: (a) after
thermal loading and (b) before thermal loading.
interface may degrade the reliability of the solder joints. Fig. 2
shows a solder bump with IMC sandwiched between solder bulk
and PCB. Furthermore, the cyclic nature of loading these modules
are subjected to while in the eld depreciates the service life of sol-
der joints of their components. HsuandChiang(2004) reportedthat
the reliability of solder joints depended strongly on the thermo-
mechanical behaviours of the materials and the geometry of the
solder joints such as standoff height and contact angle. In further-
ance of this statement, the authors believe that the solder joints
architecture and specically the prole of the solder bump joints
play a crucial role in the overall device reliability. It is pertinent to
note that component stand-off height and diameter of bond pads
(at both upper and lower interconnects of solder bump) determine
the joint geometryandare thus keyfatigue andstatic structural fac-
tors. Reichl et al. (2000) reported that thermo-mechanical aspects
of component andsystemreliabilitybecomemoreandmoreimpor-
tant with growing miniaturisation as the local physical parameters
and eld quantities show as increase in sensitivity due to non
homogeneities in local stresses, strains and temperature elds.
Generally, the impacts of static structural factors on reliability of
chip on board increases with miniaturisation and Amalu et al.
(2009) is of the view that the life expectancy of components and
systems reduce exponentially as the operating ambient tempera-
ture increases; adversely impacting long-term system reliability.
The physics of fatigue failure is usually by inducement of plastic
strain in the joint. Strain accumulation causes the joint to develop
stress which initiates crack. In their good work, Libres and Arroyo
(2010) put up hypothesis that observed bump cracks at the bulk
solder are the result of solder fatigue and extensive solder plastic
deformation during temperature cycling, coupled with a change in
the mechanical properties of the underll material at the exit side.
Furthermore, another good work by Yang and Ume (2008) cited
references suporting the claim that due to constraint of thermal
expansion, excessive strains/stresses might be induced and might
eventually initiate and propagate fatigue cracks in solder bumps.
Ladani and Razmi (2009) have reported that damage mechanism
Fig. 2. Solder bump containing IMC between the PCB and the solder bulk.
by fatigue cracks is most destructive in the presence of voids in the
solder joint.
Flip chip package is the key component which has facilitated
miniaturisation of HTEs. Li and Wang (2007) reported that ther-
mal fatigue failure, due to the fracture of solder joints which was
caused by the mismatch deformation, is frequently encountered
in ip chip assemblies. This situation could be more critical at
elevatedtemperature of operation. Althoughit is a commonknowl-
edge that the change in interconnection technology from wire
bonding to solder bumping improved high temperature perfor-
mance of FC assemblies fromthe electrical point of view, however
the extreme operating environment of HTE signicantly impacts
chip-level devices reliability by inducing untimely failures at the
board level and FC technology is not an exception. Xie et al. (2010)
believe that board level reliability concern over chip scale packages
(CSPs) is that ner pitch limits the size of solder ball attached on die
andstencil thickness usedinassembly whichleads tomuchsmaller
joint volume and standoff height while larger die-to-package ratio
typically means higher stress level caused by the CTE mismatch.
This situation may not be different for FC.
While Mallik et al. (2011) cited that the useful lifetime of elec-
tronic devices can decrease signicantly owing to the large thermal
stresses that occur especially at solder joint interfaces; Ridout and
Bailey (2007) reported that solder joints are often the cause of fail-
ure inelectronic devices. Inthe latters same paper, the authors also
stated that the intermetallic layer formed between the solder joint
and the copper pads is usually ignored, due partly to its size and
partly to the lack of material property data. Recently, IMC material
property data is available and current technology in computers has
enabledhighproductivitycomputing(HPC) requiredtohandlevery
many elements resulting fromthe inclusion of 6mIMC thickness
in the solder joint model. Thus, further study incorporating IMC
in FC solder joints is possible and needed. Such investigation will
provide further understanding of thermo-mechanical behaviour
of Pb-free solder joints in a FC assembly in HTE at elevated tem-
perature excursions characteristic of aerospace and automotive
applications as well as well-logging operations.
This investigationaims tostudyandprovide further information
on the reliability of typical and realistic Pb-free solder joints con-
taining IMC between the solder joints and the copper bond pads of
a FC48D6.3C457DC assembly; which operates in high temperature
and are subjected to strain deformations that culminate in stresses
and fatigue failure of the joints.
2. Flip chip assembly architecture
Fig. 3 shows FC architecture where Fig. 3(a) illustrates attach-
ment of the ball grid array (BGA) to the die in a daisy chain
layout, while Fig. 3(b) presents the dimensions of the BGA matrix
and Fig. 3(c) is a representation of a singular ball mount. Table 1
depicts the FC parameters whose congurations and regional plas-
tic strainare presentedinTable 2. The FCassembly is representedin
Figs. 4 and 5 is the magnied structure of one of the joints showing
force interactions when the FC is mounted on a PCB using reow
Author's personal copy
E.H. Amalu, N.N. Ekere / Journal of Materials Processing Technology 212 (2012) 471483 473
Fig. 3. Flip chip architecture showing: (a) bottomviewof ip chip plan; (b) dimensions of ip chip bump array; (c) truncated sphere solder ball before mounting and reow.
Table 2
Solder bump conguration and plastic strain.
Region (i =1,2,3) Thickness, t
i
(m) Volume, V
i
(m
3
) Regional plastic strain
i
V
i
/V
i
, i =1, 2, 3
Solder only Solder +IMC
1 6.0 1.04E+05 2.04 0
2 108.0 2.30E+06 0.887 0.102
3 6.0 1.04E+05 0.159 0
Bump volume, VT 2.51E+06
soldering process. Fig. 6 is a longitudinal cross-section of the joint
assembly.
2.1. Solder bump geometry derivation
Previously, many authors including Hsu and Chiang (2004) and
Ladani and Razmi (2009) determined solder ball geometry exclu-
sively using surface evolver software tool based on energy method.
In their good works, Chiang and Yuan (2001), Sidharth and Natekar
(2002) and Yeung and Lee (2003) used both surface evolver and
analytical method to predict solder bump geometry. In the model
creation in this study, the prole of the balls was determined using
a combination of truncated sphere theory, the force-balanced ana-
lytical method and ANSYS software design modeller (DM). The
function describing the shape of the bump was derived using the
structure in Fig. 5 a double truncated sphere model. Prior to
Fig. 4. Flip chip assembly.
Author's personal copy
474 E.H. Amalu, N.N. Ekere / Journal of Materials Processing Technology 212 (2012) 471483
Fig. 5. Force interactions in a ip chip solder joint.
mounting, the model was a truncated sphere (Fig. 3(c)) where
d
(z)
=h
0
=140m(refer to Table 1 and Fig. 3(c)).
If h
0
+y =2R
c
=, where h
0
, y, R
c
and are the ball height,
vertical distance in z direction, ball centre radius and diameter,
respectively, then:
at d
(z)
=
h
0
+y
2
, R
z
= R
_
h
0
+y
2
_
= R
c
=
89m. R
c
is also the diametric radius.
Fromtruncated sphere theory, Eq. (1):
h
0
= R
c
+
_
R
2
c
R
2
h
(1)
and
R
h
=
_
R
2
c
(h
0
R
c
)
2
(2)
where R
h
is the radius of copper pad at die bond pad intercon-
nect. Eq. (2) and parameter values in Table 1 were used to compute
the magnitude of bond pad diameter at interconnect as 146m.
Force-balanced analytical theory and DM were used sequentially
to compute the values of the two unknown parameters in Eq. (10)
height (h) and radius of the arc (R
arc
) after mounting and reow
soldering of the assembly. The algorithmis basedonmany assump-
tions which Chiang and Yuan (2001) reported. These assumptions
are:
Fig. 6. Longitudinal cross section of ip chip solder joint assembly.
(R
)
2
+PR[1 +(R
)
2
]
3]2
= 0 (6)
For equilibriumof forces acting at the plane surface at height, h:
1
n
W
Dic
= 2R
h
sin( 0
h
) +P
h
R
2
h
(7)
where W
Die
is the weight of the FC, n is the number of ball of the
die and 0
h
is the obtuse angle the surface tension makes with the
upper bond pad plane. The coefcient of W
Die
represents the share
weight component a ball experiences in supporting the FC die.
The solder volume is dened as:
v =
_
h
0
(R)
2
dz (8)
and in a similar study, Heinrich et al. (1996) approximated the
meridian dening the joints contour free surface, R
z
, to a circular
arc, R
arc
. Employing this approximation, the volume and unbal-
anced force are expressed as:
v =
_
h
0
[R
urc
(z)]
2
dz (9)
and F=
1
n
(W
Dic
)
R
h
2hR
urc
_
(R
0
+R
h
)
4R
2
urc
h
2
(R
0
+R
h
)
2
+h
2
h
2
_
for
_
h
0
140
R
urc
178
(10)
It is also assumed that a reowed FC on a PCB assembly is in
static equilibrium and thus F =0 in Eq. (10). An ideal congura-
tion of R
0
=R
h
was chosen for this investigation because variation
of the size of bond pads diameter of die in relation to the sub-
strate bond pad produces distinct ball geometry which is known
to impact solder joint reliability. Determination of values of R
arc
Author's personal copy
E.H. Amalu, N.N. Ekere / Journal of Materials Processing Technology 212 (2012) 471483 475
Table 3
Mechanical properties of assembly materials.
S/no Component Youngs modulus (GPa) C.T.E (ppm/
o
s
__
1]m
(11)
Other parameters associated with the model and which are not
dened in Table 4 such as d
p
, R, T and o are effective inelastic
Table 4
Anands constants for SnAgCu (Alamet al., 2009).
Constant Parameter Value Denition
C1 so (MPa) 39.09 Initial value of deformation resistance
C2 Q/R (1/Kelvin) 8930 Activation energy/boltzmanns constant
C3 A (1/s) 22,300 Pre-exponential factor
C4 (dimensionless) 6.0 Multiplier of stress
C4 m(dimensionless) 0.182 Strain rate sensitivity of stress
C6 h
0
(MPa) 3321.2 Hardening/softening constant
C7
S (MPa) 173.81 Coefcient for deformation resistance saturation value
C8 n (dimensionless) 0.018 Strain rate sensitivity of saturation value
C9 a (dimensionless) 1.82 Strain rate sensitivity of hardening/softening
Author's personal copy
476 E.H. Amalu, N.N. Ekere / Journal of Materials Processing Technology 212 (2012) 471483
Fig. 7. Plot of temperature prole of thermal load test condition used in ip chip test.
deformation rate, universal gas constant, absolute temperature
and effective Cauchy stress respectively. The s, a single scalar
internal variable, is the deformation resistance, whose evolution
is described by:
s =
_
sign
_
1
s
s
_
h
0
1
s
s
u
_
d
p
(12)
where s* is the saturation value of s associated with a given tem-
perature and strain rate pair and it is described by Eq. (13) and
S is
a coefcient for saturation.
s
=
S
_
d
p
/
c
(Q]R1)
_n
(13)
3.3. Loads and boundary conditions
Thermal load cycling between 38
C and 157
C and within
market environmental condition of ramp rate at 15
C/min
(IEC standard 6074925) was imposed on the simply supported
structure. Six cycles in 25 load step test condition was used in this
study. Fig. 7 shows the thermal cycle history where colour gradient
is used to identify temperature regions. The plotted thermal line
is red in colour at high temperature ambient (For interpretation of
the references to colour inthe text, the reader is referred to the web
version of the article.).
4. Results and discussion
4.1. Study on equivalent plastic strain
Inelastic deformation in solder joints induces plastic strain and
the phenomenon is captured using visco-plastic model. Excessive
plastic strain impacts the reliability of FC assembly. In this research
work, we investigate this impact in the two types of solder bumps
described in Section 3.1. The symmetry of the FC structure was
exploited and only a quarter need be studied. The depiction of the
plastic strain damage distribution on a critical corner solder bump
of the modelled quarter symmetry of the assembly is shown in
Fig. 8. Plastic strain damage on critical solder bump showing: (a) plastic strain distribution on solder +IMC bump; (b) plastic strain on solder only bump; (c) plastic strain on
Solder only region 1; (d) plastic strain on solder only region 3.
Author's personal copy
E.H. Amalu, N.N. Ekere / Journal of Materials Processing Technology 212 (2012) 471483 477
-45
5
55
105
155
-0.8
-0.3
0.2
0.7
1.2
1.7
2.2
0 5 10 15 20 25
T
e
m
p
e
r
a
t
u
r
e
(
o
C
)
P
l
a
s
c
s
t
r
a
i
n
,
Load step, L
Solder only Solder+IMC Thermal Cycle History
E
I
M
C
,
P
A
B2
B1
B
Fig. 9. Plot of equivalent plastic strain on solder bump against load step.
0.0
0.2
0.4
0.6
0.8
1.0
0 5 10 15 20 25
P
l
a
s
c
s
t
r
a
i
n
,
Load step, L
Solder only
Solder+IMC
(L, ) = (17, 0.102)
(L, ) = (17, 0.887)
Fig. 10. Plot of equivalent plastic strain for region 2 against load step.
Fig. 8. Comparison in Fig. 8(a and b) shows the effects of IMC on
plastic response of the solder joint. It can be seen that Fig. 8(a)
has a maximum damage at the interconnect between PCB IMC
and solder region 2. A comparative study in Fig. 8(c and d) also
shows that strain distribution exist along the longitudinal axis of
the bump. Graphical plots in Figs. 911 are the read out points of
plastic strain on solder bump and regions. In Fig. 9, the thermal
cycle history is plotted with dotted lines in colour gradient. It has
its ordinate axis as secondary axis and is superimposed on the plot
to aid explanation. These plots (Figs. 911) showthat SnAgCu alloy
solder joint experience volumetric plastic deformation. The rate of
change of this deformation(otherwise knownas strainrate) is quite
dependent on the composition of the bump and thus is different in
the two compositions under investigation.
The plot in Fig. 9 shows a signicant difference in plastic strain
response in the two different solder joint compositions. The trend
in solder only joint demonstrates ductile deformation on stabil-
isation. Its bumps record highest plastic strain rate in rst cycle
and higher in the second. It then stabilises in the third into steady
state for the rest of the cycle. A sharp increase in plastic strain is
observed at rst ramp down load fromstep two to three. Another
sharp increase is observed in second upper dwell region corre-
sponding to cycling fromstep ve to six with peak at step six. These
are recognised as critical load steps and determine the degree to
0.0
0.5
1.0
1.5
2.0
2.5
0 5 10 15 20 25
P
l
a
s
c
s
t
r
a
i
n
,
Load step, L
Solder only Region 1
Solder only Region 2
Solder only Region 3
Solder+IMC Region 2
(L, ) = (17,
(L, ) = (17, 0.887)
(L, ) = (17, 0.159)
(L, ) = (17, 0.102)
P
o
i
n
t
B
Fig. 11. Plot of equivalent plastic strain distribution along longitudinal sections of solder bump against load step.
Author's personal copy
478 E.H. Amalu, N.N. Ekere / Journal of Materials Processing Technology 212 (2012) 471483
which plastic strain is induced in the solder material. The phe-
nomenon observed in the rst two cycles could be regarded as
thermal shock. Solder materials under this condition could have
strain hardened and thus stabilised through stress relaxation at
the onset of third thermal cycle. Furthermore, these results indi-
cate that ramp down rates and upper dwell time of rst and second
cycles respectively are crucial parameters inaccelerated and highly
accelerated life testing (ALT and HALT) of solder joints. Since it is
a general knowledge that failure of solder joint is caused by the
average (till device failure) of accumulated plastic work per cycle,
it may be inappropriate therefore to use only one or two thermal
cycles in modelling the behaviour of solder joints as observed in
many works in the past. Bumps composed of solder +IMC do not
appear to experience thermal shock and although it attains stabil-
ity with fairly constant homogeneous amplitude of deformation in
the third load step, the difference between the second and the rest
cycle is only marginal. The deformation amplitude is higher in sol-
der +IMC than in the other. It can be inferred that fatigue failure
mechanism is predominant is this type of joint than in the solder
only.
Plot of equivalent plastic strain over the load step of region 2
for the two cases is depicted in Fig. 10. The deformation trend in
this region is the same as that of the bump (Fig. 9); however, def-
erence in magnitude exists between the two. Fig. 11 is the plot
of plastic strain over load step for all the regions of both com-
positions. In this plot, region 1 of solder only bump experienced
the highest strain while region 2 of solder +IMC bump recorded
the least value. Strain gradient is found across the regions of sol-
der only joint with maximum value at region 1 and minimum at
region 3.
4.1.1. Effect of IMC on solder joint plastic strain behaviour
Fig. 9 is the plot of read out data points of plastic strain over
the load step for the solder bump compositions. It is observed
that a higher value of plastic strain is recorded when the bump
is assumed to be composed of solder only. This is expected since
solder only bump contains more solder by volume and the higher
plastic strain value may be advantageous in cushioning the effect
of CTE mismatch and thermal shock during life operation. The plot
(solder only) can be seen to stabilise at the start of cycle three
shown as point A in the graph. An estimated difference in the plas-
tic strain (E
IMC,P
) between the two compositions was computed
at the middle (point B) of the remaining cycle because the strain
rate is fairly at steady state from cycle three to six. At point B,
the value of B1 and B2 are 2.038 and 0.0791, respectively. The
percentage variation (E
IMC,p
)% in the plastic strain value of the
two compositions using solder only composition as the reference
point is approximately 96.1%. It is a general knowledge that solder
bumps of a Sn-based solder and Au/Ni/Cu based substrate consist
of IMC; it is also known that critical parameters such as accu-
mulated plastic work density and strain energy density of solder
joints used in fatigue life models to compute service life of sol-
der joints of electronic assemblies depend on the plastic strain
of the joints. Thus, ignoring the use of IMC in model develop-
ment and analysis may have signicant effect on predicted life of
such solder joints. Muralidharan et al. (2010) in a similar thermal
cycling reliability study of solder joint reported an observed dis-
crepancy between experimental results and nite elememt model
(FEM) predictions. The discrepancies between the experimental
results and FEM predictions have been a concern and many sug-
gestions have been put forward. The authors propose, based on the
result of this investigation that one of the causes is the non use of
IMC in modelling, especially in high temperature operating devices
where IMC layer thickness is nontrivial and can grow up to 6m
in magnitude.
The contribution of E
IMC,P
may be derived numerically. The
underlying physics of thermo-mechanics involved in the plastic
strain deformation is governed by the equation:
v
1
0
= v
1
(14)
where v
T
is the change in volume of the solder bump during ther-
mal loading. v
T
is the total volume of the bump made up of three
volumetric regions. The and 0 are the volumetric expansion
coefcient and range of temperature cycling limits. From bump
geometry and conguration (Figs. 5 and 6):
v
1
= v
1
+v
2
+v
3
(15)
where 1, 2 and 3 designate the three regions of the bump, respec-
tively.
For solder only composition : v
1,(scldcr cnly)
= v
rcgicn1
+v
rcgicn2
+v
rcgicn3
(16)
For solder +IMC composition : v
1,(scldcr+lMC)
= v
lMC
+v
rcgicn2
+v
lMC
(17)
Thus, change in volume is:
v
1,(scldcr cnly)
= 0(v
rcgicn1
+v
rcgicn2
+v
rcgicn3
) (18)
v
1,(scldcr+lMC)
= 0(v
lMC
+v
rcgicn2
+v
lMC
) (19)
FromEq. (18):
v
1,(scldcr cnly)
v
rcgicn1
+v
rcgicn2
+v
rcgicn3
= 0 (20)
Similarly, Eq. (19) gives:
v
1,(scldcr+lMC)
v
lMC
+v
rcgicn2
+v
lMC
= 0 (21)
IMC in not visco-plastic and thus all changes are fromregion 2.
Eq. (21) therefore reduces to:
v
(rcgicn2)
v
lMC
+v
rcgicn2
+v
lMC
= 0 (22)
Eq. (20) may be re-written as:
v
(rcgicn1)
+v
(rcgicn2)
+v
(rcgicn3)
v
rcgicn1
+v
rcgicn2
+v
rcgicn3
= 0 (23)
Applying volume constancy condition,
v
lMC
+v
rcgicn2
+v
lMC
= v
rcgicn1
+v
rcgicn2
+v
rcgicn3
= v
1
(24)
Combination of Eqs. (22)(24) yields, E
IMC, p
:
v
(rcgicn2)
v
1
scldcr+lMC
/ =
v
(rcgicn2)
v
1
. .
lMC,p
+
v
(rcgicn1)
+v
(rcgicn3)
v
1
scldcr cnly
(25)
It can be proposed based on Eq. (25) that the smaller the IMC
thickness in a joint, the smaller the effects of its negligence in mod-
elling reliability studies of the FC solder joints. Plastic strains at the
regions may be called regional plastic strain,
i
. Their values were
read off from Fig. 11 at point B and presented in Table 2. Input of
Author's personal copy
E.H. Amalu, N.N. Ekere / Journal of Materials Processing Technology 212 (2012) 471483 479
these values in Eq. (25) may provide further understanding of the
expression. If volumetric strain relationship is expressed simply as:
v
i
=
i
v
i
i = 1, 2, 3 (26)
Then, Eq. (25) can be re-written as:
2
v
2
v
1
scldcr+lMC
/ =
2
v
2
v
1
+
1
v
1
+
3
v
3
v
1
scldcr cnly
(27)
2
v
2
v
1
scldcr+lMC
/ =
2
v
2
v
1
+
1
v
1
v
1
+
3
v
3
v
1
scldcr cnly
(28)
2
v
2j
scldcr+lMC
/ =
2
v
2j
+
1
v
1j
+
3
v
3j
scldcr cnly
(29)
where V
if(i=1,2,3)
is the regional volume fraction. Substituting
numerical values fromTable 2 into Eq. (29), E
IMC, p
:
0.09
scldcr+lMC
/ = 0.81
..
lMC,p
+0.09
scldcr cnly
(30)
It can be suggested based on Eq. (30) and as can be seen from
Figs. 911 that the presence of IMC has contributed in the change of
plastic deformation value of solder of region 2 fromapproximately
0.81 in solder only bump to 0.1 in solder +IMC. The volume fraction
of region 2 is equal in value in the two models. Thus, the decrease in
value is brought about by difference in strain magnitude. The infer-
ence is that IMC has increased boundary resistance against solder
deformation and prevents it from undergoing much deformation.
Consequently, stress at the interface boundaries between region 2
and IMC will increase and initiate boundary cracks that causes joint
failure. In this analysis, the percentage difference in plastic strain
values of the two compositions using solder only composition as
the reference is approximately 90%. The difference between this
value and the 96.1% computed earlier using Fig. 9 is about 6.0%.
4.2. Study on plastic work
A body accumulates plastic work (w
p
) when external loads
imposed on the body induce plastic strain in the body. Thermal
load on chip level assembly does the same on solder bumps used to
attachthechiponboard. Theaccumulatedplastic workis ameasure
of damage in the body and is thus referred to as a damage parame-
ter where a high value signies lowjoint reliability and vice versa.
Plastic work density (^W
p
) has been shown to be a better parame-
ter in estimating solder joint fatigue life. Numerically, the average
value of ^W
p
can be expressed as:
^W
p,uvg
=
n
i=1
^W
i
^v
i
n
i=1
^v
i
(31)
where, ^W
i
is plastic work accumulated in a cycle in element i and
^V
i
is the volume of the element. It is observedinFig. 12that solder
only joint experienced higher magnitude of nominal plastic work,
which increases continuously with load step for both joints until
failure occurs.
The change in value of plastic work density over load step
induces accumulated plastic work which is the cause of failure of
solder joints. Its plot over loadstepis showninFig. 13. It is observed
that the highest quantumleap damage occurs in load step two; the
impact of this loading could be described as thermal shock and
it was independent of joint composition. The two joints investi-
gated showdifferences in response to the load fromload step two
onwards with solder only composition developing higher inelastic
deformation. From the plot of percentage change in plastic work
over load step (Fig. 14), it is seen that the nonlinearity in the accu-
mulation of damage could give way to steady state plastic work
accumulation. This is an indication that solder joints experience
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
1 1 7 5 3
P
l
a
s
c
w
o
r
k
d
e
n
s
i
t
y
,
M
J
/
m
3
Thermal cycle number
Solder only
Solder + IMC
Fig. 12. Plot of Plastic work density against thermal cycle in solder bump.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
2 3 7
C
h
a
n
g
e
i
n
p
l
a
s
c
w
o
r
k
d
e
n
s
i
t
y
,
M
J
/
m
3
Thermal cycle number
Solder only
Solder + IMC
Fig. 13. Plot of change in plastic work density in solder bump against thermal cycle.
strain hardening and stress relaxation afterwards which could be
advantageous to joint reliability. This nding underscores the con-
cern needed in determining the real value of ^W
p,avg
. From this
view point, the use of one or two thermal cycle(s) may be inad-
equate in modelling studies. The authors recommend at least six
thermal cycles.
0.0
50.0
100.0
150.0
200.0
250.0
6 4 2 0
P
e
r
c
e
n
t
a
g
e
c
h
a
n
g
e
i
n
p
l
a
s
c
w
o
k
Thermal cycle number
Solder only
Solder + IMC
Fig. 14. Plot of percentage change in plastic work density in solder bump against
thermal cycle.
Author's personal copy
480 E.H. Amalu, N.N. Ekere / Journal of Materials Processing Technology 212 (2012) 471483
Fig. 15. Strain energy damage in solder regions showing: (a) strain energy on solder +IMC bump; (b) strain energy on solder only bump; (c) strain energy on solder +IMC
region 2; (d) strain energy on solder only region 2; (e) strain energy on solder +IMC region 3; (f) strain energy on solder only region 3.
4.3. Study on strain energy
The work done by external forces in producing deformation in
a body is stored within the body as strain energy; the strain energy
per unit volume is referred to as strain-energy density. The ther-
mal load imposed on the solder joints induces thermal stress and
the resultant deformation is stored in the body as strain energy.
Strain energy damages in solder at various regions of the two mod-
els are represented in Fig. 15. It can be seen from Fig. 15(a and
b) that IMC inuences solder joint strain energy. Fig. 15(c and d)
shows the contribution of IMC to strain energy development at
the interconnect of PCB IMC and solder region 2 for the two mod-
els. Its severe effect in the bump with IMC can be observed. A
comparison of strain energy of region 3 for the two bumps is
presented in Fig. 15(e and f) to show the character of damage
distribution of IMC and solder respectively. It is observed that
while solder damage is fairly evenly distributed; IMC damage con-
centrates at the periphery of the diametric plane of interconnect
a potential site for crack initiation. The plots of output data
of strain energy are shown in Figs. 16 and 17. Fig. 16(b) is the
section of Fig. 16(a) in dotted green colour ellipse. In the plot,
while solder only joints exhibit higher strain energy density with
-5.0E-02
3.0E-16
5.0E-02
1.0E-01
1.5E-01
2.0E-01
2.5E-01
3.0E-01
0 5 10 15 20 25
S
t
r
a
i
n
E
n
e
r
g
y
d
e
n
i
t
y
,
M
J
/
m
3
Load step, L
Solder only
Solder+IMC
-1.0E-03
1.1E-17
1.0E-03
2.0E-03
3.0E-03
4.0E-03
5.0E-03
-5 5 15 25
S
t
r
a
i
n
e
n
e
r
g
y
d
e
n
s
i
t
y
,
M
J
/
m
3
Load step, L
Solder only
Solder+IMC
a b
Fig. 16. Plot of strain energy density of solder bump against load step.
Author's personal copy
E.H. Amalu, N.N. Ekere / Journal of Materials Processing Technology 212 (2012) 471483 481
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0 5 10 15 20 25
S
t
r
a
i
n
e
n
e
r
g
y
d
e
n
s
i
t
y
,
M
J
/
m
3
load step, L
Solder only Region 1
Solder only Region 2
Solder only Region 3
Solder+IMC Region 1
Solder+IMC Region 2
Solder+IMC Region 3
-5.0E-03
0.0E+00
5.0E-03
1.0E-02
1.5E-02
2.0E-02
2.5E-02
7 12 17 22
S
t
r
a
i
n
e
n
e
r
g
y
d
e
n
s
i
t
y
(
M
J
/
m
3
)
Load step, L
Solder only Region 1
Solder only Region 2
Solder only Region 3
Solder+IMC Region 1
Solder+IMC Region 2
Solder+IMC Region 3
a
b
Fig. 17. Plot of regional strain energy density of solder against load step bump.
lower amplitude and localised resonances, solder +IMC bumps
can be seen to observe higher amplitude of oscillation. Although
it is observed that strain energy density value of solder only
bumps is higher than solder +IMC, however the higher amplitude
of oscillation of solder +IMC bumps coupled with their charac-
ter of damage (damage concentration at the periphery) indicates
that reliability of solder +IMC bumps may be more critical and
their failure will majorly be driven by fatigue crack mecha-
nism. It is the authors opinion that the higher strain energy
of solder only bump is due to the greater volume of solder it
contains which is advantageous in cushioning the effect of CTE
mismatch of the bonded joint materials. It may be important to
note that the damage in solder joint does not depend on strain
energy alone but also boundary stress which is contributed by the
presence of IMC. This opinion is supported by the non colour gra-
dient in Fig. 15(b) in comparison to Fig. 15(a).
The strain energy density for all the regions in the two
bumps is plotted in Fig. 17. Some localised resonances are
observed in the rst two cycles at load step three and six.
The instantaneous energy at work could be intrinsic energy
at rst plastic limit for step three and the other, intrinsic
energy at onset of strain hardening and stress relaxation. It
is also observed that solder only region 2 developed high-
est strain energy density within the rst cycle and solder
only region 1 was highest in the second. This phenomenon
indicates the existence (in solder only bump) of different crit-
ical plane volumetric strain energy densities which result in
shear force between planes of solder regions and also strain
energy density gradient along the longitudinal axis of the
bump.
Table 5
Effects of IMC on FC solder joint fatigue life.
Model W
W
ucc
)
1
(32)
Author's personal copy
482 E.H. Amalu, N.N. Ekere / Journal of Materials Processing Technology 212 (2012) 471483
Table 6
Deviations of predicted solder joint fatigue life frommeasured/experimental values.
Author Package Solder Model Experimental (Cycle) Predicted (Cycle) Deviation fromexperimental (%)
Lee and Lau (2004) PBGA SnAgCu CofnManson 4772 3310 31
Zhang et al. (2003) CSP SnPb Modied Darveaux 856 745 12.9
Rodgers et al. (2004) MicroBGA SnAgCu Schubert et al. 4123 2417 41.4
Lau and Lee (2004) PBGA SnPb Modied CofnManson 2826 2840 0.5
Yang and Ume (2008) FC SnPb Anand/Zahn 54 60 11.1
where N
f
is the number of repetitions or cycles to failure, the
constant W
and W
acc
for SnAgCu solder were generated from
Syed (2004) as 0.0015 and 0.5683, respectively. He determined W
and ^W
p
/W
acc
as 0.0014 and 1.32J/cm
3
, respectively. Similarly,
they reported that the value of W
and W
acc
. Consequently, the computation produced
W
Two load steps of thermal cycle history are critical and may
impact accelerated and highly accelerated life testing (ALT and
HALT) of solder joint.
Acknowledgements
The authors acknowledge the funds received from Petroleum
Technology Development Fund (PTDF), Nigeria, for the PGresearch
Author's personal copy
E.H. Amalu, N.N. Ekere / Journal of Materials Processing Technology 212 (2012) 471483 483
work reported in parts in this paper. We also acknowledge the sup-
port of EMERG staff/students and the laboratory staff of University
of Greenwich, especially Mr. Ian Cakebread and V.C. Cosgrave. In
addition, we wouldlike tothankDr. Sheufor his input andMr. Gal-
adima Aminu of PetroleumTechnology Development Fund, Abuja,
Nigeria, for his support.
References
Alam, M.O., Lu, H., Bailey, C., Chan, Y.C., 2009. Fracture mechanics analysis of solder
joint intermetallic compounds in shear test. Computational Materials Science
45, 576583.
Amalu, E.H., Ekere, N.N., Bhatti, R.S., 2009. High temperature electronics: R&D chal-
lenges and trends in materials, packaging and interconnection technology. In:
2nd International Conference on Adaptive Science & Technology, Accra, Ghana
2009, pp. 146153.
Amalu, E.H., Lau, W.K., Ekere, N.N., Bhatti, R.S., Mallik, S., Otiaba, K.C., Takyi, G., 2011.
A study of SnAgCu solder paste transfer efciency and effects of optimal reow
prole on solder deposits. Microelectronic Engineering 88, 16101617.
Chiang, K., Yuan, C., 2001. An overviewof solder bump shape prediction algorithms
with validations. IEEE Transactions on Advanced Packaging 24 (2), 158162.
Heinrich, S.M., Schaefer, M., Schroeder, S.A., Lee, P.S., 1996. Prediction of solder joint
geometries in array-type interconnects. ASME Journal of Electronic Packaging
118, 114121.
Hsu, Y., Chiang, K., 2004. Double layers wafer level chip scale package (DL-WLCSP)
solder joint shape prediction, reliability and parametric study. In: Inter Society
Conference on Thermal Phenomena, Las Vegas, pp. 310316.
Johnson, R.W., Evans, J.L., Jacobsen, P., Thompson, J.R., Christopher, M., 2004. The
changing automotive environment: high-temperature electronics. IEEE Trans-
actions on Electronics Packaging Manufacturing 27 (3), 164176.
Ladani, L.J., Razmi, J., 2009. Interaction effect of voids and standoff height on ther-
momechanical durability of bga solder joints. IEEE Transactions on Advanced
Packaging 9 (3), 348355.
Lau, D., Lee, S.W.R., 2004. Computational analyses on the effects of irregular condi-
tions during accelerated thermal tests on board level solder joint reliability. In:
Electronics Packaging Technology Conference, Singapore, pp. 516521.
Laurila, T., Vuorinen, V., Kivilahti, J., 2005. Interfacial reactions between lead-free
solders andcommonbasematerials. Materials ScienceandEngineering49(12),
160.
Lee, S.W.R., Lau, D., 2004. Computational model validation with experimental
data from temperature cycling tests of PBGA assemblies for the analysis of
board level solder joint reliability. In: 5th International Conference on Thermal
and Mechanical Simulation and Experiments in Micro-Electronics and Micro-
Systems, EuroSimE 2004, France, pp. 115120.
Li, X., Wang, Z., 2007. Thermo-fatigue life evaluation of SnAgCu solder joints in ip
chip assemblies. Journal of Materials Processing Technology 183, 612.
Libres, J., Arroyo, J.C., 2010. Investigation of bump crack and deformation on Pb-free
ip chip packages. In: In: Electronic Components and Technology Conference,
Las Vegas, NV.
Mallik, S., Ekere, N., Best, C., Bhatti, R., 2011. Investigation of thermal management
materials for automotive electronic control units. Applied Thermal Engineering
31 (23), 355362.
MGC, 2009. High performance FR-4 for multi-layered PWB high performance FR-4
for multi-layered PWB copper clad laminates. Retrieved 29th March, 2011, from
www.mgc.co.jp/eng/products/lm/btprint/lineup/fr4.html.
Muralidharan, G., Kurumaddali, K., Kercher, A.K., Leslie, S.G., 2010. Reliability of
Sn3.5Ag solder joints in high temperature packaging applications. In: Elec-
tronic Components and Technology Conference, Las Vegas, NV.
Nguyen, T.T., Lee, D., Kwak, J.B., Park, S., 2010. Effect of glue on reliability of ip
chip BGA packages under thermal cycling. Microelectronics Reliability 50 (7),
10001006.
Normann, R.A., 2005. First High-Temperature Electronics Products Survey 2005.
Sandia National Laboratories, California, pp. 143.
Rodgers, B., Punch, J., Ryan, C., Waldron, F., 2004. Experimental and numerical
evaluationof SnAgCuandSnPbsolders usingamicroBGAunder acceleratedtem-
perature cycling conditions. In: International Mechanical Engineering Congress
and Exposition (IMECE2004), Anaheim, CA, USA, pp. 17.
Reichl, H., Schubert, A., Topper, M., 2000. Reliability of ip chip and chip size pack-
ages. Microelectronics Reliability 40, 12431254.
Ridout, S., Bailey, C., 2007. Reviewof methods to predict solder joint reliability under
thermo-mechanical cycling. Fatigue & Fracture of Engineering Materials and
Structures 30 (5), 400412.
Sakuma, K., Sueoka, K., Kohara, S., Matsumoto, K., Noma, H., Aoki, T., Oyama, Y., Nishi-
waki, H., Andry, P.S., Tsang, C.K., Knickerbocker, J.U., Orii, Y., 2010. IMC bonding
for 3D interconnection. In: Electronic Components and Technology Conference
(ECTC), Las Vegas, NV, pp. 864871.
Sidharth, R.B., Natekar, D., 2002. Solder joint shape and standoff height prediction
and integration with FEA-based methodology for reliability evaluation. In: Elec-
tronic Components and Technology Conference, San Diego, CA, pp. 17391744.
Stoyanov, S., Bailey, C., Desmulliez, M., 2009. Optimisation modelling for thermal
fatigue reliability of lead-free interconnects in ne-pitch ip-chip packaging.
Soldering & Surface Mount Technology 21 (1), 1124.
Syed, A., 2004. Accumulated creep strain and energy density based thermal fatigue
life prediction models for SnAgCu solder joints. In: Electronic Components and
Technology Conference, Las Vegas, NV, USA, pp. 737746.
Xie, W., Lee, T., Liu, K.-K., Xue, J.-C., 2010. Pb-Free solder joint reliability of ne pitch
chip-scale packages. In: Electronic Components and Technology Conference, Las
Vegas, NV, pp. 15871590.
Xu, L., Pang, J.H.L., Prakash, K.H.P., Low, T.H., 2005. Isothermal and thermal cycling
aging on IMC growth rate in lead-free and lead-based solder interface. IEEE
Transactions on Electronics Packaging Manufacturing 28 (3), 408414.
Yang, J., Ume, C.I., 2008. Thermomechanical reliability study of ip-chip solder
bumps: using laser ultrasound technique and nite element method. In: Elec-
tronic Components and Technology Conference, pp. 611622.
Yeung, B.H., Lee, T.T., 2003. Evaluation and optimization of package processing and
design through solder joint prole prediction. IEEE Transactions on Advanced
Packaging 26 (1), 6874.
Zahn, B.A., 2002. Impact of ball via congurations on solder joint reliability.
In: Electronic Components and Technology Conference, San Diego, CA, pp.
14751483.
Zhang, L., Sitaraman, R., Patwardhan, V., Nguyen, L., Kelkar, N., 2003. Solder joint
reliability model with modied Darveauxs equations for the micro SMD wafer
level-chipscale package family. In: Electronic Components andTechnology Con-
ference, NewOrleans, LA, pp. 572577.
Zhang, Z., Park, S., Darbha, K., Master, R.N., 2010. Impact of usage conditions on
solder joint fatigue life. In: Electronic Components and Technology Conference,
Las Vegas, NV, pp. 1419.