[ 27 ]
Analysis of crack growth in solder joints
Keywords
Reliability, Lead-free soldering,
Fatigue, Intermetallics,
Crack propagation
Abstract
Understanding crack growth in
solder joints is important for
predicting the fatigue life of solder
interconnects. In this paper, crack
propagation in solder joints made
of two solder alloys,
62Sn/36Pb/2Ag (by weight), a
commonly used solder paste for
SMT reflow applications, and
96.5Sn/3.5Ag (by weight), a lead-
free solder alloy, was examined
during thermal cycling. Based on
these observations, the rate of
crack propagation was estimated.
Microstructural changes in the
solder during thermal cycling were
also studied.
Soldering & Surface Mount
Technology
11/3 [1999] 27–32
© MCB University Press
[ISSN 0954-0911]
The author wishes to thank
Drs J. Hu and E. Jih for helpful
discussions and assistance.
Received March 1999
Revised May 1999
Introduction
In electronic modules, solder joints are commonly used to
interconnect electronic components to the electronic cir-
cuitry on the substrate. Solder interconnects perform the
important functions of electrical connection, mechanical
support, and thermal dissipation.
In automotive applications, the electronic modules are
subject to extreme operating conditions, such as cyclic
temperature variations, vibration, and humidity, often
imposed simultaneously. Owing to the coefficient of ther-
mal expansion (CTE) mismatch between the component,
the solder, and the substrate material, when the module
experiences temperature changes in service, cyclic thermo-
mechanical stresses/strains are generated in the solder joint
(Frear, 1990, 1994; Lau, 1991). Further, significant creep
also occurs in the solder joint (Hwang, 1990), as the service
temperature is relatively high compared with the melting
temperature of the solder alloy. In other words, the homolo-
gous temperature is relatively high; for example, at 125°C,
the homologous temperature is 0.88 for 62Sn/36Pb/2Ag and
0.81 for 96.5Sn/3.5Ag. Eventually, due to the interactions of
thermomechanical fatigue and creep and microstructural
coarsening, cracks are initiated in the solder joint, and
failure occurs when the crack propagates through the solder
interconnect. It is therefore important to understand the rate
of crack propagation in the solder joint in order to be able to
predict the fatigue life of solder interconnects.
The purpose of this study is to examine the crack propa-
gation in solder joints, using two solder alloys:
62Sn/36Pb/2Ag and 96.5Sn/3.5Ag (all percentages by
weight). The 62Sn/36Pb/2Ag solder is widely used in SMT
(surface mount technology) reflow applications. The
96.5Sn/3.5Ag solder has recently been evaluated as a
potential lead-free solder for automotive electronic modules
(Chada et al., 1996, 1997; Chada et al., 1999; Igoshev et al.,
1998; Lloyd, 1995; Lu et al., 1999; Ren et al., 1997a,
1997b; Ren et al., 1997c; Shangguan et al., 1994; Shang-
guan and Achari, 1994, 1995; Shangguan and Gao, 1996,
1997). The melting temperature is 179-189°C for
62Sn/36Pb/2Ag and 221°C for 96.5Sn/3.5Ag. Other physi-
cal and mechanical properties of these solder alloys have
been reported in the literature (Hwang, 1992, 1996; ITRI,
1992; Klein Wassink, 1989; Lloyd, 1995; Shangguan and
Achari, 1995; Shangguan and Gao, 1996, 1997).
Experimental procedure
For the experimental work, test boards were built on FR4
printed wiring boards (PWB) substrates, using 100W , 2512
leadless ceramic chip resistors (approximately 6.3 · 3.2 ·
0.56mm). Even though the 2512 leadless ceramic chip
resistor has not been widely used on PWBs in the industry,
it was chosen for this experiment because its relatively large
size would generate large shear strains in the solder, thus
shortening the test time. 62Sn/36Pb/2Ag solder paste and
96.5Sn/3.5Ag solder paste, with rosin mildly activated
(RMA) no-clean flux, were used, and the solder paste was
deposited to the solder pad on the substrate, using a 10mil
(0.254mm) stencil. The boards were then reflowed through
an IR (infra-red) reflow oven, in air. The reflow peak tem-
perature was approximately 30°C above the liquidus tem-
perature of the solder, with approximately 60 seconds dwell
time above the solder liquidus temperature.
After reflow, several solder joints were cross-sectioned
and scanning electron microscope (SEM) pictures were
taken to examine the solder fillet shape and microstructure.
In order to be able to observe crack propagation in the
solder joint, selected samples from the assembled test
boards were cut lengthwise, using a diamond saw at a very
low speed, through the middle of the resistor (Figure 1),
providing a flat surface for the observation of crack initia-
tion and propagation. These samples, along with samples
that were not cut, were subjected to thermal shock testing
(–40 to 125°C, 18 minutes dwell time at each temperature,
air-to-air). Samples were taken out of the thermal shock
chamber at 250 cycle intervals for observation of solder
joint cracks under SEM, and the length of cracks (if present)
was measured.
Dongkai Shangguan
Visteon Automotive Systems, Ford Motor Co., Dearborn, Michigan, USA
Figure 1
Schematic illustration showing the sample being cut
through the middle (not to scale)
Chip Resistor
Termination
Solder
Figure 2
SEM pictures of the cross sections of solder joints after
reflow (1) 96.5Sn/3.5Ag; (2) 62Sn/36Pb/2Ag
(1)
(2)

[ 28 ]
Dongkai Shangguan
Analysis of crack growth in solder
joints
Soldering & Surface Mount
Technology
11/3 [1999] 27–32
To examine the microstructural changes in the solder, SEM
microprobe elemental mapping was performed on cross-
sectioned solder joints.
Results and discussion
1. General observations
Selected samples were cross-sectioned after reflow to
examine the formation of the solder joints. Figure 2 shows
the cross section of a 62Sn/36Pb/2Ag solder joint and a
96.5Sn/3.5Ag solder joint. It can be seen that the solder
joint shapes of the two alloys are in general very similar.
The average stand-off height after reflow is 20 m m for the
62Sn/36Pb/2Ag group and 17m m for the 96.5Sn/3.5Ag
group.
For thermal shock testing, samples that were cut in half
and those that were not cut were subjected to thermal shock
test simultaneously. Examination has indicated that the cut-
through samples and the un-cut samples have experienced
the same failure mode.
2. Crack growth
Figure 3 shows pictures of solder joints of 96.5Sn/3.5Ag for
a 2,512 resistor at 250, 500, 750, 1,000, 1,250, and 1,500
cycles. This is typical of the crack growth observed in all of
the samples.
(1)
(2)
Figure 3
SEM pictures of the 96.5Sn/3.5Ag solder joints for a ceramic chip resistor at: (1) 250; (2) 500; (3) 750; (4) 1,000;
(5) 1,250 and (6) 1,500 cycles. (Scale bar: 100m m)
(continued)