SAC 305 produces 10 times fewer mid-chip solder balls than does SnPb.

In May, I reported Pb-free solder pastes create fewer mid-chip solder balls and solder bridges than do SnPb pastes. Why this phenomenon occurs was not well understood at the time. I suspect it may have something to do with superior hot slump resistance of more thermally stable paste fluxes, but our chief paste formulator suggested it may be more likely a result of the oxide film that forms on the molten alloy; it’s more tenacious in Pb-free alloys than in eutectic SnPb.

To get to the bottom of this, I tested three pastes head-to-head. The first was a standard SnPb paste; the second a standard Pb-free (SAC 305) paste, and the third a hybrid paste of SnPb solder powder combined with Pb-free flux. The logic behind the test was simple: If lower defect rates were due to the flux, then the hybrid should behave more like Pb-free and produce fewer solder balls. If lower defect rates were due to the alloy, the hybrid should behave more like SnPb and produce more solder balls.

A few details on the experiment: My test vehicle has 400 chip components total, 100 each of sizes 0402 through 1206. Of these 100 each, 50 are mounted horizontally and 50 vertically. My test stencil was 0.005"-thick laser cut stainless steel, and for each block of chip components, half the apertures were rectangular, while the other half were radius inverted home plate designs, also known as the “Roundie” (Figure 1). Both were sized 1:1 with the pad. We ran 12 boards with each paste; six with a ramp profile and six with a soak profile, all in an air environment. This gave us 4800 opportunities for solder balls per paste type.


The results were crystal clear. Dr. Michael Liberatore’s insights were right on the money! The SnPb alloy produced about 10 times more solder balls than the Pb-free alloy, regardless of the flux vehicle. The Pb-free paste produced 62 mid-chip solder balls, equivalent to a 1.3% occurrence rate. The SnPb produced 504 solder balls, equivalent to a 10.5% occurrence rate. And the hybrid paste, which had SnPb alloy and Pb-free flux, produced 616 solder balls, or a 12.8% occurrence rate. (Note use of the term “occurrence rate” rather than “defect rate”; criteria under which MCSBs are considered defects vary widely.)

Another interesting datum collected was the performance of the Roundie versus the rectangular aperture. We first investigated and documented the Roundie about three years ago. The investigation was prompted by the lack of spread of Pb-free solders on OSP. In the SnPb era, we were able to limit solder ball production by reducing the stencil aperture area relative to the pad. When Pb-free solders began gaining in popularity, we found that the practice of aperture cropping often led to exposed copper along pad perimeters and corners. We needed a better way to deposit solder paste to the corners, while still limiting paste volumes and solder ball generation. The aperture design that emerged as the best in that 2004 experiment was what we now call the Roundie, and it became part of our Pb-free aperture library. In this recent experiment, a total of 1182 solder balls were created, but only 17 came from the Roundie aperture (16 SnPb; one Pb-free). That’s an occurrence rate of 0.2%. Although we did not need to fine-tune our aperture design to this degree in SnPb processing, it is nice to know this is a robust design regardless of alloy used.

What’s this month’s Pb-free lesson learned? The numbers speak for themselves. With the same boards, stencils, components and (analogous) profiles, produced on the same assembly line on the same day by the same technician, SAC 305 produced 10 times fewer MCSBs than did SnPb. This is the first absolute benefit of Pb-free solder I’ve been able to quantify. Although this is a small perk in a huge challenge, it’s still a perk.

Does the Pb-free alloy produce fewer solder bridges on fine-pitch devices? I did not even attempt to gather data on this question. While my test vehicle has 400 components, only four are QFPs, so generating a statistically significant sample size represents a considerable effort, not a quick test. For now, I’ll continue to rely on the anecdotal evidence provided by my associates who have personally experienced the alloy transition process. With respect to intrusive reflow, again I don’t have a statistically significant sample size, but this recent test did generate good indications. The Pb-free alloy appeared to perform comparably with SnPb on smaller overprints and ramp profiles, but seemed to outperform SnPb in larger overprints and soak profiles – perhaps yet another perk as we continue down this rocky road to Pb-free soldering.

Au. note: Many thanks to Dr. Michael Liberatore for guidance on the physical mechanisms at work in this phenomenon, and to Esse Leak for assembling the boards, counting nearly 1200 solder balls, and screening more than 1400 pin-in-paste locations. We would not have learned this lesson without their help.

Chrys Shea is an R&D applications engineering manager at Cookson Electronics (cooksonelectronics.com); chrysshea@cooksonelectronics.com. Her column appears monthly.

• Prev
• Next