Solid-State Reactions Between Ni and Tin-Silver-Copper Solders with
Different Cu Concentrations
W. C. Luo, C. E. Ho, J. Y. Tsai, Y. L. Lin, and C. R. Kao*
Department of Chemical & Materials Engineering
National Central University
Jhongli City, Taiwan
(*E-mail: kaocr@hotmail.com
Phone/Fax: +886-3-4227382)
Abstract
It had been reported in the literature that, during reflow of SnAgCu solders over the
Ni-bearing surface finishes, a slight variation in Cu concentration produced completely
different reaction products. With increasing Cu concentration, the reaction product switched
from (Ni1-xCux)3Sn4 to (Cu1-yNiy)6Sn5+(Ni1-xCux)3Sn4, then to (Cu1-yNiy)6Sn5. In this study,
we extended our earlier effort to investigate the effect of a small perturbation in the Cu
concentration on the solid-state aging reaction between the SnAgCu solders and Ni.
Specifically, five Sn3.9AgxCu solders (x=0.2, 0.4, 0.5, 0.6, and 0.8) were reacted with Ni at
180 oC. It was found that the strong sensitivity to Cu composition disappeared after
solid-state aging at high temperatures for a sufficient period of time. For all the Cu
concentrations studied, the same type of intermetallic compounds formed at the interface
after aging. A layer of (Cu1-yNiy)6Sn5 over a layer of (Ni1-xCux)3Sn4 was found at the interface.
This study showed that the initial difference in the intermetallic compounds right after reflow
can be aged out at high temperatures. The growth mechanisms for (Cu1-yNiy)6Sn5 and
(Ni1-xCux)3Sn4 are different, and are pointed out in this study.
Keywords: SnAgCu, interfacial reaction, surface finish, lead-free solder.
1. Introduction
The PbSn solder is an easily accessible material with very good mechanical properties,
however the Pb toxicity leads to the banning of Pb in solders. Two recent European Union
directives, RoHS ((Directive on the restriction of the use of certain hazardous substances in
electrical and electronic equipment) and WEEE (Directive on Waste Electrical and Electronic
Equipment), require new electrical and electronic equipments produced after 1st July 2006 to
be lead-free. Several review papers had been published on the status of Pb-free solders [1-4].
The SnAgCu solder family is regarded as one the most promising lead-free replacements for
the PbSn solders. The SnAgCu solder family has compositions (wt. %) near the Sn-Ag-Cu
ternary eutectic at Sn(3.50.3)Ag(0.90.2)Cu [5], and the Sn3.9Ag0.6Cu solder has been
recommended by a major industrial consortium [6].
Nickel is a very common metal surface finish used in electronic packages. The function
of Ni is to serve as a solderable diffusion barrier layer to prevent the rapid reaction between
the solder and the Cu layer below. The reactions between Ni and solders had been studied
before [7-26]. It had been reported that the reaction of Ni with liquid SnAgCu or SnCu
solders was very sensitive to the Cu concentration in the solders [12-15]. At low Cu
concentrations (≦ 0.2 wt.%), only a continuous (Ni1-xCux)3Sn4 layer formed at the interface.
When the Cu concentration increased to 0.4 wt.%, a continuous (Ni1-xCux)3Sn4 layer and a
small amount of discontinuous (Cu1-yNiy)6Sn5 particles formed at the interface. When the Cu
concentration increased to 0.5 wt.%, the amount of (Cu1-yNiy)6Sn5 increased and
(Cu1-yNiy)6Sn5 became a continuous layer. Beneath this (Cu1-yNiy)6Sn5 layer was a very thin
but continuous layer of (Ni1-xCux)3Sn4. At higher Cu concentrations (0.6-3.0 wt.%),
(Ni1-xCux)3Sn4 disappeared, and only (Cu1-yNiy)6Sn5 was present. These studies show that a
precise control over the Cu concentration in solders is needed to produce consistent results
during reflow, which is a reaction between liquid solders and solid substrate.
The objective of this study is to investigate whether this strong Cu concentration
dependency also exists for the reaction between Ni and solid SnAgCu solders. In other words,
we would like to know whether this dependency still exist during the aging of the solder
joints.
2. Experimental
Five different SnAgCu solders (Sn3.9Ag0.2Cu, Sn3.9Ag0.4Cu, Sn3.9Ag0.5Cu,
Sn3.9Ag0.6Cu, Sn3.9Ag0.8Cu) were prepared from 99.99% purity elements. A balance with
a 0.0001 g precision was used to weigh the elements, producing a maximum composition
error of 0.01 wt.%. Mixtures of the pure elements with the chosen composition were melted
at 850℃ for 250 hours to produce the solder ingots with different Cu concentrations. For
every reaction, a small piece of solder (200±10 mg) was cut from a ingot. The concentrations
of Ag and Cu in such small solder pieces were checked by ICP (Introduced Couple Plasma)
spectrum analysis to ensure the composition is the same as the ingot. Nickel disks (6.35 mm
diameter × 0.50 mm thick, 99.995% pure) were utilized to react with the solder pieces.
Before reaction, each Ni disk was metallurgically polished on both surfaces. The 1-μm
diamond abrasive was used as the last polishing step and then Ni disks were cleaned with
acetone, etched in a 50 vol. % HCl solution (in methanol) for 30 sec, and fluxed with a mildly
active rosin flux. Each solder piece was placed on a Ni disk and then reflowed through a
reflow oven. The peak reflow temperature was fixed at 250℃, and heating rate and cooling
rate were both fixed at 1℃/s, respectively. The time the solder was in the molten state was
120 s. After reflow, the samples were aged at 180℃ for as long as 5000 hrs.
The samples were then removed from the vials, mounted in epoxy, sectioned by using a
low-speed diamond saw, and metallurgically polished in preparation for characterization. The
reaction zone for each sample was examined using an optical microscope and a scanning
electron microscope (SEM). The compositions of the reaction products were determined
using an electron microprobe, operated at 20 keV. In microprobe analysis, the concentration
of each element was measured independently, and the total weight percentage of all elements
was within 100  1% in each case. For every data point, at least four measurements were
made and the average value was reported.
3. Results
The interfaces between Ni and SnAgCu solders of different compositions were similar to
what had been reported in the literature [12-15]. When the Cu concentration was 0.2 wt.%,
only (Ni1-xCux)3Sn4 formed at the interface. When the Cu concentration increased to 0.4 wt.%,
in addition to a continuous (Ni1-xCux)3Sn4 layer, a small amount of discontinuous
(Cu1-yNiy)6Sn5 particles formed between the (Ni1-xCux)3Sn4 layer and the solder. When the Cu
concentration increased to 0.5 wt.%, the (Cu1-yNiy)6Sn5 phase became a continuous layer over
the (Ni1-xCux)3Sn4 layer. At higher Cu concentrations (0.6 and 0.8 wt.%), (Ni1-xCux)3Sn4 was
not detected, but a continuous (Cu1-yNiy)6Sn5 layer formed between Ni and the solder.
Figure 1 shows the interfaces for the Sn3.9Ag0.2 Cu solder with different aging time.
For the sample right after reflow and for the one after 150 hours aging, only a layer of
(Ni1-xCux)3Sn4 was at the interface. However, when the aging time increased to 500 hours and
longer, a layer of (Cu1-yNiy)6Sn5 and a layer of (Ni1-xCux)3Sn4 formed over the Ni layer as
shown in Fig. 1 (c) and (d). These two compounds (Cu1-yNiy)6Sn5 and (Ni1-xCux)3Sn4 are
based on the Cu6Sn5 and Ni3Sn4 crystal structures, respectively, as established using X-ray
diffraction in an earlier study [12]. Right after reflow, the Cu atoms scattered randomly
throughout the entire solder joint. However, after aging for sufficient amounts of time, some
of the Cu went back to the interface and formed (Cu1-yNiy)6Sn5. Figure 2 is a EPMA line-scan
for the Sn3.9Ag0.2Cu solder joint that had been aged at 180 ℃ for 1000 hours. As can be
seen in Fig. 2, substantial amount of Ni can be dissolved in (Cu1-yNiy)6Sn5, reaching about
(Cu0.57Ni0.43)6Sn5. The compound (Ni1-xCux)3Sn4 can also dissolved appreciable amount of Cu,
reaching about (Ni0.8-0.9Cu0.2-0.1)3Sn4, according to the data in Fig. 2. It should be noted that
the composition of (Cu1-yNiy)6Sn5 and (Ni1-xCux)3Sn4 are quite similar to what had been
reported for the solid-state reaction between SnCu solders and Ni by Chen et al. [26].
The formation of a (Cu1-yNiy)6Sn5 and (Ni1-xCux)3Sn4 double-layer after reflow was not
limited to the Sn3.9Ag0.2Cu solder. Figure 3 shows the interfaces for the Sn3.9Sg0.6Cu
joints. As shown in Fig. 3 (a), only a layer of Cu1-yNiy)6Sn5 was at the interface right after
reflow. This is different from the Sn3.9Ag0.2Cu, where only (Ni1-xCux)3Sn4 was present right
after reflow. As the aging time increased, a thin layer of (Ni1-xCux)3Sn4 formed below
(Cu1-yNiy)6Sn5 after 150 hours of aging, as shown in Fig. 3 (b). Both (Cu1-yNiy)6Sn5 and
(Ni1-xCux)3Sn4 grew thicker with the aging time. The situation for the Sn3.9Ag0.8Cu was
similar. Figure 4 shows the interfaces for the Sn3.9Sg0.8Cu joints. One major difference
between Sn3.9Ag0.8Cu and Sn3.9Ag0.6Cu was that it too much longer for (Ni1-xCux)3Sn4 to
formed below (Ni1-xCux)3Sn4. As shown in Fig. 4 (c), a clear (Ni1-xCux)3Sn4 layer was visible
only after 3500 hours of aging. Figure 5 is a EPMA line-scan for the Sn3.9Ag0.8Cu solder
joint that had been aged at 180 ℃ for 5000 hours. As can be seen in Fig. 2, substantial
amount of Ni can be dissolved in (Cu1-yNiy)6Sn5, reaching about (Cu0.57-0.69Ni0.43-0.31)6Sn5.
Figure 6 shows the interfaces for solders with different compositions that had been aged
at 180 oC for 5000 hours. For all Cu concentrations used, a layer of (Cu1-yNiy)6Sn5 and a layer
of (Ni1-xCux)3Sn4 formed over the Ni layer as shown in Figure 6. The observation that, after
aging, different solders produced the same compounds at the interface is quite surprising.
These results are in sharp contrast to the interfaces right after reflow, where the reaction
products depended strongly on the Cu concentration of the solder.
The thickness of (Ni1-xCux)3Sn4 is shown in Fig. 7. As can be seen here, (Ni1-xCux)3Sn4
grew thicker as the aging time increased. In the reaction of Sn3.9Ag0.4Cu/Ni, the thickness
of (Ni1-xCux)3Sn4 is the highest compared to other solder compositions (Fig. 8). This is
probably because the (Cu1-yNiy)6Sn5 layer for this composition is among the thinnest. For the
Sn3.9Ag0.4Cu/Ni reaction, the (Ni1-xCux)3Sn4 layer was not visible until the aging time
became longer than 2000 hours. Figure 8 shows that the (Cu1-yNiy)6Sn5 layer for this solder
composition is the highest. In summary, the (Ni1-xCux)3Sn4 thickness correlates inversely with
that of (Cu1-yNiy)6Sn5. This indicated that (Cu1-yNiy)6Sn5 is an effective diffusion barrier that
hindered the growth of (Ni1-xCux)3Sn4.
Figure 8 shows that the thickness of the (Cu1-yNiy)6Sn5 layer increased as the Cu
concentration in the solder increased. This phenomenon is quite reasonable because the
amount of Cu available increased with the Cu concentration in the solder. It is interesting to
estimate how much Cu in solders had gone back to the interface to become (Cu1-yNiy)6Sn5
after 5000 hours of aging. Alternatively, we can ask if all Cu had gone back to the interface to
form (Cu1-yNiy)6Sn5, how thick (Cu1-yNiy)6Sn5 would had been? Figure 9 is a comparison
between the measured thickness of (Cu1-yNiy)6Sn5 and the estimated thickness by assuming
all Cu had formed (Cu1-yNiy)6Sn5 at the interface. In calculating the estimated thickness, the
density of (Cu1-yNiy)6Sn5 was taken to be the same as that of Cu6Sn5, which was reported to
be 8.28 g/cm3 [27]. The solder used in each sample was 200 mg, and the wetting area (the
area that solder was in direct contact with the Ni substrate) of each sample was 0.16 cm 2. The
values for the composition of (Cu1-yNiy)6Sn5 were from the EPMA measurement. It can be
seen from Fig. 9 that the measured thickness was always slightly thinner than that of the
estimated thickness. In other words, after 5000 hours of aging, most of the Cu atoms had
gone back to the interface.
4. Discussion
According to earlier studies [12-15], the reaction products right after reflow were very
sensitive to the Cu composition in the solder. With increasing Cu concentration, the reaction
product switched from (Ni1-xCux)3Sn4 to (Cu1-yNiy)6Sn5+ (Ni1-xCux)3Sn4, then to
(Cu1-yNiy)6Sn5. However, this strong sensitivity to the Cu composition disappeared when the
solder joints were subjected to aging at high temperatures for sufficient amounts of time.
Figure 6 shows that aging at 180 oC for 5000 hours can make all the solder joint have the
same type of intermetallics compounds at the interface, i.e. a layer of (Cu1-yNiy)6Sn5 over a
layer of (Ni1-xCux)3Sn4. This result suggests that any initial difference in the strength of the
solder joints due to Cu concentration may disappear as the solder joints are subjected to
solid-state aging at high temperatures for a sufficient period of time.
A recently established Cu-Ni-Sn 240 oC ternary isotherm, shown in Figure 10 [28],
provides rationalization for such a strong dependence on the Cu concentration for the solder
joints right after reflow. The temperature of the isotherm (240 oC) is quite close to the
temperature used in this study for the reflow (250 oC), and the isotherm should provide a
good approximation. According to this isotherm, the phase fields that are in equilibrium with
the
Sn
phase
[denoted
as
(Sn)]
include
one
three-phase
field,
(Sn)+(Cu1-yNiy)6Sn5+(Ni1-xCux)3Sn4, and two two-phase fields, (Sn)+(Ni1-xCux)3Sn4 and
(Sn)+(Cu1-yNiy)6Sn5. Notice that the (Sn) single-phase field is very small in size, and this is
the reason why a small change in Cu concentration can produce completely different results.
When the Cu concentration in solder was high (x=0.6 and 0.8), the interface represented a
tie-line in the (Sn)+(Cu1-yNiy)6Sn5 two-phase field, and (Cu1-yNiy)6Sn5 formed next to the
solder. When the Cu concentration was low (x=0.2), the interface represented a tie-line in the
(Sn)+(Ni1-xCux)3Sn4 two-phase field, and (Ni1-xCux)3Sn4 forms next to the (Sn) phase. When
the Cu concentration was in-between, the (Sn)+(Cu1-yNiy)6Sn5+ (Ni1-xCux)3Sn4 three-phase
field dominated, and both (Cu1-yNiy)6Sn5 and (Ni1-xCux)3Sn4 formed.
Next, let us discuss the reasons for the strong Cu sensitivity to the Cu composition to
disappear after solid-state aging. According to the Cu-Ni-Sn isotherm in Figure 10,
(Cu1-yNiy)6Sn5 is not in thermodynamic equilibrium with Ni. Therefore, for the cases of high
Cu concentrations, where there was only a layer of (Cu1-yNiy)6Sn5 over Ni, there was driving
force for the (Ni1-xCux)3Sn4 phase to nucleate and grow between the (Cu1-yNiy)6Sn5 layer and
the Ni layer. For the cases of low Cu concentrations, where these was only a layer
(Ni1-xCux)3Sn4 over Ni, the Cu inside the solder would diffuse back to the interface and
formed a layer of (Cu1-yNiy)6Sn5 over the (Ni1-xCux)3Sn4 layer. In this case, the Cu inside the
solder existed in the form of Cu6Sn5. The compound Cu6Sn5 would decompose and release
the Cu atoms. The Cu atoms then diffused back to the interface to react with Sn and Ni to
form (Cu1-yNiy)6Sn5. The driving force for Cu6Sn5 to resettle back to the interface is to seek
Ni so that the composition of (Cu1-yNiy)6Sn5 can approach (Cu0.57Ni0.43)6Sn5. Kinetically, the
factor enables this to occur is that Cu is a fast diffuser in Sn, and the diffusion rate of Cu in
Sn is fast enough for the process to happen. In fact, the growth mechanism for (Cu1-yNiy)6Sn5
is quite similar to the well-known (Au1-zNiz)Sn4 layer formed in the solder joints that utilize
the Au-bearing surface finishes [18, 21, 22].
5. Conclusions
In this study, the effect of a small perturbation in the Cu concentration on the solid-state
aging reaction between the SnAgCu solders and Ni was investigated. Specifically, five
Sn3.9AgxCu solders (x=0.2, 0.4, 0.5, 0.6, and 0.8) were reacted with Ni at 180 oC. It was
found that the strong sensitive to the Cu composition after reflow disappeared after the
solid-state aging at high temperatures for a sufficient period of time. For all the Cu
concentrations studied, the same type of intermetallics compounds formed at the interface
after aging. A layer of (Cu1-yNiy)6Sn5 over a layer of (Ni1-xCux)3Sn4 was found at the interface.
This study shows that the initial difference in the intermetallics compounds right after reflow
can be aged out at high temperatures. The growth mechanisms for (Cu1-yNiy)6Sn5 and
(Ni1-xCux)3Sn4 are different. The (Cu1-yNiy)6Sn5 layer grew by the reaction of Cu atoms from
the solder. In fact, the growth mechanism for (Cu1-yNiy)6Sn5 is quite similar to the
well-known (Au1-zNiz)Sn4 formed in the solder joints that utilize the Au-bearing surface
finishes. The (Ni1-xCux)3Sn4 layer grew by the reaction of the Ni layer and the (Cu1-yNiy)6Sn5
layer.
Acknowledgment.
This work was supported by the National Science Council of R.O.C.
through grants NSC-92-2216-E-008-006 and NSC-92-2214-E-008-003.
Reference
1.
J. Glazer, Inter. Mater. Rev., vol. 40, p.65, 1995.
2.
M. Abtew and G. Selvaduray, Mater. Sci. & Eng. R, vol. 27, p.95, 2000.
3.
K. Suganuma, Current Opinion Solid State Mater. Sci., vol. 5, p.55, 2001.
4.
K. Zeng and K. N. Tu, Mater. Sci. & Eng. R, vol. 38, p.55, 2002.
5.
K. W. Moon, W. J. Boettinger, U. R. Kattner, F. S. Biancaniello, and C. A. Handwerker, J.
Electron. Mater, vol. 29, p.1122, 2000.
6.
http://www.nemi.org/.
7.
K. N. Tu and K. Zeng, Mater. Sci. & Eng. R, vol. 34, p.1, 2001.
8.
K. N. Tu, T. Y. Lee, J. W. Jang, L Li, D. R. Frear, K. Zeng, and J. K. Kivilahti, J. Appl.
Phys., vol. 89, p.4843, 2001.
9.
K. N. Tu and R. D. Thompson, Acta Metall., vol. 30, p.947, 1982.
10. Z. Mei, A. J. Sunwhoo, and J. W. Morris, Jr., Metall. Trans. A, vol. 23, p.857, 1992.
11. C. E. Ho, Y. L. Lin, J. Y. Tsai, and C. R. Kao, J. Chin. Inst. Chem. Eng., vol. 34, p.387,
2003.
12. C. E. Ho, Y. L. Lin, and C. R. Kao, Chem. Mater., vol. 14, p.949, 2002.
13. W. T. Chen, C. E. Ho, and C. R. Kao, J. Mater. Res., vol. 17, p.263, 2002.
14. L. C. Shiau, C. E. Ho, and C. R. Kao, Solder. Surf. Mt. Tech., vol. 14, p.25, 2002.
15. C. E. Ho, R. Y. Tsai, Y. L. Lin, and C. R. Kao, J. Electron. Mater., vol. 31, p.584, 2002.
16. C. E. Ho, L. C. Shiau, and C. R. Kao, J. Electron. Mater., vol. 31, p.1264, 2002.
17. W. H. Tao, C. Chen, C. E. Ho, W. T. Chen, and C. R. Kao, Chem. Mater., vol. 13, p.1051,
2001.
18. C. E. Ho, S. Y. Tsai, and C. R. Kao, IEEE Trans. Adv. Pack., vol. 24, p.493, 2001.
19. C. Chen, C. E. Ho, A. H. Lin, G. L. Luo, and C. R. Kao, J. Electron. Mater., vol. 29,
p.1200, 2000.
20. C. M. Liu, C. E. Ho, W. T. Chen, and C. R. Kao, J. Electron. Mater., vol. 30, p. 1152,
2001.
21. C. E. Ho, W. T. Chen, and C. R. Kao, J. Electron. Mater., vol. 30, p.379, 2001.
22. C. E. Ho, R. Zheng, G. L. Luo, A. H. Lin, and C. R. Kao, J. Electron. Mater., vol. 29,
p.1175, 2000.
23. C. E. Ho, Y. M. Chen, and C. R. Kao, J. Electron. Mater., vol. 28, p.1231, 1999.
24. M. S. Lee, C. Chen, and C. R. Kao, Chem. Mater., vol. 11, p.292, 1999.
25. M. S. Lee, C. M. Liu, and C. R. Kao, J. Electron. Mater., vol. 28, p.57, 1999.
26. W. T. Chen, R. Y. Tsai, Y. L. Lin, and C. R. Kao, J. SMT, vol. 15, p.40, 2002.
27. H. P. R. Frederikse, R. J. Fields, and A. Feldman, J. Appl. Phys., vol. 72, p.2879, 1992.
28. C. H. Lin, S. W. Chen, and C. H. Wang, J. Electron. Mater., vol. 31, p.907, 2002.
Figure Captions
Fig. 1
The cross-section view for the Sn3.9Ag0.2Cu/Ni interfaces that had aged at
180 oC for (a) 0 hours, (b) 150 hours, (c) 500 hours, and (d) 1000 hours.
Fig. 2
EPMA line-scan across the reaction zone for the sample shown in Fig. 1 (d).
The intermetallic compound present are (Cu1-yNiy)6Sn5 and (Ni1-xCux)3Sn4.
Fig. 3
The cross-section view for the Sn3.9Ag0.6Cu/Ni interfaces that had aged at
180 oC for (a) 0 hours, (b) 150 hours, (c) 2000 hours, and (d) 5000 hours.
Fig. 4
The cross-section view for the Sn3.9Ag0.8Cu/Ni interfaces that had aged at
180 oC for (a) 0 hours, (b) 2000 hours, (c) 3500 hours, and (d) 5000 hours.
Fig. 5
EPMA line-scan across the reaction zone for the sample shown in Fig. 4 (d).
The intermetallic compound present are (Cu1-yNiy)6Sn5 and (Ni1-xCux)3Sn4.
Fig. 6
The cross-section view for the solders/Ni interfaces that had aged at 180 oC for
5000 hours for the (a) Sn3.9Ag0.2Cu/Ni interface, (b) Sn3.9Ag0.4Cu/Ni
interface, (c) Sn3.9Ag0.5Cu/Ni interface, (d) Sn3.9Ag0.6Cu/Ni interface and
(e) Sn3.9Ag0.8Cu/Ni interface. For all the solder compositions, both
(Ni1-xCux)3Sn4 and (Cu1-yNiy)6Sn5 formed.
Fig. 7
The thickness of (Ni1-xCux)3Sn4 versus the aging time at 180 oC.
Fig. 8
The thickness of (Cu1-yNiy)6Sn5 versus the aging time at 180 oC.
Fig. 9
The measured thickness of (Cu1-yNiy)6Sn5 after aging at 180 oC for 5000 hours.
Also shown is the estimated thickness obtained by assuming all Cu in the
solder had come back to the interface to form (Cu1-yNiy)6Sn5.
Fig. 10
The CuNiSn isotherm at 240 oC. This isotherm was adapted from Lin et al.
[28].