Phase Equlibria and Thermal Analysis Near the
Au-Pb-Sn Ternary Eutectic

Daniel Lewis, Michael Notis, and Angela Grusd

 

1. Introduction

In microelectronic packaging the joining of electronic components to printed wiring boards is often accomplished by soldering. Typically tin-lead eutectic solder is used for this purpose, however, the service environment of the printed circuit board may alter the required composition of the solder alloy to either a high-tin or high-lead solder. To improve the wetting and bonding behavior between the solder and the copper-based conductor metal the contacts are plated with various materials. Typical materials include eutectic solder, nickel, palladium, or corrosion-resistant noble metals. In some processes, such as tape automated bonding or TAB, the contacts used to join the component leads to the silicon IC chip are plated with gold for this purpose.

P>During the soldering operation the molten solder will dissolve the metallization and/or the base contact metal and precipitate intermetallics that will form the bond responsible for giving the solder joint it's strength. If the metallization dissolves too rapidly or is too thick the integrity of the solder joint can be compromised by forming too large a volume fraction of the intermetallic phase. To control the amount of intermetallic in a solder joint it is necessary to understand the dissolution rates and phase equilibria controlling the process.

If the gold concentrations in tin-lead solder reach too high a level embrittlement of the solder joint will occur due to precipitation of intermetallic AuSn4. The phase equilibria between tin-lead-gold is an important tool that can be used to understand and design more reliable solder joints and component contact metallizations.

The tin-lead-gold system was first studied by A. Prince in 1966 and again by G. Humpston in 1984. Both have determined the ternary eutectic composition to be within 2-4 at% Au. Prince's value lies near 2 at% Au and Humpston's value lies near 4 at% Au. Figure 1 shows the proposed ternary liquidus projection in the tin-rich corner. A series on seven alloys were made at compositions e at compositions spread around the two proposed ternary eutectic points with at least one alloy in all three primary phase fields. Metallography and DSC were used to determine which of the two compositions were closest to the true ternary eutectic. This information will help microelectronic component assemblers determine the limit of gold solubility in tin-lead solder before large precipitates of AuSn4 embrittle the solder joints.

2. Experimental Overview

Because the ternary eutectic consists of three three-phase reaction lines meeting at a single point it is possible to chose an alloy series where each alloy lies in a different two-phase region of the liquidus of the phase diagram. This will produce a different primary phase in each alloy make it possible to visually examine the alloys to determine their position with respect to the true eutectic. The middle-most composition of the alloy expansion was chosen to lie about midway between the two proposed ternary eutectic compositions. The resolution of the alloy spread was chosen to be 2 at%. This allows each alloy to reside in a different primary phase field.

Figure 2 shows a schematic of the phase diagram near the eutectic and the expansion that was chosen for this experiment. The proposed eutectic compositions (squares) by Prince and Humpston are also labeled in addition tolabeled in addition to the compositions of the alloys (circles) used in this study.

Alloys were made by melting the pure materials in a ¼" test tube over a Bunsen burner. To prevent oxidation during melting a few drops of glycerin were added to the test tube. Samples were chosen to be 3 grams so that the final shape of the ingot would be nearly spherical to help reduce any heat transfer effects. Samples were melted three or more times in insure that good mixing was achieved. After solidification samples were cut in half, mounted in epoxy, and polished using standard metallographic procedures. The solution of 0.04 micron silica provided an etch so that phases could easily be identified without etching. To improve the contrast for the photomicrographs Karapella's etching solution of 5g FeCl3, 2mL HCl, and 99mL methanol was used to darken the tin phase. In one case, however, a solution of 2mL HNO3 and 98 mL alcohol (2% Nital) was used due to the large volume fraction of tin present in the sample.

3.0 Solidification, Phase Diagram, and Microstructure Relationships

A typical liquidus projection and isothermal sections around a ternary eutectic reaction are shown schematically in Figure 3. There are three important features on the liquidus surface: primary phase fields, three phase reaction lines, and invariant reactionnd invariant reaction points. If the temperature and composition are such that an alloy lies in a primary phase field below the liquidus surface a liquid phase and one solid phase, the primary phase, would be in equilibrium. If the alloy temperature and composition lie on a three phase reaction line then the liquid phase and two solid phases would be in equilibrium. If the temperature and composition lie at an invariant reaction point then the liquid and three solid phases are in equilibrium. Figure 4 illustrates these ideas.

4.0 Results and Discussion

4.1 Sn-3Au-21Pb

Figure 5a shows the microstructure for this alloy located between Humpston and Prince's proposed eutectic compositions. This microstructure is an SEM backscatter image so the lightest phase is lead-rich, the darkest phase is tin-rich and the gray needles are AuSn4. The primary phase in this microstructure is AuSn4 which can be seen as a large lath like structure at the bottom of the micrograph. Around the lath lead is seen coming out of solution. This indicates that the next reaction in this solidification occurs along the AuSn4+Pb+liquid three phase reaction line. Final solidification occurs at the ternary eutectic as shown in Figure 4b. Note that the lead-rich phase preferentially nucleates on the AuSn4 needles and rarely are the three phases in mutual contact.

Presence of the AuSn4 needles as a primary phase suggests that Prince's original composition for the ternary eutectic was more accurate than Humpston's. With this evidence the three phase reaction lines can be redrawn to meet at Prince's eutectic composition. These lines are drawn as dotted lines on Figure 2.

4.2 Sn-5Au-19Pb

This alloy lies in the AuSn4 primary phase field and lies close to a straight line drawn from the stoichiometric compound AuSn4 and Prince's eutectic composition. This suggests that solidification occurred through the AuSn4+liquid phase field.

Above a ternary eutectic three three-phase tie triangles exist separated by three two-phase regions. If an alloy composition were to lie on a straight line between the eutectic point and a single phase region the solidification of the alloy would occur from a the phase field directly to the ternary eutectic. The microstructure in Figure 6 suggests this type of solidification due to the absence of a mutually growing two phase structure. The microstructure contains primary AuSn4 laths surrounded by ternary eutectic.

4.3 Sn-3Au-23Pb

Figure 7 shows an alloy near the three phase reaction lree phase reaction line for Pb+AuSn4+liquid. This is evidenced by the largest feature in the microstructure consisting of an AuSn4 needle surrounded by a lead-rich phase as predicted by the phase diagram. This also provides evidence for the eutectic point to be located nearer Prince's composition because if Humpston's composition were used a primary lead-rich phase would be expected. But, as this alloy shows, lead-rich and AuSn4 are solidifying cooperatively indication that a three phase reaction line is nearby.

4.4 Sn-1Au-23Pb

This alloy lies near the three phase reaction line extending from the tin-lead eutectic in the primary tin phase field and it was expected that there would be a small amount of primary tin, a three phase reaction product between tin and lead with final solidification occurring at the ternary eutectic. Figure 8 shows nodules of tin-lead eutectic-like structures, outside these nodules are regions consisting of ternary eutectic. Etching was done in 2% Nital due to the large volume fraction of tin in this sample.

4.5 Sn-3Au-19Pb

This alloy is located in the primary AuSn4 phase field near the AuSn4+Sn+liquid three-phase reaction line and a primary needle of AuSn4 can be seen in Figure 9. Because the alloys composition is close to the three-phase reactio three-phase reaction line it is expected that a small volume fraction of AuSn4 growing alone will be present in the microstructure. After the growth of primary AuSn4, the phase diagram predicts that coupled growth of tin and AuSn4 will occur. Because AuSn4 is present as the primary phase and in the three-phase reaction it is difficult to tell where the primary phase ends and the three-phase reaction begins. From the micrograph of Figure 8 regions containing AuSn4 and tin are seen with ternary eutectic regions separating them. This agrees reasonably well with the phase diagram although the exact features in the microstructure cannot be completely explained from the phase diagram.

4.6 Sn-1Au-21Pb

This alloy is located in the primary tin phase field and in Figure 10 the primary tin phases can easily be seen. Just outside the tin phase a three-phase reaction product can be seen as a tin-lead lamellar structure, this corresponds to the three-phase reaction line extending from the tin-lead eutectic reaction. Figure 10 shows that the final solidification occurs at the ternary eutectic where it appears that all phases are in mutual contact.

4.7 Sn-5Au-21Pb

This alloy lies in the primary AuSn4 phase field and in Figure 11 the large laths of AuSn4 can easily be seen. The three phe seen. The three phase reaction between AuSn4+lead+liquid solidifies next with final solidification at the ternary eutectic.

5. DSC Results

In addition to the peak temperature of the DSC curve there are two other important temperatures that should be reported when for a DSC experiment. They are the first liftoff temperature and the peak tangent temperature. Because the electronic analysis equipment may lag behind the reactions that we are trying to measure, recording the first liftoff and the tangent temperatures may help to better estimate reaction temperatures. Figure 12 shows a typical DSC curve and lists the important temperatures that were measured during this experiment. Table I shows the temperature profile used for the DSC experiments.
 


Table I. Temperature profile used for the DSC experiments

 
Type of Experiment Temperature Range  Rate
Heating -1 Room Temperature-250°C  10°C/min
Isothermal Hold 250°C  N/A
Cooling -1 250-50°C  10°C/min 10°C/min
Heating - 2 50-160°C  5°C/min
Heating - 2 160-200°C  1°C/min
The heating - 1 and cooling -1 runs were used as a conditioning run to stabilize the samples in the aluminum DSC sample pans. The final heating run from 160 to 200°C at 1°C per minute will be analyzed for invariant reaction temperatures. Table II shows the results for the DSC experiments.
 


Table II. DSC Results for Au-Pb-Sn alloys around the ternary eutectic between 160-200°C.

 
DTH="78">176.5°
Alloy First Liftoff First Peak Tangent  Return to Baseline? Second Peak  Peak to Peak Final Melting 
Sn-3Au-21Pb 175.7°  176.5° No 179.3°  0.7° 181.2°
Sn-5Au-19Pb 176.0°  176.0°  176.8° Yes N/A  N/A 180.6°
Sn-3Au-23Pb 175.9°  176.7° No 179.7°  0.5° 181.8°
Sn-1Au-23Pb 176.6°  177.1° No 180.5°  2.0° 183.4°
Sn-3Au-19Pb 175.6°  176.0° No 178.5°  0.7° 182.1°
Sn-1Au-21Pb 175.9°  177.1° No 180.8°  1.8° 187.8°
Sn-5Au-21Pb 176.4°  176.5° Yes N/A  Yes N/A  N/A 180.6°

6. Conclusions

All microstructure of all the alloys agree with the phase diagram provided that the eutectic point reported by Prince is chosen as the ternary eutectic composition. However, thermal analysis indicates that Humpston's eutectic composition is the correct choice for the ternary eutectic.

7. References

  1. Prince, A., J. Less Common Metals, 12, 107-116 (1967)
  2. Humpston, G., and B.L. Davies, Metal Science, 18, 329-331 (1984)
>