Sei sulla pagina 1di 4

Journal of Alloys and Compounds 476 (2009) 138141

Contents lists available at ScienceDirect

Journal of Alloys and Compounds


journal homepage: www.elsevier.com/locate/jallcom

Creep behavior of eutectic 80Au/20Sn solder alloy


G.S. Zhang a , H.Y. Jing a , L.Y. Xu a, , J. Wei b , Y.D. Han a
a
b

School of Material Science and Engineering, Tianjin University, Tianjin 300072, China
Singapore Institute of Manufacturing Technology, 71 Nanyang Drive, Singapore 638075, Singapore

a r t i c l e

i n f o

Article history:
Received 28 July 2008
Received in revised form 20 August 2008
Accepted 3 September 2008
Available online 31 October 2008
Keywords:
Eutectic 80Au/20Sn solder alloy
Creep
Constitutive model
Creep mechanism

a b s t r a c t
Eutectic 80Au/20Sn solder alloy is widely used in high power electronics and optoelectronics packaging
in which the creep property of the solder joint is essential to meet the global demand for longer operating
lifetime in their applications. In this study, the tensile creep behavior of bulk eutectic 80Au/20Sn solder
alloy is reported and compared with 63Sn37Pb solder joint. The creep strain rate increases and creep
lifetime decreases as the applied stress level and temperature increase. The 80Au/20Sn solder alloy shows a
superior anti-creep performance over the 63Sn37Pb solder joint. The experimental data were successfully
t with Dorn model and Garofalo model. However, the application of Garofalo model resulted in a lower
estimated variance of error terms as compared to the Dorn model. Grain boundary sliding is the possible
creep mechanism within the given stress level and temperature. The nucleation, accumulation and further
growth of microvoids lead to the creep rupture.
2008 Elsevier B.V. All rights reserved.

1. Introduction
The traditional tinlead solders have been widely employed
as electrical interconnects in electronic industries. However, the
toxic Pb can cause harmful inuence on environment and health.
On the other hand, the tinlead solders would no longer satisfy
the reliability requirement in high power electronic and optoelectronic components with the smaller size of solder joint and
higher mechanical, thermal and electrical load in which the excellent creep resistance is essential. Hence, it is urgent to develop
the lead-free solders with excellent creep and fatigue resistance.
In the recent decade, a great amount of lead-free solders have
been reported and the promising candidates mainly include the
Sn-based alloys such as SnAg, SnCu, SnZn, SnBi, SnSb, AuSn,
and SnAgCu systems. Among the lead-free candidate solders,
the eutectic 80Au/20Sn solder alloy is particularly attractive in
high power electronics and optoelectronics as hermetic sealing and
die attachment material because it has excellent high-temperature
performance, high mechanical strength, high electrical and thermal conductivity, and can also offer uxless soldering [16]. The
manufacturing process for such applications has been investigated
extensively, with emphasis on the bonding quality, the metallurgical interaction and formation of AuSn solder bump [711]. Besides
the manufacturing and solderability, the thermomechanical properties of 80Au/20Sn solder alloy have been reported [1214].

Corresponding author. Tel.: +86 22 2740 2439; fax: +86 22 2740 5020.
E-mail address: xulianyong@tju.edu.cn (L.Y. Xu).
0925-8388/$ see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2008.09.009

Nevertheless, the creep behavior of 80Au/20Sn solder has been seldom reported in the literature. The 80Au/20Sn solder has obvious
creep response because of its lower melting point. The creep deformation can lower the reliability and performance of the electrical
components signicantly and creep damage is vital for the reliability of solder joint. Therefore, it is necessary to gain an insight into
the creep behavior of eutectic 80Au/20Sn solder alloy.
Creep characteristics of solder alloys can be studied by different
experimental techniques such as conventional tensile creep [15],
uniaxial tensile [16], stress relaxation [17], impression [18], and
indentation testing methods [14]. The conventional tensile creep
test is time-consuming and difcult for the material with the tendency of geometric instabilities, but it is very important because the
tensile stress state promotes the solder joint failure. In this study,
the tensile creep behavior of bulk eutectic 80Au/20Sn solder alloy
was investigated at different temperatures and stress levels and the
experimental results were compared with 63Sn37Pb solder joint.
The creep constitutive models for description of high temperature
creep behavior and low cycle fatigue property of 80Au/20Sn solder
alloy and the possible creep deform mechanisms were proposed.
2. Experimental procedure
The material employed was as-cast 80Au/20Sn solder alloy with a melting point
of 279.4 0.1 C. In order to get the similar microstructure with the solder joints in
the actual packages, the air cooling method was used when the solder bars were
cast. Subsequently, small rectangular specimens were prepared by machining from
the high purity as-cast solder bars. The specimens had a total length of 30 mm,
a gauge length of 20 mm, and a central cross-section of 0.4 mm 1.55 mm. Next,
the gauge section of each specimen was ground carefully with SiC papers (#220,
500, 1200, and 2400) and polished using 1 m diamond past. Afterwards, the spec-

G.S. Zhang et al. / Journal of Alloys and Compounds 476 (2009) 138141
imens were annealed at 60 C for 24 h in a N2 atmosphere to eliminate the residual
stresses.
The tensile creep tests were performed on a micro-force test system (Tytron
895.20A-02). The system has the capability to maintain a constant loads (load variation <5%) for a long time. The system was equipped with a specially designed
environmental chamber rated from 65 to 250 C for running creep tests at low
and high temperatures. TestStarII software was used to control the system and collect the data. The tests were conducted at three different temperatures of 25, 75 and
125 C with a constant stress control. An extensometer with a gauge of 10 mm and
resolution of 0.1 m was employed to measure the strain of the specimen. The specimens were held at the test temperature for 30 min before loading. Three samples
were tested for each condition and the results were averaged.
Following creep testing, metallurgical examinations were conducted for the
specimens. The standard procedures of metallographic preparation were sample
mounting, grinding and polishing, chemical etching, microscope examination and
measurements. The bonded samples were cold-mounted with epoxy for 8 h and
polished. The mold was ground using SiC paper of grit 600 to expose the interface then SiC paper of grit 1200 was used to further grind the materials until the
cross-section of the sample was visible. Next, the sectioned area was polished
with Type K alumina suspension (liquid diamond) of 3, 1, and 0.5 m sequentially until the interface and internal microstructure of the solder joint can be
seen clearly under an optical microscope. A nal touchup using polishing suspensions (COL-K) allowed better surface nishing for microstructural analysis.
The samples were then examined with a combination of optical microscope and
SEM/EDX.

3. Results and discussion


3.1. Creep curves
Fig. 1 shows the creep curves of 80Au/20Sn solder alloy. It can be
seen that the creep curves display typical primary and secondary
steady-state creep stages. The tertiary creep deformation characteristic can also be seen clearly at 125 C when the applied stress
is 25 MPa. The creep characteristics distinguish obviously at different stress levels and temperatures. The creep strain increases with
an increase in applied stress level and temperature. Then creep

139

Table 1
The steady-state creep rate and creep lifetime of 80Au/20Sn solder alloy.
Temperature (K)

Creep stress (MPa)

Creep rate (1/s)

Creep lifetime (h)

298
298
298
298
298
298
298
348
348
348
348
348
348
348
398
398
398
398
398
398
398

100
125
150
175
200
250
300
25
37.5
50
60
75
87.5
100
5
7.5
10
12.5
15
16.7
25

1.58e8
2.84e8
4.55e8
8.80e8
9.26e8
1.32e7
2.28e7
1.98e7
4.88e7
8.49e7
1.20e6
2.14e6
2.79e6
3.61e6
7.04e7
2.30e6
6.34e6
8.74e6
1.57e5
2.01e5
6.57e5

132.215
123.52
114.45
101.20
49.90
47.26
42.67
145.67
97.66
64.61
53.11
13.31
12.03
9.02
119.70
113.54
47.25
22.96
17.34
16.23
3.05

strain rate at any given time can be determined by differentiating creep strain versus time and the minimum rate was taken as
the creep strain rate of steady-state stage. The total time from the
beginning of primary stage to the end of tertiary stage is dened
as the creep lifetime. The effect of stress level and temperature
on the steady-state creep rate and creep lifetime are shown in
Table 1. It indicates that the creep strain rate increases and creep
lifetime decreases sharply with increasing applied stress level and
temperature.
3.2. Constitutive model
The homologous temperature of 80Au/20Sn solder alloy at room
temperature exceeds 0.5 so the creep is considered as the dominant deformation-controlling mechanism. Considering that the
steady-state stage takes up the most of the creep time of nal failure, the secondary creep strain region is chosen to represent the
entire process in most of the constitutive models. In the steadystate creep stage, the creep strain rate can be generally described
by WeertmanDorn equation [19] as follows:
=

AGb
kT

 b p   n
G

 Q

D0 exp

kT

(1)

where is the steady-state creep strain rate, G is the shear modulus,


b is the Burgers vector, k is Boltzmanns constant, T is the absolute
temperature, A is a constant, d is the grain size, p is the grain size
exponent,  is the applied stress, D0 is the frequency factor, n is the
creep stress exponent and Q is the activation energy of the deformation process. From the above equation, it can be seen clearly
that the steady-state creep strain rate is related to the properties of
the solder alloy itself (shear modulus, grain size, etc.) and service
conditions (service temperature and applied stress). The smaller
the grain size, or the higher the service temperature, or the higher
the applied stress, the higher the steady-state creep strain rate and
the shorter the creep lifetime. As the Boltzmann constant is equal to
the ratio of the gas constant (R) to the Avogadro constant, and some
constants can be converted into a new constant, the above equation can be simplied as the Dorn power law constitutive model
[20] given by

 Q

= A1  n exp
Fig. 1. Tensile creep curves of 80Au/20Sn solder alloy. (a) Under different stress
levels at 125 C; (b) at different temperatures with the applied stress of 25 MPa.

RT

where A1 is a complex constant.

(2)

140

G.S. Zhang et al. / Journal of Alloys and Compounds 476 (2009) 138141

It is assumed that the deformation is dominated by one creep


mechanism in the whole stress range applied, and therefore the
stress exponent can be taken as a constant at any given temperature.
Then we have the calculation as follows.
Taking the natural logarithmic transformation on both sides of
Eq. (2), then we have:
ln = ln A1 + nln 

Q
RT

(3)

It is clear that at a given temperature the creep stress exponent


can be calculated by linear regression of the experimental data. For
the bulk 80Au/20Sn solder alloy, the calculated stress exponents
at 25, 75 and 125 C are 2.37, 2.10 and 2.77, respectively. The creep
stress exponent and creep activation energy are stress level applied
and temperature dependent in Dorn model, the average value can
be calculated by tting all the experimental results together. Eq. (4)
shows the tted Dorn model:

= 8.37 105  2.07 exp

1.02 105
RT

(4)

The hyperbolic sine type Garofalo model [18] describing creep


ow for low and high stresses where the simple power law breaks
down is also used to predict the steady-state creep behavior of
solders. The model is given by


 Q 

= A2 [sinh()]n exp

(5)

RT

In this equation, is the stress coefcient. Data are t by iterative


multivariable nonlinear regression method, the following t was
obtained:
= 4.62 1015 [sinh(2 105 )]

2.07

exp

1.02 105
RT

(6)

Fig. 2 compares the Dorn model and Garofalo model for the
steady creep rate versus applied stress curves. Clearly, both models
are able to provide an acceptable description of eutectic 80Au/20Sn
solder alloy experimental results over the present experimental
stresstemperature range. However, the Garofalo model exhibits
a lower estimated variance of error terms as compared to the Dorn
model with the present data. The better agreement between Garofalo model and experimental results is evident.
For the eutectic 80Au/20Sn solder alloy, the calculated creep
stress exponent and creep activation energy is 2.07 and 102 kJ/mol,
respectively by tensile creep test, which are different from n = 2.55,
and Q = 18.95 kcal/mol (79.29 kJ/mol) obtained by Morgan based
on uniaxial compression creep testing [21]. The variations can be
explained by differences in testing methods, microstructure, specimen preparation, measuring errors, and data processing method.

Fig. 2. Comparison on steady-state creep rate obtained from experimental results


and t models. (a) Dorn model; (b) Garofalo model.

morphology of the specimens after testing. The grain boundary sliding along the loading direction and the cuplike depressions across
the specimen can be observed. The size of the depressions depends
on the number and distribution of microvoids that are nucleated.
It is known that the creep deformation by microvoids growth and
coalescences mechanisms generally relates to intergranular creep
fracture process [23]. The fractography does not vary signicantly
but for the reduction of fracture surface. This veries the creep
mechanism which does not change and indicates high mechanical
strength of 80Au/20Sn solder even at high temperature and applies
stress level.
From the discussion above, the grain boundary sliding can be
the dominant creep mechanism. At high homologous temperature

3.3. Creep failure mechanism


When the Dorn power law creep relation is valid, the value
of creep stress exponent is often used to identify the creep
deformation-controlling mechanism. The creep stress exponent
does not change signicantly over the wide applied stress level
and the testing temperature range. This suggests that the change
in applied stress and temperature does not change the creep rate
controlling mechanism. Grivas et al. investigated the deformation
behavior of PbSn solder and pointed out that the deformation
was controlled by grain boundary (GB) sliding near the creep stress
exponent close to 2 [22]. Further studies should be performed to
clarify the creep mechanism of the eutectic 80Au/20Sn solder alloy.
The microstructure often changes during the creep process; the
creep deformation mechanisms can be largely claried by analyzing the creep fractography. Fig. 3 shows a typical fractography

Fig. 3. Typical morphology of the fractography of eutectic 80Au/20Sn solder alloy.

G.S. Zhang et al. / Journal of Alloys and Compounds 476 (2009) 138141

141

steady-state creep rate and higher creep resistance compared


with the 63Sn37Pb solder.
(2) The Dorn power law creep stress exponent does not change
signicantly with increasing the stress, and therefore both
Dorn model and Garofalo model are suitable for the eutectic 80Au/20Sn solder over the tested stress and temperature
ranges. However, the latter ts the experimental data more
successfully.
(3) The grain boundary sliding is probably the most dominant
creep mechanism in this experiment and the nal intergranular fracture via nucleation, accumulation and further growth
of microvoids.
Acknowledgements
Fig. 4. Comparison of the steady-state creep model curves between 80Au20Sn and
63Sn37Pb solder.

and under the action of applied stress, grain boundary sliding takes
place and leads to the formation of microvoids. The nucleation,
accumulation and further growth of microvoids nally results in
the intergranular fracture of typical creep.
3.4. Comparison with 63Sn37Pb solder
The creep behavior of the traditional eutectic 63Sn37Pb solder
has been widely reported by many researchers. For instance, the
constant-load creep tests of eutectic 63Sn37Pb solder were performed at temperatures of 25, 75 and 125 C by Zhang et al. [24] The
comparison between the t steady-state creep constitutive model
curves of 63Sn37Pb and eutectic 80Au/20Sn solder is shown in
Fig. 4. It can be seen that the eutectic 80Au/20Sn solder has lower
steady-state creep rate and the steady-state creep rate does not
increase as fast as 63Sn37Pb solder with an increase in the applied
stress. This attributes to the good mechanical properties of (AuSn)
and  (Au5 Sn) phase in eutectic 80Au/20Sn solder. The tiny and 
phases strengthen pure Sn grain boundaries and block grain boundary sliding, thus the eutectic 80Au/20Sn solder has superior creep
resistance and higher creep activation energy than the 63Sn37Pb
eutectic solder.
4. Conclusions
Relationships between tensile creep behavior and microstructure of eutectic 80Au/20Sn solder alloy was studied and compared
with the 63Sn37Pb solder. Two steady-state creep constitutive
models were employed to describe the observed creep behaviors.
The study supports the following conclusions:
(1) The eutectic 80Au/20Sn solder alloy exhibits the typical creep
deformation characteristics. The creep strain increases and
creep lifetime decreases with the improved applied stress level
and temperature. The 80Au/20Sn solder alloy exhibits a lower

This work is sponsored by the Doctor Subject Foundation of


China under grant no. 20050056035 and the Natural Science Foundation of China under grant no. 50575160.
References
[1] A. Larsson, S. Forouhar, J. Cody, R.J. Lang, IEEE Photon. Technol. Lett. 2 (1990)
307309.
[2] G.W. Yang, R.J. Hwu, Z.T. Xu, X.Y. Ma, IEEE J. Quant. Electron. 35 (1999)
15351541.
[3] W. Pittroff, G. Erbert, G. Beister, et al., IEEE Trans. Adv. Packag. 24 (2001)
434441.
[4] J. Zhao, L. Li, W.M. Wang, Y.C. Lu, IEEE Photon. Technol. Lett. 15 (2003)
15071509.
[5] Y. Qu, S. Yuan, C.Y. Liu, B.X. Bo, G.J. Liu, H.L. Jiang, IEEE Photon. Technol. Lett. 16
(2004) 389391.
[6] X.S. Liu, M.H. Hu, H.K. Nguyen, C.G. Caneau, M.H. Rasmussen, IEEE Trans. Adv.
Packag. 27 (2004) 640646.
[7] D.S. Ellis, J.M. Xu, IEEE J. Select. Top. Quant. Electron. 3 (1997) 640648.
[8] D.A. Ackerman, P.A. Morton, R.F. Kazarinov, T. Tanbun-Ek, R.A. Logan, Appl. Phys.
Lett. 66 (1995) 466468.
[9] S.J. Sweeney, L.J. Lyons, A.R. Adams, D.A. Lock, IEEE J. Select. Top. Quant. Electron.
9 (2003) 13251331.
[10] J.W.R. Teo, F.L. Ng, L.S.K. Goi, et al., Microelectron. Eng. 85 (2008) 512517.
[11] W. Jiang, Cryst. Res. Technol. 41 (2006) 150153.
[12] A. Yamauchi, Y. Arai, Proceedings of the Electronic Components and Technology
Conference, 2001, pp. 242246.
[13] G. Chen, Phys. Rev. B 57 (1998) 1495814973.
[14] Y.C. Liu, J.W.R. Teo, S.K. Tung, K.H. Lam, J. Alloys Compd. 448 (2008) 340343.
[15] M.D. Mathew, H. Yang, S. Movva, K.L. Murty, Metall. Mater. Trans. 36A (2005)
99105.
[16] X. Chen, G. Chen, M. Sakane, IEEE Trans. Compon. Packag. Technol. 28 (2005)
111116.
[17] S.G. Jadhav, T.R. Bieler, K.N. Subramanian, J.P. Lucas, J. Electron. Mater. 30 (2001)
11971205.
[18] J.J. Stephens, D.R. Frear, Metall. Mater. Trans. A 30 (1999) 13011313.
[19] A. Schubert, H. Walter, R. Dudek, et al., Proceedings of the International Symposium on Advanced Packaging Materials Processes, 2001, pp. 129134.
[20] R.S. Whitelaw, R.W. Neu, D.T. Scott, Trans. ASME, J. Electron. Packag. 121 (1999)
99107.
[21] H.S. Morgan, Trans. ASME, J. Electron. Packag. 113 (1995) 350354.
[22] D. Grivas, K.L. Murty, J.W. Morris Jr., Acta Metall. 27 (1979) 731737.
[23] A.J. McEvily, Metal failures: mechanisms, in: Analysis and Prevention, John
Wiley & Sons, New York, 2002, pp. 163-180.
[24] Q. Zhang, A. Dasgupta, P. Haswell, ASME Electron. Photon. Packag. 3 (2003)
215224.

Potrebbero piacerti anche