Thermal Expansion of Several Sn-based Intermetallic Compounds

Thermal Expansion of Several Sn-based
Intermetallic Compounds
Nan Jiang*, J. A. Clum*, R. R. Chromik** and E. J. Cotts**
*Department of Mechanical Engineering, **Department of Physics
Binghamton University,
State University of New York
Binghamton, NY 13902-6016
When two metals are joined by solder, the joining interface undergoes a reaction to yield
one or more intermetallic layers. Intermetallic compounds can also grow at the interface of the
solder and the substrate during storage [1] at ambient temperatures. While the presence of
intermetallic compounds is an indication that a good metallurgical bond has formed, these
compounds are often more brittle than the solder and may be deleterious to a joint’s mechanical
integrity. When an electronic package encounters thermal fluctuations, cyclical shear strain is
imposed on solder joint interconnections due to the differences in thermal expansion of the printed
wire board, solder, and components. This thermal cycling may lead to a condition of thermal
fatigue and eventual solder joint failure.
Although the values of thermal expansion coefficients for various solders and printed wire
board laminates have been tabulated, literature values for the coefficients of thermal expansion of
solder-metal intermetallics are scarce, and disagreement is found between the few existing
measurements[2,3]. As solder joints are made ever smaller, the intermetallic occupies an ever
greater proportion of the solder joint. It is clear that values of the physical properties of solder
alloys are needed for the detailed analysis, modeling and testing of simulated and actual joints.
Thus we seek to determine values of the coefficient of thermal expansion for several Sn-based
intermetallic compounds. Five intermetallic compounds of special interest in the electronic
packaging industry are Cu 6 Sn 5 , Cu 3 Sn, PdSn 4 , PdSn 2 and Ni 3 Sn 4 . Previously determined room
temperature coefficients of thermal expansion [2, 3] for the above materials include 16.3 +/- 0.3,
19.0 +/- 0.3 and 13.7 0.3 ppm/ o C for Cu 6 Sn 5 , Cu 3 Sn and Ni 3 Sn 4 , respectively. In contrast,
Reichenecker [4] measured the coefficients of thermal expansion for Cu 6 Sn 5 and Cu 3 Sn to be 20.0
and 18.4 ppm/ o C in the temperature range of 25 o C to 100 o C. There have been no published values
for the thermal expansion coefficients of the Pd-Sn intermetallics. In this study, the coefficients of
thermal expansion of all five intermetallic compounds were experimentally determined.
The primary reason for the lack of thermal expansion values for those intermetallics lies in
the difficulty of preparing a single phase bulk solder alloy. Because all five intermetallic
compounds in this study form by peritectic reactions, single phase, homogenous compounds of
practical size can not be produced with a casting method with a relatively slow cooling rate (~
5 o C/second). For this reason, we utilized a more rapid solidification technique: arc melting,
followed by long term annealing. Annealing temperatures of 50 o C below the corresponding
liquidus temperature and annealing times of 30 to 75 days were used.
To verify homogeneity, the samples were examined by x-ray diffraction analysis. In every
case patterns corresponding to the appropriate intermetallic compound were observed. Samples
were also examined using an environmental scanning electron microscope (ESEM)
1

Cu 6 Sn 5 Cu 3 Sn PdSn 2 PdSn 4 Ni 3 Sn 4
60 o C 18.3 18.2 13.6 15.5 14.6
80 o C 18.5 18.4 13.7 15.6 14.9
100 o C 18.6 18.5 13.8 15.7 15.1
120 o C 19.0 18.7 13.9 15.8 15.2
140 o C 19.4 19.0 14.2 16.0 15.7
160 o C 20.2 19.4 14.3 16.2 16.0
TABLE I
Thermal expansion coefficients of five intermetallic compounds in the temperature of
60 o C - 160 o C. All values have the unit of ppm/ o C.
equipped with an energy dispersive spectroscopy (EDX) detector. Cu 3 Sn alloys appeared to be
homogenous, with no observable segregation or porosity. In Cu 6 Sn 5 samples, Sn-rich areas were
observed. The area fraction of this Sn-rich area was determined to be on the order of 3%. Defects
in the form of micropores were discovered in the arc melted compounds of PdSn 4 , PdSn 2 , and
Ni 3 Sn 4 . The porosity of these three samples is on the order of 2-5%.
Thermal expansion tests were performed with a commercially available thermomechanical
analyzer (TMA). The specimen was heated in a stepwise manner, waiting for complete thermal
Thermal expansion coefficient (ppm/ o C)
28
26
24
22
20
18
16
40 60 80 100 120 140 160 180
Temperature ( o C)
Figure 1: A plot of the coefficients of thermal
expansion as a function of temperature for Cu 3 Sn (),
Cu 6 Sn 5 (O), Cu (a), and Sn (/). Values for Cu 3 Sn and
Cu 6 Sn 5 were obtained in this experiment, values for
Cu and Sn were obtained from reference 7. The total
uncertainty of the measurements was estimated to be
2% from the combination of two sources: the standard
deviation of the measurements themselves and that of
calibration values.
Thermal Expansion Coefficient (ppm/ o C)
18
17
16
15
14
13
12
11
40 60 80 100 120 140 160 180
Temperature ( o C)
Figure 2: A plot of the coefficients of thermal
expansion as a function of temperature for PdSn 2 (a),
PdSn 4 (O), and Pd (). Values for PdSn 2 and PdSn 4
were obtained in this experiment, values for Pd were
obtained from reference 7. The total uncertainty of the
measurements was estimated to be 2% from the
combination of two sources: the standard deviation of
the measurements themselves and that of calibration
values.
2

equilibrium at each intervening temperature. Thermal expansion measurements were made on
heating from 50 o C to 170 o C with an isothermal hold of 30 minutes at each 20 o C interval. A
heating rate of 10 o C/min was used. NIST reference materials [5] SRM 731 (Borosilicate Glass),
SRM736 (Copper) and SRM 738 (Stainless Steel) were used for instrument calibration. Three
samples of each material were produced and tested, and for each material between ten and fifteen
runs were performed. Several (approximately 25) runs were also performed with a ramping
method for each material where the samples were heated from 50 o C to 170 o C at a constant heating
rate of 10 o C. Results obtained with this method did not differ significantly from the stepwise
heating method. Table 1 lists coefficients of thermal expansion for all five intermetallic
compounds as a function of temperature with the unit of ppm/ o C. The total uncertainty of the
measurements was estimated to be 2% from the combination of two sources: the standard deviation
of the measurements themselves and that of the calibration values.
At temperatures between 25°C and 100°C we observed that the thermal expansion
coefficient of Cu 6 Sn 5 is essentially the same as that of Cu 3 Sn (Figure 1), while both are
significantly higher than that of Cu. Another experiment was conducted to investigate the thermal
expansion of a phase mixture of Cu-Sn
compounds. Samples of the nonstoichiometric
material CuSn (a mixture of
Cu 6 Sn 5 and Cu 3 Sn) were produced by the
arc- melting method and the thermal
expansion coefficients of this phase
mixture were determined. The coefficient
of thermal expansion of this phase mixture
was the same within experimental error as
those of Cu 6 Sn 5 and Cu 3 Sn. Thus, a single
thermal expansion curve may be used to
model the behavior of Cu 6 Sn 5 , Cu 3 Sn or
the phase mixture, in this temperature
range.
We note that the coefficients of
thermal expansion of the intermetallics
investigated were found to be significantly
different than those of Sn or the base
metal. Comparing the thermal expansion
of the two Pd/Sn compounds, PdSn 4 shows
12% higher expansion than that of PdSn 2
and roughly 30% higher expansion than
that of Pd (Figure 2). The Ni 3 Sn 4
intermetallic compound showed about 6%
higher thermal expansion than that of Ni
(Figure 3). The coefficients of thermal
expansion for solder depend on its
composition ranging from 20 to 29 ppm/ o C
[6], significantly larger than any of the
observed values of thermal expansion
coefficients.
Extrapolation of the data of Fig. 1
indicates that the room temperature
coefficient of thermal expansion for both
Thermal expansion coefficient (ppm/ o C)
16.0
15.6
15.2
14.8
14.4
14.0
13.6
60 80 100 120 140 160
Temperature ( o C)
Figure 3: A plot of the coefficients of thermal
expansion as a function of temperature for Ni 3 Sn 4 (a),
and Ni (). Values for Ni 3 Sn 4 was obtained in this
experiment, values for Ni were obtained from
reference 7. The total uncertainty of the
measurements was estimated to be 2% from the
combination of two sources: the standard deviation of
the measurements themselves and that of calibration
values
3

Cu 6 Sn 5 and Cu 3 Sn is 18 ppm/ o C. This value can be compared to previously determined values of
16.3 and 20.0 ppm/ o C for Cu 6 Sn 5 and 19.0 and 18.4 ppm/ o C for Cu 3 Sn [3,4]. The differences
between these results may be the result of differences in the microstructure of the samples
investigated. For instance, the samples of Ref. 3 were small grained (nanometer length scale) from
HIP (hot isostatic pressed) material, in contrast to the very large grained material of the present
study.
Our results demonstrate that although differences exist between the coefficient of thermal
expansion of the intermetallics in the respective systems, more emphasis should be placed on the
greater differences that exist between the intermetallics and their respective basis metals and/or
solders. These differences provide insight on the long term effect of thermal cycling performance
of soldered joints.
Acknowledgments
This research was funded by the Integrated Electronic Engineering Center at the State University of
New York at Binghamton. The IEEC receives funding from the New York State Science and
Technology Foundation, the National Science Foundation, and a consortium of industrial members.
References
[1] Woodgate, R.W, Handbook of Machine Soldering, (John Wiley and Sons, New York,
1983, Chapter 2).
[2] R. J. Fields, S. R. Low III, and G. K. Lucey, Jr, The Metal Science of Joining, (The
Minerals, Metals & Materials Society, 1992), pp. 165-173.
[3] R. J. Schaefer, F. S. Biancaniello, and R. D. Jiggetts, The Metal Science of Joining, (The
Minerals, Metals & Materials Society, 1992), pp. 175-181.
[4] W. J. Reichenecker, Insulation/Circuits, 27, 104 (1981).
[5] National Institute of Standard and Technology, Gaithersburg, MD 20899.
[6] K. R. Stone, R. Duckett, S. Muckett, and M. E. Warwick, Brazing and Soldering, 4, 20
(1983).
[7] Y. S. Touloukian, Thermal expansion, metallic elements and alloys, (IFI/Plenum, 1975).
4