Materials Transactions, Vol. 45, No. 3 (2004) pp. 765 to 775
Special Issue on Lead-Free Soldering in Electronics
#2004 The Japan Institute of Metals
Development of Sn-Based, Low Melting Temperature Pb-Free Solder Alloys
Paul Vianco, Jerome Rejent and Richard Grant
Sandia National Laboratories, Albuquerque, NM 87185, USA
Low temperature, Sn-based Pb-free solders were developed by making alloy additions to the starting material, 96.5Sn-3.5Ag (mass%). The
melting behavior was determined using Differential Scanning Calorimetry (DSC). The solder microstructure was evaluated by optical
microscopy and electron probe microanalysis (EPMA). Shear strength measurements, hardness tests, intermetallic compound (IMC) layer
growth measurements, and solderability tests were performed on selected alloys. Three promising ternary alloy compositions and respective
solidus temperatures were: 91.84Sn-3.33Ag-4.83Bi, 212 C; 87.5Sn-7.5Au-5.0Bi, 200 C; and 86.4Sn-5.1Ag-8.5Au, 205 C. A quaternary alloy
had the composition 86.8Sn-3.2Ag-5.0Bi-5.0Au and solidus temperature of 194 C. The shear strength of this quaternary alloy was nearly twice
that of the eutectic Sn-Pb solder. The 66Sn-5.0Ag-10Bi-5.0Au-10In-4.0Cu alloy had a solidus temperature of 178 C and good solderability on
Cu. The lowest solidus temperature of 159 C was realized with the alloy 62Sn-5.0Ag-10Bi-4.0Au-10In-4.0Cu-5.0Ga. The contributing factor
towards the melting point depression was the composition of the solid solution, Sn-based matrix phase of each solder.
(Received October 2, 2003; Accepted December 11, 2003)
Keywords: lead-free solder, tin-based, low melting temperature
1.
Introduction
At present, the Sn-Ag-Cu, Sn-Ag-Bi and Sn-Ag-Bi-Cu
alloy families are at the forefront of Pb-free solder research
and development activities. However, these solders have
significantly higher melting temperatures than has the
eutectic 63Sn-37Pb (mass%) alloy that melts at 183 C. The
higher process temperatures raise the following concerns: (1)
thermal damage to components (e.g., the plastic molding
compounds of surface mount devices) and laminate materials; (2) an increased tenacity of flux residues; and (3)
increased maintenance costs for soldering equipment.1–4)
Lead-free solder studies were performed in the early to
middle 1990s in response to bills proposed in the United
States Congress that sought to ban or heavily tax Sn-Pb
solders used in electronics.5–7) Product development engineers desired an alternative solder having a lower melting
temperature than 63Sn-37Pb in order to use less expensive
printed circuit board laminate materials and component
molding compounds.
Off-the-shelf solders were limited that possessed a melting
temperature similar to, or lower than, that of the 63Sn-37Pb
alloy. The eutectic composition 91Sn-9Zn has a melting
temperature of 199 C.8) However, the Zn component caused
poor solder paste shelf life, processing, and long-term
reliability properties; albeit, several companies have demonstrated the successful assembly of printed circuit cards with
the 91Sn-9Zn solder. Bismuth additions have been explored
to improve the performance of Sn-Zn solders.9)
The 58Bi-42Sn alloy has received the greatest attention as
a low melting temperature, Pb-free alternative to 63Sn-37Pb
solder.10–13) The melting temperature of the eutectic 58Bi42Sn solder is 139 C.14) The solderability properties of this
composition are generally very good.15) However, the 58Bi42Sn alloy has several drawbacks. First, the 139 C melting
temperature is generally less than the activation temperatures
of current flux materials. The higher processing temperatures
that would be required to activate the solder paste flux would
lessen the advantage of the low melting temperature of the
solder. Second, the relative brittleness of the 58Bi-42Sn alloy
poses a reliability concern for low-cost consumer electronics
because such products are often subjected to mechanical
shock.
A research and development project was begun, the
objective of which was to develop a Sn-based, Pb-free alloy
that possessed a solidus temperature that was equal to, or less
than, the 183 C eutectic temperature of the 63Sn-37Pb
solder. It was also desirable to minimize the pasty range of
the alloy, preferably to less than ten Celsius degrees. The
base material for the alloy development effort was the
96.5Sn-3.5Ag solder. Ternary, quaternary, and higher order
alloys were developed through the addition of other metal
elements. The Differential Scanning Calorimeter (DSC) was
used to identify the solidus temperature of each successive
alloy composition. Project constraints prohibited a strict
analysis of the liquidus temperature for each composition.
Rather, the DSC peak width provided a qualitative metric of
the proximity of the liquidus temperature to the solidus
temperature. The development effort used metallographic
cross sections to document alloy microstructures and electron
probe microanalysis (EPMA) to determine phase compositions. Several ancillary experiments were also performed on
promising compositions to evaluate solid state intermetallic
compound layer growth, hardness, and shear strength.
2.
Experimental Procedures
2.1 Alloy fabrication techniques
The alloys were fabricated on a hot plate in an air
environment. A 0.5 kg charge of 96.5Sn-3.5Ag (mass%)
alloy was melted into a stainless steel crucible at 350 C to
400 C. Dissolution of the crucible structure was not
observed. Elemental additions were made to the molten
96.5Sn-3.5Ag solder in the form of wire, strip, or powders.
Atomic emission spectroscopy/inductively coupled plasma
(AES/ICP) was used to confirm the composition of selected
alloys during the study.
The alloy specimens that were used for microscopy and
phase identification were first exposed to the DSC thermal
schedule to assure that the respective microstructures that
P. Vianco, J. Rejent and R. Grant
were developed by a consistent cooling rate. Phase identification was performed on the sample cross sections using
electron probe microanalysis (EPMA). The electron beam
was operated at 15 keV and a beam current of 21 nA. Several
particles of similar morphology were evaluated in order to
eliminate x-ray sample volume effects.
(Note: Coefficients placed in front of the element indicate
mass%; subscripts represent atomic%.)
2.2 Differential Scanning Calorimeter measurements
The Differential Scanning Calorimeter (DSC) was used to
determine the solidus temperature of each alloy composition.
Sample masses ranged from 5 to 20 mg. Both heating and
cooling rates were 10 C/min. In several cases, alloy specimens were also evaluated with 1 C/min heating and cooling
rates to assure the absence of low temperature or satellite
peaks. No differences were observed between the two
heating/cooling rates. The samples were scanned between
25 C and 300 C. Two heating and cooling cycles were
performed. The second cycle was used to generate the actual
melting properties data.
The solidus temperature of the alloy was determined from
the heating portion of the DSC cycle, using the following
convention. First, a ‘‘baseline’’ was constructed under the
transformation peak. Then, a second line was constructed at
the inflection point of the leading edge of the transformation
peak, having the same slope as that of the trace at the
inflection point. The solidus temperature was designated by
the intersection between the latter line and the baseline.
Because the objective of the present study required that a
large number of alloy variations be explored, the exact
determination of the liquidus temperature was omitted in
favor of a qualitative assessment based upon the shape and
width of the transformation peak.
3.
120
100
Heat Flow, P/(mW)
766
80
60
40
20
0
150
250
200
Temperature, T/°C
Fig. 1 DSC thermogram of the 96.5Sn-3.5Ag solder. The heating rate was
10 C/min.
Sn-rich
matrix
Ag3Sn
particles
Results and Discussion
3.1 Ternary compositions
The DSC thermogram of the 96.5Sn-3.5Ag solder is shown
in Fig. 1. The alloy has a eutectic temperature of 221 C; the
heat of transformation or heat of melting, H, was 66 J/g.
The optical micrograph in Fig. 2 illustrates the 96.5Sn-3.5Ag
microstructure. There is a Sn-rich phase (dendrite arms)
between which are regions of finely dispersed Ag3 Sn
particles in a Sn-rich matrix. Initially, the same Ag/Sn ratio
of 0.036 was kept in subsequent, higher order alloys in order
to maximize the melting point depression role of Ag.
Three ternary solders were developed from the Sn-Ag
binary alloy; those compositions were: (1) 91.84Sn-3.33Ag4.83Bi, (2) 87.5Sn-7.5Au-5.0Bi, and (3) 86.4Sn-5.1Ag8.5Au. The DSC thermogram of the 91.84Sn-3.33Ag4.83Bi alloy is shown in Fig. 3. This alloy has a solidus
temperature of 212 C, a H of 54 J/g, and a relatively
narrow peak width. The microstructure of the 91.84Sn3.33Ag-4.83Bi solder, which is shown in the SEM/backscattered electron (BSE) image in Fig. 4, has a 96Sn-4Bi
solid solution matrix; Ag3 Sn particles; and Bi particles
grouped in clusters or that decorate the Ag3 Sn particles. The
4.83 mass% Bi addition maximized the decrease in the
solidus temperature without forming 58Bi-42Sn regions
100 µ m
Fig. 2 Optical micrograph of the 96.5Sn-3.5Ag solder. The cooling rate
was 10 C/min.
having a solidus temperature of 139 C.16) The 91.84Sn3.33Ag-4.83Bi alloy has been the subject of several materials
and assembly process studies.17–20)
The 87.5Sn-7.5Au-5.0Bi and 86.4Sn-5.1Ag-8.5Au compositions illustrated the role of Au as a melting temperature
depressant. Gold was selected under the presumption that it
would have similar melting point depression properties as Ag
because both combine in solid-solution.21,22) The 87.5Sn6.0Au-6.5Bi composition had a relatively narrow DSC peak
and a solidus temperature of 212 C. The 86.5Sn-9.0Au-4.5Bi
composition had a narrower transformation peak and a
solidus temperature of 202 C. The optimum ternary composition, 87.5Sn-7.5Au-5.0Bi had a solidus temperature of
200 C and a H of 50 J/g; the DSC thermogram is shown in
Fig. 5. The SEM/BSE photograph in Fig. 6 shows the
microstructure of the 87.5Sn-7.5Au-5.0Bi alloy. There was a
Development of Sn-Based, Low Melting Temperature Pb-Free Solder Alloys
767
50
70
60
Heat Flow, P/(mW)
Heat Flow, P/(mW)
40
50
40
30
20
30
20
10
20
0
150
200
250
Temperature, T/°C
Fig. 3 DSC thermogram of the 91.84Sn-3.33Ag-4.83Bi solder. The
heating rate was 10 C/min.
96Sn-4Bi
0
200
150
250
Temperature, T/°C
Fig. 5 DSC thermogram of the 87.5Sn-7.5Au-5.0Bi solder. The heating
rate was 10 C/min.
AuSn4
Bi
96Sn-4Bi
Ag3Sn
Bi
AuSn4
Fig. 4 SEM/BSE photograph of the 91.84Sn-3.33Ag-4.83Bi solder. The
cooling rate was 10 C/min.
96Sn-4Bi solid solution matrix phase; a AuSn4 phase in the
form of blocky particles and needles; and elemental Bi
particles.
The elements Ag and Au were added to Sn to form a SnAg-Au ternary alloy, resulting in the 86.4Sn-5.1Ag-8.5Au
composition that was confirmed by AES-ICP to within 95%
confidence intervals of Sn, 1.5 (mass%); Ag, 0.33; and Au,
0.06. The solidus temperature and H were 205 C and 55 J/
g, respectively; the corresponding DSC thermogram is shown
in Fig. 7. The melting point depression capabilities of Au
were documented. The 86.4Sn-5.1Ag-8.5Au microstructure
is shown in Fig. 8. There was a 96Sn-4Bi matrix phase;
needle-shaped and blocky AuSn4 particles; long needles of
the Ag3 Sn phase (inset photograph); and smaller needles of
Ag4 Sn3 . The latter stoichiometry was not observed in the
Fig. 6 SEM/BSE photograph of the 87.5Sn-7.5Au-5.0Bi solder. The
cooling rate was 10 C/min.
equilibrium Ag-Sn binary alloy phase diagram.23) The fact
that the Ag4 Sn3 stoichiometry was identified in similar
particles of other compositions lent credibility to this finding.
The high Sn, Au containing Pb-free solders may have
applications in hybrid microcircuit (HMC) assemblies. It was
hypothesized that the Au content would reduce the rate of
thick (or thin) film dissolution by the molten alloy, as well as
lower the rate of solid-state intermetallic compound layer
growth in HMCs. Studies are underway to substantiate these
hypotheses.
768
P. Vianco, J. Rejent and R. Grant
50
100
80
Heat Flow, P/(mW)
Heat Flow, P/(mW)
40
60
40
20
10
20
0
30
200
150
250
Temperature, T/°C
0
150
250
200
Temperature, T/°C
Fig. 7 DSC thermogram of the 86.4Sn-5.1Ag-8.5Au solder. The cooling
rate was 10 C/min.
Fig. 9 DSC thermogram of the 86.8Sn-3.2Ag-5.0Bi-5.0Au solder. The
heating rate was 10 C/min.
96Sn-4Bi
Bi
Ag3Sn
AuSn4
96Sn-4Bi
Ag3Sn
Ag4Sn3
AuSn4
Fig. 10 SEM/BSE photograph of the 86.8Sn-3.2Ag-5.0Bi-5.0Au solder.
The cooling rate was 10 C/min.
Fig. 8 SEM/BSE photograph of the 86.4Sn-5.1Ag-8.5Au solder. The
cooling rate was 10 C/min.
3.2 Quaternary compositions
Potential quaternary compositions were developed with
Ag, Au, and Bi additions to Sn. The 86.8Sn-3.2Ag-5.0Bi5.0Au composition had a solidus temperature of 194 C and a
H equal to 53 J/g; however, its relatively wide transformation peak (Fig. 9) indicated a large pasty range. The
second alloy, 84.5Sn-3.0Ag-5.0Bi-7.5Au, had a solidus
temperature of 215 C and narrower transformation peak.
Because of the lower solidus temperature, the 86.8Sn-3.2Ag5.0Bi-5.0Au was selected for further analysis. Shown in
Fig. 10 is an SEM/BSE photograph showing the 86.8Sn3.2Ag-5.0Bi-5.0Au microstructure. The matrix phase was a
solid solution of 96Sn-4Bi. Distributed in the matrix were:
blocky and needle particles of AuSn4 ; particles of Ag3 Sn;
and lastly elemental Bi particles. Further variations of the Sn,
Ag, Bi, and Au contents did not improve the alloy properties.
Several ancillary experiments were performed on the
benchmark 86.8Sn-3.2Ag-5.0Bi-5.0Au composition and the
two ternary alloys, 86.4Sn-5.1Ag-8.5Au and 87.5Sn-7.5Au5.0Bi. Shown in Table 1 is the Knoop microhardness of these
solders as well as the 96.5Sn-2.5Ag, 91.84Sn-3.33Ag-4.83Bi,
and 63Sn-37Pb alloys. The minor differences between
hardness numbers generated by 25 g and 50 g suggested that
values obtained with a 15 g load should be comparable. The
three Au-containing solders had hardness levels that exceeded that of the 96.5Sn-3.5Ag composition. Two of three
solders were harder than the 91.84Sn-3.33Ag-4.83Bi composition.
Development of Sn-Based, Low Melting Temperature Pb-Free Solder Alloys
Table 1 Knoop microhardness (15 s) of selected solder alloys.
Solder alloy (mass%)
Knoop microhardness
Table 2
769
Ring-in-plug shear strength of selected solder alloys.
Solder alloy (mass%)
Ring-in-plug shear strength (MPa)
86.4Sn-5.1Ag-8.5Au (15 g)
23 3
87.5Sn-5.0Bi-7.5Au
80 2
87.5Sn-5.0Bi-7.5Au (15 g)
86.8Sn-3.2Ag-5.0Bi-5.0Au (15 g)
38 3
40 2
86.8Sn-3.2Ag-5.0Bi-5.0Au
96.5Sn-3.5Ag
84 2
55 1
96.5Sn-3.5Ag (25 g)
15 2
60Sn-40Pb
40 2
96.5Sn-3.5Ag (50 g)
17:4 0:7
91.84Sn-3.33Ag-4.83Bi
80 10
60Sn-40Pb (25 g)
13:2 0:6
60Sn-40Pb (50 g)
15 2
91.84Sn-3.33Ag-4.83Bi (50 g)
26 1
The shear strength of 87.5Sn-7.5Au-5.0Bi and 86.8Sn3.2Ag-5.0Bi-5.0Au solder joints was determined by the ringand-plug test (10 mm/min). The joint gap was 0.178 mm.
The joint microstructures are shown in Figs. 11(a) and 12(a).
The shear strengths of the 86.8Sn-3.2Ag-5.0Bi-5.0Au and
87.5Sn-7.5Au-5.0Bi solders appear in Table 2 (four tests per
composition). The two Au-containing solders had exceptionally high strengths. The failure paths were located in the bulk
solder for both alloys (Figs. 11(b) and 12(b)) and indicated
relatively little ductility.
(a)
Solid state intermetallic compound growth experiments
were performed on Cu couples made with the 87.5Sn-7.5Au5.0Bi and 86.8Sn-3.2Ag-5.0Bi-5.0Au solders. The aging
conditions were 170 C for 40 days and 100 days. Shown in
Figs. 13(a) and (b) are optical micrographs of the 87.5Sn7.5Au-5.0Bi/Cu and 86.8Sn-3.2Ag-5.0Bi-5.0Au/Cu couples, respectively, after aging for 100 days (Similar structures
were observed after 40 days; they were simply thinner.).
Electron probe microanalysis traces identified multiple sublayer compositions that are documented schematically in
Fig. 14. The Cu3 Sn sub-layer adjoined the Cu substrate. The
Cux Auy Snz sub-layer located between Cu3 Sn layer and the
87.5Sn-7.5Au-5.0Bi solder changed composition as a func-
(b)
Cu ring
Cu ring
87.5Sn-7.5Au-5.0Bi
87.5Sn-7.5Au-5.0Bi
Crack
Cu plug
100 µ m
20 µ m
Fig. 11 (a) Optical micrograph of the gap of the ring-and-plug test sample fabricated with the 87.5Sn-7.5Au-5.0Bi solder. (b) The failure
path of the tested sample was located in the bulk solder.
(a)
(b)
Cu ring
Cu ring
86.8Sn-3.2Ag-5.0Au-5.0Bi
86.8Sn-3.2Ag-5.0Au-5.0Bi
Crack
Cu plug
100 µ m
20 µm
Fig. 12 (a) Optical micrograph of the gap of the ring-and-plug test sample fabricated with the 86.8Sn-3.2Ag-5.0Au-5.0Bi solder. (b) The
failure path of the tested sample was located in the bulk solder.
770
P. Vianco, J. Rejent and R. Grant
(b)
(a)
87.5Sn-7.5Au-5.0Bi
86.8Sn-3.2Ag-5.0Au-5.0Bi
(CuAu) Sn
(Cu33Au)66Sn55
(Cu33Au)66Sn55
Cu33Sn
Cu33Sn
Cu
Cu
20 µ m
20 µm
Fig. 13 Optical micrographs of: (a) 87.5Sn-7.5Au-5.0Bi/Cu couple and (b) 86.8Sn-3.2Ag-5.0Au-5.0Bi/Cu, both following solid-state
aging at 170 C for 100 days.
Table 3
Intermetallic compound layer thicknesses for an aging temperature of 170 C and aging times of 40 and 100 days.
Intermetallic compound layer (mm)
Aging time
Solder alloy
(days)
(mass%)
40
100
Cu3 Sn
Cu6 Sn5 or
Cu-Au-Sn
Total
87.5Sn-5.0Bi-7.5Au
3:2 0:6
7:8 1:4
11:0 1:2
86.8Sn-3.2Ag-5.0Bi-5.0Au
96.5Sn-3.5Ag
2:7 0:6
2:7 0:5
8:8 2:3
8:6 2:3
11:5 2:0
11:3 2:6
100Sn
5:5 1:1
5:3 2:7
10:8 2:6
87.5Sn-5.0Bi-7.5Au
4:5 0:9
11:3 2:8
15:7 2:7
86.8Sn-3.2Ag-5.0Bi-5.0Au
4:5 0:8
11:8 2:7
16:3 2:6
96.5Sn-3.5Ag
4:2 0:8
11:0 2:9
15:2 2:8
100Sn
7:7 1:4
7:3 2:6
15:0 2:3
40 days
Cu
Cu33Sn
100 days
Cu
Cu33Sn
(Cu55Au22)66Sn55
(Cu33Au)66Sn55
(CuAu) Sn
87.5Sn7.5Au5.0Bi
87.5Sn7.5Au5.0Bi
40 days
Cu
Cu33Sn
(Cu33Au)66Sn55
86.8Sn3.2Ag5.0Au-5.0Bi
100 days
Cu
Cu33Sn
(Cu33Au)66Sn55
86.8Sn3.2Ag5.0Au-5.0Bi
Fig. 14 Schematic diagram of the intermetallic compound sub-layer
compositions for the 87.5Sn-7.5Au-5.0Bi/Cu and 86.8Sn-3.2Ag-5.0Au5.0Bi/C couples aged at 170 C for 40 days and 100 days. (Layer
thicknesses representations are not to scale.)
tion of the aging conditions; the layer composition remained
unchanged for the 86.8Sn-3.2Ag-5.0Bi-5.0Au/Cu couples.
The Cux Auy Snz sub-layers had concentration gradients of Cu
and Au that decreased and increased, respectively, from the
Cu substrate to the solder field due the mutual solubility of
Au and Cu. The thicknesses of the sub-layers and total layers
were compared to those of 96.5Sn-3.5Ag/Cu and 100Sn/Cu
couples in Table 3. All of the layer thickness values were
nearly identical to those of the 96.5Sn-3.5Ag binary alloy.
Thus, the presence of Au in one sub-layer did not have a
significant impact on the growth rate of that sub-layer or the
total layer thickness.
3.3 Fifth- and sixth-order compositions
Additions of In and Cu additions were made to the 86.8Sn3.2Ag-5.0Bi-5.0Au composition. The thermograms were
scrutinized for the 118 C peak of the potential 52In-48Sn
phase.24) Initially, 5 mass% In was added at the expense of Sn
and Ag; the Ag/Sn ratio was 0.038. The resulting alloy had
the composition 81.9Sn-3.1Ag-5.0Bi-5.0Au-5.0In. The In
addition did not significantly alter the solidus temperature
(194 C) nor the peak width vis-á-vis the 86.8Sn-3.2Ag5.0Bi-5.0Au alloy. Low temperature peaks did not appear in
the DSC thermogram that would have been associated with
eutectic In-Sn or Bi-Sn regions in the solder.
Copper additions were made to the 81.9Sn-3.1Ag-5.0Bi5.0Au-5.0In composition, resulting in the sixth-order composition 80.1Sn-2.9Ag-5.0Bi-5.0Au-5.0In-2.0Cu. The solidus temperature was 191 C; the peak width was slightly
reduced as compared to that of the Sn-Ag-Bi-Au-In alloy.
Also, there was no appearance of low temperature peaks.
The 80.1Sn-2.9Ag-5.0Bi-5.0Au-5.0In-2.0Cu alloy was
further modified by varying Bi and In between the combinations of: (1) Bi:10, In:5; (2) Bi:10, In:10; and (3) Bi:5,
Development of Sn-Based, Low Melting Temperature Pb-Free Solder Alloys
In:10. It was observed that Bi was more effective at
decreasing the solidus temperature than was In. Combining
10 mass% Bi and 10 mass% In with 2 mass% Cu decreased
the solidus temperature to 180 C, a value that was similar to
the eutectic temperature of the 63Sn-37Pb solder. The
transformation peak was wider as compared to the peaks of
the 86.9Sn-3.2Ag-5.0Bi-5.0Au and 80.1Sn-2.9Ag-5.0Bi5.0Au-5.0In-2.0Cu compositions. However, the peak broadened towards the low-temperature values without the development of a high-temperature shoulder or satellite peaks.
More importantly, low temperature peaks associated with the
higher Bi and In concentrations did not appear in the DSC
thermograms. The final composition was 70.4Sn-2.6Ag10.0Bi-5.0Au-10.0In-2.0Cu having a solidus temperature of
180 C and H equal to 31 J/g.
At this point, the need to maintain the Ag/Sn mass% ratio
at approximately 0.036 was examined. It was confirmed that,
given the highly complex microstructure of these higher
order alloys, the melting temperature depressant function of
Ag had become masked, even when the value was doubled
and then tripled with no effect on the solidus temperature or
peak width.
The Cu concentration was increased to 4 mass%. The Bi
and In additions were maximized at 10 mass% each. The
resulting solder composition was 66Sn-5.0Ag-10Bi-5.0Au10In-4.0Cu. The DSC thermogram appears in Fig. 15. No
low temperature peaks were observed and the peak width was
largely unchanged from its predecessor compositions. The
solidus temperature was 178 C. The heat of transformation,
H, was 31 J/g. Upon cooling the alloy (10 C/min), the
DSC peak was relatively narrow, indicating that solidification of all of the phases occurred at nearly the same time.
The microstructure of the 66Sn-5.0Ag-10Bi-5.0Au-10In-
100
Heat Flow, P/(mW)
80
60
40
20
0
125
175
225
Temperature, T/°C
Fig. 15 DSC thermogram (10 C/min heating) of the alloy 66Sn-5.0Ag10Bi-5.0Au-10In-4.0Cu. The onset temperature ranged from 177 C to
179 C.
771
100 µ m
Fig. 16 Optical micrograph showing the microstructure of the 66Sn5.0Ag-10Bi-5.0Au-10In-4.0Cu solder. The cooling rate was 10 C/min.
4.0Cu solder is illustrated by the optical micrograph in
Fig. 16. A large number of small, evenly distributed particle
phases were generated in the microstructure. The phase
compositions were determined by EPMA and labeled in
Fig. 17. The matrix phase had the composition of 88Sn6.5Bi-5.5In. Bismuth was present as elemental particles as
well as BiIn intermetallic compound particles. The 110 C
solidus temperature of the BiIn compound was not observed
in the DSC thermogram. One of the larger particle phases had
the composition (Au,Ag,Cu)3 (Sn,In)2 . The Au, Ag, and Cu
were present in an atomic percent ratio of 11:2.5:1,
respectively. The Sn and In were present in a ratio of atomic
percents equal to 1:2. Therefore, this particle appeared to be
based upon the AuIn stoichiometry, but with the presence of
Ag and Cu due to their mutual solubility with Au, and the
presence of Sn due to its mutual solubility with In. The other
particle phase was Ag3 (Sn,In). Tin and In were present at
nearly equal amounts — 10 at% and 13 at%, respectively —
due to their mutual solubility. This particle is based upon the
Ag3 Sn stoichiometry.
An interesting phase in the 66Sn-5.0Ag-10Bi-5.0Au-10In4.0Cu solder was that of particles that had a dark contrast in
the core and light contrast around the edges (Fig. 17). The
EPMA determined that there was no significant concentration
differences between the two regions. Both the edge and core
regions had the stoichiometry, (Cu8 Au)6 (Sn6 In)5 that resembled Cu6 Sn5 . The mutual solubility of Au in Cu and In in Sn
resulted in the presence of Au and In in the particle
composition.
The 66Sn-5.0Ag-10Bi-5.0Au-10In-4.0Cu alloy appeared
to be a promising, Pb-free alternative for second-level printed
wiring assemblies. Sessile drop experiments were performed
on Cu substrates, using a rosin-based, mildly activated
(RMA) flux; a test temperature of 250 C, and hold times of
15 s and 30 s. The contact angles that were measured for the
two time periods were 23 3 and 18 3 , respectively.
More rigorous meniscometer/wetting balance tests were also
performed, using similar test parameters.25) The measured
contact angle and solder-flux interfacial tension were
34:2 0:7 and 418 9 dynes/cm, respectively. These
772
P. Vianco, J. Rejent and R. Grant
(Au, Ag)3(Sn, In)2
Bi
88Sn-6.5Bi-5.5In
(Cu, Au) (Sn, In)
BiIn
Ag3(Sn, In)
Fig. 17 SEM/BSE photographs identifying the phases that formed in the 66Sn-5.0Ag-10Bi-5.0Au-10In-4.0Cu solder. The cooling rate
was 10 C/min.
Table 4 Solderability parameters for the 66Sn-5.0Ag-10Bi-5.0Au-10In4.0Cu and other high-Sn, Pb-free solders (RMA flux, 250 C).
Solder alloy
Contact angle
Solder-flux interfacial
(mass%)
( )
tension (dynes/cm)
66Sn-5.0Ag-10Bi-5.0Au
34:2 0:7
418 9
96.5Sn-3.5Ag
36 3
460 30
95.5Sn-3.9Ag-0.6Cu
40 5
500 40
91.84Sn-3.33Ag-4.83Bi
31 5
420 30
10In-4.0Cu
values are very similar to those of other high-Sn, Pb-free
solders as illustrated in Table 4.26)
The solid-state intermetallic compound growth behavior
was studied between the 66Sn-5.0Ag-10Bi-5.0Au-10In4.0Cu solder and Cu substrates. The aging experiments that
were performed with temperatures of 70 C, 100 C, 135 C,
and 170 C and aging times of 10 days, 40 days, and 100 days.
The composition of the intermetallic compound layer was
determined by EPMA. The intermetallic compound layer
exhibited a single composition, (Cu, Au)6 (Sn, In)5 , that is
based upon the Cu6 Sn5 compound. There was a limited
presence of Au and In as indicated by the ratios for Cu:Au
and Sn:In of 19:1 and 17:1, respectively.
The rate kinetics of intermetallic compound layer growth
were determined by a multivariable linear regression analysis
of thickness data, using the phemenological equation:
y ¼ yo þ Atn exp½H=RT
ð1Þ
where y is the intermetallic compound layer thickness (m) at
time (s), t; yo is the initial layer thickness, 1:92 106 m; A
is a constant; n is the time exponent; H is the apparent
activation energy (J/mol); R is the universal gas constant
(8.314 J/mol-K); and T is temperature (K). The regression
analysis was performed with lnðy yo Þ as the dependent
variable and lnðtÞ and (1=T) as the independent variables; the
coefficients were lnðAÞ, n, and H=R. The kinetics parameters A, n, and H are listed in Table 5, along with those of
other high-Sn, Pb-free solders.27) The time exponent indicated that the layer growth was based upon a diffusioncontrolled mechanism. The similar values of both n and H
amongst the alloys in Table 5 indicated that the solid-state
intermetallic compound layer growth was relatively insensitive to the widely different solder compositions and microstructures.
Several qualitative observations were made from the
higher order alloys that were developed to this point. First,
the various phases usually precipitated into uniform distributions of relatively small particles throughout the Sn-based
matrix phase, thereby avoiding the development of large
crystal structures that can potentially embrittle complex
alloys. Secondly, the low temperature phases were avoided in
spite of the relatively high concentrations of Bi and In.
Several pseudo-quantitative trends were also documented
during the course of arriving at the 66Sn-5.0Ag-10Bi-5.0Au10In-4.0Cu alloy. (1) An increase in the Au concentration
from 5 mass% to 7.5 mass% had no significant effect on the
solidus temperature and served to only widen the melting
Table 5 Solid-state intermetallic compound growth kinetics parameters for 66Sn-5.0Ag-10Bi-5.0Au-10In-4.0Cu and other high-Sn, Pbfree solders (RMA flux, 250 C). The error term is a 95% confidence interval.
Solder alloy
A
(mass%)
(m/sn )
66Sn-5.0Ag-10Bi-5.0Au-
Time exponent, n
Apparent activation
energy, H (kJ/mol)
0.161
0:51 0:35
59 13
10In-4.0Cu
96.5Sn-3.5Ag
1:83 106
0:50 0:30
57 8
95.5Sn-3.9Ag-0.6Cu
1:05 103
0:58 0:08
50 4
91.84Sn-3.33Ag-4.83Bi
0:57 106
0:46 0:15
49 9
Development of Sn-Based, Low Melting Temperature Pb-Free Solder Alloys
3.4 Seventh- and higher-order compositions
Further alloy additions were made to the 66Sn-5.0Ag10Bi-5.0Au-10In-4.0Cu composition in an effort to decrease
the solidus temperature. Those studies, which focused on the
elements Fe, Ni, Al, Zn, and Ga, were less comprehensive
than the investigations of the previous solders, being limited
to primarily DSC and microstructural evaluations. The
additions were made generally at the expense of the Sn
content and in some cases, albeit to a lesser extent, the Au
content.
Two-to-three mass% additions of Fe caused a one-to-two
Celsius degree decrease in the solidus temperature. The phase
distribution in the 64Sn-5.0Ag-10Bi-4.0Au-10In-4.0Cu3.0Fe composition (solidus temperature of 177 C and H
of 24 J/g) was very similar to that of the 66Sn-5.0Ag-10Bi5.0Au-10In-4.0Cu alloy with two notable exceptions. First,
the matrix phase had a higher solid solution concentration of
Bi and In: 82Sn-10Bi-8In. Second, FeSn2 particles were
uniformly distributed within the matrix. No low temperature
peaks were observed in the DSC thermograms. This alloy
was particularly difficult to make due to the high melting
temperature of Fe and its relatively slow dissolution into
molten Sn.
The addition of Ni resulted in the composition, 64Sn5.0Ag-10Bi-4.0Au-10In-4.0Cu-3.0Ni, having a solidus temperature of 174 C and H of 25 J/g. The Ni additions caused
the formation of long (Cu, Ni)x Sny particles. This Ni-bearing
solder was difficult to fabricate because, like the Fe additions,
Ni has a high melting temperature and low dissolution rate in
molten Sn.
The addition of 3 mass% Al resulted in the alloy, 64Sn5.0Ag-10Bi-5.0Au-10In-4.0Cu-2.0Al, having a liquidus temperature of 173 C and H of 19 J/g. Unfortunately, the peak
width increased by approximately three Celsius degrees over
that of the basis alloy.
Next, the Zn additions were studied. Two mass% Zn was
added in lieu of either 2 mass% Sn or 2 mass% Cu. The
resulting solidus temperatures and peak widths were similar
to those of the baseline 66Sn-5.0Ag-10Bi-5.0Au-10In-4.0Cu
solder. Interestingly enough, the addition of 4 mass% Zn and
the elimination of Au resulted in the 67Sn-5.0Ag-10Bi-10In4.0Cu-4.0Zn composition that had a solidus temperature of
178 C (H ¼ 30 J/g). All of the Zn-containing solders
showed no low temperature peaks. The microstructures had a
relatively uniform distributions of particles. Meniscometer/
wetting balance solderability tests (RMA flux, 250 C, 30 s)
were performed on Cu with the 66Sn-5.0Ag-10Bi-5.0Au-
100
80
Heat Flow, P/(mW)
peak. In fact, the higher Au content diminished the capability
of Cu additions to depress the solidus temperature. (2)
Changing the In concentration from 5 mass% to 10 mass%
had little impact on the solidus temperature. However, in a
synergistic effect, that 5–10 mass% In content allowed Bi
additions of 5–10 mass% to decrease the solidus temperature
by as much as ten Celsius degrees without the appearance of
a low temperature peak on the DSC thermogram. (3) The
melting properties became less sensitive to the ratio of Ag/
Sn. Also, maintaining the Ag content to less than 5 mass%
enhanced the ability of the Cu additions to decrease the
solidus temperature.
773
60
40
20
0
125
175
225
Temperature, T/°C
Fig. 18 DSC thermogram of the 62Sn-5.0Ag-10Bi-4.0Au-10In-4.0Cu5.0Ga solder. The heating rate was 10 C/min.
10In-2.0Cu-2.0Zn and 67Sn-5.0Ag-10Bi-10In-4.0Cu-4.0Zn
solders. The contact angles were 29 5 and 26 1 ,
respectively, indicating relatively good solderability.28)
Lastly, the element Ga was investigated. A 2 mass% Ga
addition had very little impact on the solidus temperature and
peak width of the resulting alloy. The DSC thermogram
exhibited an absence of low-temperature peaks. The microstructure had a relatively uniform distribution of particles.
The quantity of Ga was increased to 5 mass%, resulting in the
solder composition 62Sn-5.0Ag-10Bi-4.0Au-10In-4.0Cu5.0Ga. The solidus temperature was measured at 159 2 C. The DSC peak (Fig. 18) was very shallow (Hmelting of
25–27 J/g) and broadened towards the low temperatures;
there were no shoulders appearing on the transformation peak
nor were low temperature peaks in the DSC thermogram. The
alloy exhibited very good solderability on Cu (RMA flux,
250 C, 30 s), having a contact angle of 22 3 . This alloy
exhibited very low ductility as is illustrated by the optical
micrograph in Fig. 19 in which there was considerable
particle pull out from the cross sections. An EPMA
determination of phase compositions was not performed for
this alloy.
Role of alloy additions on solidus temperature
A similarity was observed between the solidus temperature
and Sn contents of the multi-component, Pb-free alloys and
those of the eutectic Sn-Pb solder. The combined quantities
of the alloy additions, Ag, Bi, Au, In, etc., were as effective as
an equivalent amount of Pb in terms of depressing the solidus
temperature of Sn.
The microstructures of the multicomponent solders were
comprised of intermetallic compound particles and a Snbased, matrix phase. The matrix phase was either pure Sn, as
in the case of the baseline alloy 96.5Sn-3.5Ag, or a solid
solution of Bi, or Bi and In, in Sn. Because the melting
774
P. Vianco, J. Rejent and R. Grant
ln(1/x) (x = mole fraction of Sn)
5.0
Particle
pull-out
4.9
4.8
4.7
4.6
4.5
0
0.0 10
-4
1.0 10
1.5 10
-1
T -T
100 µ m
Fig. 19 Optical micrograph of the 62Sn-5.0Ag-10Bi-4.0Au-10In-4.0Cu5.0Ga solder. The cooling rate was 10 C/min.
temperatures of the intermetallic compounds were typically
higher than that of the matrix phase, it was construed that the
composition of the matrix phase may have been primarily
responsible for melting point (solidus temperature) depression observed with the complex alloy compositions. In the
event that this hypothesis was accurate and the matrix phase
behaved as an ideal solid solution, then the solidus temperature (1=Ts, alloy ) and the mole fraction of the solvent element
Sn (x) would be related according to the following equation:29)
lnð1=xÞ ¼ H=Rð1=Ts, alloy 1=Tmelt, Sn Þ
-5
5.0 10
ð2Þ
where H is the heat of transformation of the solvent
component Sn, 59 J/g;30) R is the universal gas constant
(8.314 J/mol-K), and Tmelt, Sn is the melting temperature of
the pure Sn (505 K). The slope of a linear regression analysis
executed between the dependent variable lnð1=xÞ and the
independent variable (1=Ts, alloy 1=Tmelt, Sn ) returns the
value of H=R. Therefore, the appropriateness of the
aforementioned assumptions would be substantiated if the
value of H calculated from the experimental data and
equation (2) was equivalent to 59 J/g.
The Sn-based, matrix phase composition was determined
for the following alloys: 96.5Sn-3.5Ag; the three ternary
alloys; the 86.8Sn-3.2Ag-5.0Bi-5.0Au quaternary alloy, as
well as the 66Sn-5.0Ag-10Bi-5.0Au-10In-4.0Cu and 64Sn5.0Ag-10Bi-4.0Au-10In-4.0Cu-3.0Fe alloys. The regression
analysis returned a value of H equal to 57 J/g, which is
very close to that of pure Sn. The experimental data have
been plotted in Fig. 20, together with the regression analysis
line. Therefore, it appears that the primary contributing factor
towards the melting point depression exhibited by the
multicomponent solders was the composition of the solid
solution, Sn-based matrix phase.
Lastly, it was confirmed that the intermetallic compound
particle phases had a minimal role in the solidus temperature
reduction. The same linear regression analysis was performed, but with the mole fraction of Sn (x) determined from
the total alloy solute concentration. That analysis returned a
value of H equal to the 125 J/g, which was nearly twice
-1
o
/K
-4
-4
2.0 10
-4
2.5 10
-1
Fig. 20 Logarithm of the reciprocal of the mole fraction of Sn in the solid
solution matrix as a function of the difference of reciprocal solidus
temperature and reciprocal melting temperature of Sn, both in K.
that of pure Sn. Therefore, the intermetallic compound
particle phases had a lesser role (if any at all) in the melting
point depression exhibited by the multicomponent solders.31)
4.
Conclusions
(1) The development of low temperature, Sn-based, Pb-free
solders used the method of multiple alloy additions to
realize a reduced solidus temperature and minimized
pasty range. The starting material was the eutectic
composition 96.5Sn-3.5Ag (mass%).
(2) Three ternary solders were developed from the Sn-Ag
binary alloy. Their compositions and solidus temperatures were: 91.84Sn-3.33Ag-4.83Bi, 212 C; 87.5Sn7.5Au-5.0Bi, 200 C; and 86.4Sn-5.1Ag-8.5Au, 205 C.
The alloys exhibited narrow DSC peaks indicative of
small pasty ranges and uniform distributions of particle
phases within their respective microstructures.
(3) A quaternary alloy was investigated that had a
composition and solidus temperature of 86.8Sn3.2Ag-5.0Bi-5.0Au and 194 C, respectively. The peak
was relatively broad, indicating an increased pasty
range. The microstructure was comprised of a 96Sn-4Bi
solid solution matrix phase and uniform distribution of
AuSn4 , Ag3 Sn, and Bi particles.
(4) The Au-containing ternary and quaternary alloys had
considerably higher strength than the binary Sn-Ag or
Sn-Pb solders, yet similar solid-state reaction rates
when present in solder/Cu couples.
(5) The additions of In and Cu formed the 66Sn-5.0Ag10Bi-5.0Au-10In-4.0Cu alloy having a solidus temperature of 178 C. The alloy microstructure was comprised
of an 88Sn-6.5Bi-5.5In matrix phase and uniformly
distributed particle phases of BiIn, (Au, Ag, Cu)3 (Sn,
In)2 , Ag3 (Sn, In), and (Cu, Au)6 (Sn, In)5 . This alloy
exhibited a solderability contact angle (34:2 0:7 )
and solid-state intermetallic compound layer growth
kinetics, both on Cu, that were comparable to the same
properties of other Sn-based, Pb-free solders.
(6) Further alloy additions were made to the 66Sn-5.0Ag10Bi-5.0Au-10In-4.0Cu that included Fe, Ni, Al, Zn,
Development of Sn-Based, Low Melting Temperature Pb-Free Solder Alloys
and Ga. The elements Fe, Ni, and Zn in concentrations
of 2–5 mass% had very little impact on the solidus
temperature. A 3 mass% Al addition caused a five
Celsius degree drop in the solidus temperature. The
lowest solidus temperature of 159 C was realized with
composition
62Sn-5.0Ag-10Bi-4.0Au-10In-4.0Cu5.0Ga. No other low temperature peaks were observed.
(7) The contributing factor towards the melting point (i.e.,
solidus temperature) depression exhibited by the multicomponent solders was the composition of the solid
solution, Sn-based matrix phase.
Acknowledgments
The authors wish to thank D. Susan for his thorough
review of the manuscript. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin
Company, for the US Dept. of Energy’s National Nuclear
Security Administration under contract DE-AC0494AL85000.
REFERENCES
1) G. Swan, A. Woosley, N. Vo, T. Koschmieder and T. Chong: Proc. of
the Technical Conference APEX 2001, (IPC SMEMA, 2001) pp. LF2-6
1–LF2-6 2.
2) A. Rae: Circuits Assembly July 2003, pp. 22–25.
3) M. Kelly, D. Colnago, V. Sirtori, J. Bath, S. Tan, L. Teo, C. Grosskopf,
K. Lyjak, C. Ravenelle and E. Kobeda: Proc. of the Surface Mount
Tech. Association International Conference, (SMTA, 2002) CD link.
4) M. Arra, D. Geiger, D. Shangguan, S. Yi, F. Grebenstein, H.
Fockenberger, K. Kerek and H. Wong: Proc. of the Surface Mount
Tech. Association International Conference, (SMTA, 2001) CD link.
5) J. Lieberman and W. Bradley: Toxic Lead Reduction Act: S.2637, (US
Senate, 1990).
6) B. Cardin and H. Waxman: Pb-Based Paint Hazard Abatement Act:
H.2922, (US House of Rep., 1991).
7) H. Reid: Lead Exposure Reduction Act: S.729, (US Senate, 1993).
8) Binary Alloy Phase Diagrams – Vol. 2, ed. by T. Massalski, (TMS,
1986) p. 286.
9) Y. Yamagishi: Proc. IPC Works 2000, (IPC, 2000) PP. S-01-1-1–S-018-4.
10) L. Felton, C. Raeder and D. Knorr: J. Metals 45 (1993) 28–32.
11) J. Glazer: Int. Mater. Rev. 40 (1995) 65–93.
12) Z. Mei, F. Hua and J. Glazer: Proc. Surface Mount International,
(SMTA, 1999) CD link.
13) F. Hua, Z. Mei and J. Glazer: Proc. 48th Electronic Components and
775
Tech. Conference, (IEEE/EIA, 1998) pp. 277–283.
14) Binary Alloy Phase Diagrams — Vol. 2, ed. by T. Massalski (TMS,
1986) p. 540.
15) A. Jackson, A. Artaki and P. Vianco: Critical Materials and Processes
in a Changing World, ed. by B. Rasmussen, et al., (SAMPE, 1994)
pp. 381–393.
16) Footnote #1: The presence of Bi strengthened the alloy as both a solute
as well as a precipitate phase. An additional attribute of Bi was that it
lowered the molten solder surface tension, thereby improving its
solderability.
17) P. Vianco and J. Rejent: J. Electronic Mater. 28 (1999) 1131–1138.
18) P. Vianco and J. Rejent: J. Electronic Mater. 28 (1999) 1139–1144.
19) P. Vianco and P. Mizik: Critical Materials and Processes in a
Changing World, ed. by B. Rasmussen, et al., (SAMPE, 1994) pp. 366–
380.
20) P. Vianco, I. Artaki, A. Jackson and J. Sampala: Proc. Surface Mount
International, (SMTA, 1994) pp. 437–448.
21) Binary Alloy Phase Diagrams – Vol. 1, ed. by T. Massalski, (TMS,
1986) p. 7.
22) The cost penalty of adding Au to the solder would only be realized if
the alloy were used in a bulk form such as solder wire or as the contents
of a wave soldering machine. There is very little added cost to a solder
paste containing Au because the expense of solder paste arises largely
from fabrication of the metal powder particles.
23) Binary Alloy Phase Diagrams — Vol. 1, ed. by T. Massalski, (TMS,
1986) p. 71.
24) Binary Alloy Phase Diagrams — Vol. 2, ed. by T. Massalski, (TMS,
1986) p. 141.
25) P. Vianco: The Metal Science of Joining, ed. by M. Cieslak, et al.,
(TMS, 1992) pp. 265–284.
26) P. Vianco: Soldering Handbook, (AWS, 2000) p. 51.
27) P. Vianco, P. Hlava, A. Kilgo and J. Rejent: Environmentally Friendly
Electronics — Lead-Free Technology ed. by J. Hwang, (Electrochemical, 2001) pp. 436–483.
28) The solder 66Sn-5.0Ag-10Bi-5.0Au-10In-2.0Cu-2.0Zn was actually
formed into both wire and solder paste for experiments on the assembly
of through-hole and surface mount test vehicles, respectively. In both
cases, the solder joints exhibited reasonable wetting-and-spreading as
well as fillet formation as first indicated by the solderability test contact
angles. The fillet-lifting defect was significantly reduced in the
through-hole joints; no damage was observed in the surface mount
interconnections.
29) K. Denbigh: The Principles of Chemical Equilibrium, (Cambridge,
1971) pp. 260–261.
30) G. Carter: Principles of Physical and Chemical Metallurgy, (ASM,
International, 1979) pp. 118–119.
31) It was interesting to note that the elemental metals that have melting
temperatures in the range of 30 C–350 C have H values of 20 J/g–
80 J/g. These latter values are very similar those recorded for the Snbased, Pb-free solder alloys.