mechanical shock durability studies of sn–ag–cu–ni … · the design of the test vehicle,...
TRANSCRIPT
62
Transactions of The Japan Institute of Electronics Packaging Vol. 2, No. 1, 2009
Mechanical Shock Durability Studies of Sn–Ag–Cu–Ni BGA Solder
Joints on Electroless Ni–P/Au Surface FinishFumiyoshi Kawashiro*, Hajime Yanase*, Masato Ujiie*, Takaki Etou* and Hiroshi Okada**
*NEC Electronics Corporation, 1120 Shimokuzawa, Sagamihara, Kanagawa 229-1198, Japan
**Senju Metal Industry Co.,Ltd, 1, Matsuyama, Mohka, Tochigi 321-4346, Japan
(Received July 30, 2009; accepted November 24, 2009)
Abstract
In this study, 60 solder compositions were examined in an effort to improve the integrity against impact loads of the
solder joints on an electroless Ni–P/Au surface finish. The Ag, Cu, and Ni contents in the Sn-based solder varied from
0 to 4.5 wt%, 0 to 2.0 wt%, and 0 to 0.05 wt%, respectively. Impact shear tests were performed to investigate solder joint
integrity after solder ball reflowing, after reflow soldering twice more, after storage at room temperature for 168 hours,
and after storage at 150°C for 1,000 hours. According to the results, the Ag content should be as low as possible, the Cu
content should be from 0.5 to 0.7 wt%, and the Ni content should be as high as possible. The Ag and Ni contents should
be determined in consideration of the wettability and the board-level reliability of the solder joints.
Keywords: Impact Shear Test, Electroless Ni–P Au, Sn–Ag–Cu–Ni Solder, Automotive, Flip-chip BGA
1. Introduction1.1 Trends
With the miniaturization and very large scale integration
of circuits in Si devices, the latest semiconductors for auto-
motive products such as car navigation systems require
fine-pitch flip-chip BGA packages (FCBGA) whose BGA
solder joints consist of solder balls and electroless Ni–P/
Au (ENIG) surface finish pads, as shown in Figure 1. Lead-
free solders like SnAg3Cu0.5, which are very brittle and
much weaker than tin-lead ones against impact loads, are
used for the BGA solder joints of the package.[1–10]
These solder joints tend to come off easily during handling
or transportation. To resolve this problem, the improve-
ment of the solder joints against impact forces and board-
level thermal cyclic properties is required.
1.2 Impact loads During TransportationTo determine the direction of the impact loads during
transportation, FCBGA packages were packed into a tray
container on a transporter, and the gravitational accelera-
tions of the packages were monitored. These accelerations
are plotted in Figure 2. The graph shows that the shock
that came from the horizontal direction (the X-axis) was
higher than that from the vertical direction (the Z-axis).
Therefore, impact shear testing rather than pull testing
was used to measure the impact properties of solder joints.
2. Experiments2.1 Test vehicle
The design of the test vehicle, which was 27 mm × 27
mm and had 720 BGA pads with a 0.8 mm pitch, is shown
Fig. 1 Structure of FCBGA and its pad surface finish.
Fig. 2 Impact accelerations on a package in the tray con-tainer during transportation.
63
in Figure 3. The BGA pad surface finish consisted of
ENIG, with Ni plating and Au plating from 1.6 to 2.0 μm
and from 0.07 to 0.10 μm, respectively. The content of P in
the Ni plating varied from 9 to 11 wt%. After feeding water-
soluble flux to the substrate pads, 0.500 mm diameter
solder balls were attached to the black solder pads shown
in Figure 3, which had a pad opening diameter of 0.400
mm. Next, the solder balls were reflowed at 250°C as
shown in Figure 4 and the flux residue was removed using
de-ionized water.
2.2 Compositions of solder ballIn this study, 57 solder alloys were examined in an effort
to improve the integrity of the solder joints against impact
loads. The concentration of Ag, Cu, and Ni in the Sn based
solder varied from 0 to 4.5 wt%, 0 to 2.0 wt%, and 0 to 0.05
wt%, respectively, as shown in Table 1. In addition, three
solder alloys consisting of SnPb37, SnCu0.7Ni0.05, and
SnAg2.3Ni0.08 were examined.
2.3 Impact shear testingAn impact shear tester (Dage 4000-HS) was used to
investigate the impact properties of the solder joints, as
shown in Figure 5.[11] The speed of impact testing ranged
from 10 mm/s to 4,000 mm/s. Both the shear load and the
fracture energy of solder joint were plotted. Figure 6
shows the relationship between the shear load and the
fracture energy of the solder joints. According to our pre-
vious studies, the fracture energy of solder joints can
describe their integrity more clearly than the shear
load.[12–13] Therefore, the fracture energy of the solder
joints was also used in this study to assess the solder joint
integrity. Impact shear tests were performed immediately
after solder ball reflowing, after reflow soldering twice
more, after storage at room temperature for 168 hours, and
after storage at 150°C for 1,000 hours.
Fig. 3 27 mm × 27 mm/ 0.8 mm Pitch/ FCBGA.
Fig. 4 Reflow profile.
Table 1 Compositions of Sn based solder balls.
Ag 0
Cu 0 0.1 0.5 0.7
Ni 0 0.0 0.0 0.0
Ag 0.5
Cu 0 0.1 0.5 0.7 1
Ni 0 0.015 0.05 0 0.015 0.05 0 0 0.015 0.05 0
Ag 1
Cu 0 0.1 0.5 0.7 1 2
Ni 0 0.015 0.05 0 0.015 0.05 0 0 0.015 0.05 0 0
Ag 2
Cu 0 0.1 0.5 0.7
Ni 0 0.015 0.05 0 0.015 0.05 0 0 0.015 0.05
Ag 3
Cu 0 0.1 0.5 0.7 2.0
Ni 0 0.015 0.05 0 0.015 0.05 0 0 0.015 0.05 0.0
Ag 3.5 4 4.5
Cu 0 0 0.1 0.5 0.7 0 0.1 0.5 0.7
Ni 0 0 0 0 0 0 0 0 0
Fig. 5 Impact shear tester.
Fig. 6 Relationship between shear load and fracture energyof solder joint.
Kawashiro et al.: Mechanical Shock Durability Studies of Sn–Ag–Cu–Ni (2/7)
64
Transactions of The Japan Institute of Electronics Packaging Vol. 2, No. 1, 2009
3. Results and Discussion3.1 Shear speed
The minimums of the fracture energies of the Sn–Ag–Cu
solder joints are plotted as functions of their Cu content in
Figure 7. Tests were conducted on 30 solder joint points
under each condition. The speeds of the impact shear tests
were 10, 100, 1,000 and 4,000 mm/s. Figure 8 represents
the rates of the interfacial fracture modes at each shear
speed. The results in Figure 7 show that the sensitive
shear speeds were 10 mm/s and 100 mm/s. Furthermore,
Figure 8 shows that the fractures occurred mainly in the
solder bulk at the speed of 10 mm/s, and almost all the
fractures occurred in the intermetallic compound layer at
the speeds of 1,000 mm/s and 4,000 mm/s. Therefore, the
most sensitive shear speed to distinguish the impact prop-
erty seems to be 100 mm/s.
3.2 Cu content effect on impact shear propertyFrom Figure 7, the fracture energies of SnAgxCu0.5 and
SnAgxCu0.7 solder joints were higher than those of
SnAgxCu0 , SnAgxCu0.1, where x varied from 0 to 4.5 wt%.
It seems that these differences were caused by changes in
the intermetallic compound (IMC) composition. For exa-
mple, the SEM images in Figure 13 show cross-sections of
the solder joints of SnAg0Cuy alloys on ENIG pads and
fractures due to the impact shear tests, where y varied
from 0 to 0.7 wt%. The compositions of the IMC layers
were identified using an EPMA (electron probe micro-
analyzer) and EDX (energy dispersive X-ray fluorescence
spectroscopy). Table 2 shows the typical composition of
each observed IMC layer. The SnAg0Cu0 alloy on ENIG
pads formed a Ni3Sn4 IMC as the first IMC layer. There
were Ni3(Sn,P)2 IMC and P-rich Ni layers between the first
IMC layer and Ni–P surface finish. A fracture due to an
impact shear test occurred between Ni3(Sn,P)2 and
Ni3Sn4, where there were some voids. The SnAg0Cu0.1
alloy formed a (Ni,Cu)3Sn4 IMC as the first IMC layer.
There were Ni3(Sn,P)2 IMC and P-rich Ni layers between
the first IMC layer and the Ni–P surface finish. A fracture
Fig. 7 Fracture energies of SnAgCu solder joint on ENIG.
Fig. 8 Interfacial fracture rate of SnAgCu solder joint onENIG.
Table 2 Typical composition of IMC layers.
Layer Observed composition [at%]
Ni-P layer Cu0.0 Ni90.0 Sn0.0 P10.0
P-rich Ni layer Cu0.0 Ni73.8 Sn1.9 P24.3
Ni3(Sn,P)2 Cu0.0 Ni58.2 Sn25.6 P16.2
Ni3 Sn4 Cu0.0 Ni45.9 Sn54.1 P0.0
(Ni,Cu)3 Sn4 Cu6.5 Ni34.5 Sn59.3 P0.0
(Cu,Ni)6 Sn5 Cu37.8 Ni18.6 Sn43.6 P0.0
Fig. 9 Fracture energies of SnAgCu solder joints as a func-tion of Ag content.
65
also occurred between the Ni3(Sn,P)2 and Ni3Sn4. Mean-
while, the SnAg0Cu0.5 and SnAg0Cu0.7 alloys formed a
(Cu,Ni)6Sn5 IMC as the first IMC layer. The Ni3(Sn,P)2
IMC layer was too thin to observe. Fractures occurred in
the solder bulk. According to these results, the Cu content
should be from 0.5 to 0.7 wt%, and a (Cu,Ni)6Sn5 IMC pro-
vides good durability against impact loads.
3.3 Ag content effect on impact shear propertyFigure 9 shows the averages of the fracture energy of
the Sn–Ag–Cu solder joints as a function of Ag content.
The averages of the fracture energy of the Sn–Ag–Cu solder
joints tended to decrease as the Ag content increased.
Also, the Cu content seemed to have little effect on the
fracture energies. Therefore, the microstructures of the
SnAgxCu0 alloys were investigated, where x varied from 0
to 3 wt%. Figure 10 shows SEM images of the microstruc-
tures of the SnAgxCu0 alloys, where the Ag content affects
the Ag3Sn intermetallic compound dispersion and Sn grain
size. The relationship between the fracture energy of the
SnAgxCu0 solder joints and Vickers hardness is plotted in
Figure 11. The fracture energy of the solder joints
decreased as the Vickers hardness of the solder alloy
increased. For the high Ag content solder alloy SnAg3, the
microstructure of the solder alloy has a finely dispersed
Ag3Sn IMC and a minute Sn grain size, which make the
solder hard and the solder joint brittle. Therefore, the Ag
content should be as low as possible to obtain good dura-
bility against impact loads.
3.4 Ni content effect on impact shear propertyThe effect of the Ni content on the fracture energy of the
Sn–Ag–Cu–Ni alloys is shown in Figure 12. It can be seen
that the fracture energies of the SnAgxCu0.7Ni0.05 solder
joints were higher than those of the SnAgxCuyNiz solder
joints, where x, y and z varied from 0 to 4.5, 0 to 0.7 and 0
to 0.05 wt%, respectively. It seems that these differences
were caused by changes in the intermetallic compound
(IMC) composition. For example, the SEM images in
Figure 13 show cross sections of the solder joints of the
SnAg3Cu0.7NiZ alloys on ENIG pads and the fractures due
to the impact shear tests. The compositions of the IMC lay-
ers were identified using an EPMA and EDX and are shown
in Table 2. The fracture energies of the SnAg3Cu0.7NiZalloys increased as the Ni content increased. The
SnAg3Cu0.7Ni0 alloy on ENIG pads formed a (Cu,Ni)6Sn5
IMC as the first IMC layer. There was a P-rich Ni layer
between the first IMC layer and the Ni–P surface finish. A
fracture due to an impact shear test occurred at the first
IMC layer. The SnAg3Cu0.7Ni0.05 alloy also formed a
(Cu,Ni)6Sn5 IMC as the first IMC layer. The P-rich Ni
layer was too thin to observe, which implies that the
consumption rate of Ni from the Ni–P surface finish
decreased. A fracture occurred between the first IMC
layer and the Ni–P surface. Based on these results, the Ni
content should be as high as possible, and a (Cu,Ni)6Sn5
IMC provides good durability against impact loads.
Fig. 10 Microstructures of SnAgxCu0 alloys (x: 1, 0, 3).
Fig. 11 Microstructures of SnAgxCu0 alloys (x: 1, 0, 3). Fig. 12 Fracture energies of SnAgCuNi solder joints as afunction of Ni content.
Kawashiro et al.: Mechanical Shock Durability Studies of Sn–Ag–Cu–Ni (4/7)
66
Transactions of The Japan Institute of Electronics Packaging Vol. 2, No. 1, 2009
3.5 Effects of pre-conditionsThe effects of pre-conditions on the fracture energy of
Sn–Ag–Cu solder joints are plotted in Figure 14 of the data
for the Sn–Ag–Cu–Ni and common solder alloys like
SnPb37, SnCu0.7Ni0.05, SnAg1Cu0.1Ni0.015 and SnAg2.3Ni0.08
are plotted in Figure 15. The pre-conditions written as As
reflowed and Reflow × 2 indicate that the impact tests were
performed immediately after solder ball reflowing, and
after reflow soldering twice more, respectively. RT 168 h
and HT 1000 h mean that the tests were performed after
storage at room temperature (RT) for 168 hours and after
storage at a high temperature (HT) of 150°C for 1,000 hours,
respectively. The fracture energies of the SnAg3Cu0.5 alloys
were too low compared with those of SnCu0.7Ni0.05,
SnAg1Cu0.5, and SnAg2Cu0.7Ni0.05. Although the fracture
energy of SnAg1Cu0.15Ni0.015 was very high after reflow
soldering, it became worse after the pre-conditions were
applied. The fracture energy of the SnPb37 alloy was also
high after reflow soldering, and it also became worse after
storage at 150°C for 1,000 hours. The fracture energies of
the SnAg2.3Ni0.08 alloys were also very low. These results
indicate that such solder alloys as SnCu0.7Ni0.05,
SnAg1Cu0.7Ni0.05, and SnAg2Cu0.7Ni0.05, with Ag content
not more than 3 wt%, Cu content from 0.5 to 0.7 wt%, and
Ni content not less than 0.05 wt%, seem to form solder
joints that are robust against impact loads under the vari-
ous pre-conditions.
3.6 Melting point of solder and board level reliabilityFCBGA packages for automotive products have a thick
substrate and a Cu-based lid. Therefore, the reflow temper-
ature during board-level assembly is lower than that of
packages for portable products. Solder balls whose melt-
Fig. 13 IMCs of various solder alloys (left side: SnAg0Cuy (y: 0, 0,1, 0.5, 0.7), right side: SnAg3.0Cu0.7Niz (z: 0, 0.05)].
Fig. 14 Effects of pre-conditions on the fracture energies ofSnAgCu solder joints.
Fig. 15 Effects of pre-conditions on the fracture energies ofvarious solder joints.
67
ing points are higher than the reflow temperature can
cause solder joint failures such as open joints during
board-level assembly. Therefore, the melting point of the
solder, including the liquidus temperature, should be as
low as possible. Liquidus temperatures of solders are
shown in Figures 16 and 17. Preferred compositions
should be within 2–3 wt% for Ag, 0.5–0.8 wt% for Cu, and
not more than 0.05 wt% for Ni. Moreover, the board-level
thermal cyclic property should be taken into consider-
ation.[14]
4. ConclusionsThe mechanical shock durabilities of Sn-based solder
alloys, with Ag, Cu and Ni contents varying from 0 to 4.5
wt%, 0 to 2.0 wt%, and 0 to 0.05 wt% respectively, on an elec-
troless Ni–P/Au surface finish, were investigated by
impact shear tests at the speed of 100 mm/s under various
pre-conditions. The results show that the following points
are important for solder alloys to perform well:
(1) The Ag content should not be more than 3.0 wt%,
and solder alloys with 0 wt% Ag content will be
stronger than the others against mechanicals
shocks.
(2) The Cu content should be from 0.5 to 0.7 wt%.
(3) The Ni content should not be less than 0.05 wt%.
(4) Such solder alloys as SnCu0.7Ni0.05, SnAg1Cu0.7Ni0.05,
and SnAg2Cu0.7Ni0.05 seem to form solder joints
that are robust against mechanical shocks under
various pre-conditions.
(5) The Ag and Ni contents should be determined
with consideration for the wettability and the
board-level reliability of solder joints.
References
[1] M. Amagai., Y. Toyoda, et al., “High Drop Test Reli-
ability: Lead-free Solders”, Proc. 54th Electronic
Components and Technology Conf., pp. 1304–1309.
[2] D. Cavasin, M. Anani, et al., “Board Level Character-
ization of Pb-free Sn–3.5Ag Versus Eutectic Pb–Sn
BGA Solder Joint Reliability”, Proc. 57th Electronic
Components and Technology Conf., pp. 129–135.
[3] W. H. Zhu, L. Xu, et al., “Drop Reliability Study of
PBGA Assemblies with SAC305, SAC105 and
SAC105–Ni Solder Ball on Cu–OSP and ENIG Sur-
face Finish”, Proc. 58th Electronic Components and
Technology Conf., pp. 1667–1672.
[4] A. T. Valotaa, A. Losavioa, et al., “High Speed Pull
Test Characterization of BGA solder joints”, Proc.
7th. Int. Conf. on Thermal, Mechanical and Mult-
iphysics Simulation and Experiments in Micro-Elec-
tronics and Micro-Systems, pp. 1–6.
[5] H. Kim, M. Zhang, et al., “Improved Drop Reliability
Performance with Lead Free Solders of Low Ag Con-
tent and Their Failure Modes”, Proc. 57th Electronic
Components and Technology Conf., pp. 962–967.
[6] X. J. Zhao, J. F. J. M. Caers, et al., “Improvement of
Mechanical Impact Resistance of BGA Packages
with Pb-free Solder Bumps”, Proc. 2006 Electronics
Packaging Technology Conf., pp. 174–178.
[7] X. K. Zhu, B. Qi, et al., “Mechanical Test and Analysis
on Reliability of Lead-free BGA Assembly”, Proc
2005 Electronics Packaging Technology Conf., pp. 1–
5.
[8] I. Shohji, H. Watanabe, et al., “Impact Properties of
Lead-Free Sn–Ag–Cu–Ni–Ge Solder Joint with Cu
Electrode”, Materials Transactions, Vol. 49, No. 7
(2008), pp. 1513–1517.
[9] I. Shohji, T. Osawa, et al., “Impact Properties of Sn–
0.75Cu Lead-free Solder Ball Joint”, Key Engineering
Materials Vols., 385–387 (2008), pp. 745–748.
[10] A. Syed, J. Scanlan, et al., “Impact of Package
Design and Materials on Reliability for Temperature
Cycling, Bed, and Drop Loading Conditions”, Proc.
Fig. 16 Liquidus temperatures of SnAgCu alloys.
Fig. 17 Ni content effect on liquidus point of SnAg3Cu0.5
alloys.
Kawashiro et al.: Mechanical Shock Durability Studies of Sn–Ag–Cu–Ni (6/7)
68
Transactions of The Japan Institute of Electronics Packaging Vol. 2, No. 1, 2009
58th Electronic Components and Technology Conf.,
pp. 1453–1461.
[11] K. Newman, “BGA Brittle Fracture – Alternative Sol-
der Joint Integrity Test Methods”, Proc. 55th Elec-
tronic Components and Technology Conf., pp. 170–
179.
[12] F. Kawashiro, Y. Kurita, et al., “Development of BGA
Attach Process with High Density SiP Technology
“SMAFTI””, 13th Symposium on Micro jointing and
Assembly Technology in Electronics (MATE 2007),
Feb. 2007, pp. 49–54.
[13] F. Kawashiro, M. Ujiie, et al., “Reliability Studies of
Sn–Ag–Cu BGA Solder Joints on Ni/Cu/Au Surface
Finish for SMAFTI Packaging Technology”, Proc.
58th Electronic Components and Technology Conf.,
pp. 283–289.
[14] S. K. Kang, P. Lauro, et al., “Evaluation of Thermal
Fatigue Life and Failure Mechanisms of Sn–Ag–Cu
Solder Joints with Reduced Ag Contents”, Proc. 54th
Electronic Components and Technology Conf., pp.
661–667.