# Alloy Melting

Folks,

Dear Dr. Ron,

Recently we had a solderability problem with tin-finished component leads and SAC305 solder paste.  One of our engineers claimed that the problem was that the tin finish melts at too high a temperature (Tm= 232°C) for the SAC305 solder paste (Tm = 219°C) to melt it.

My understanding is that certainly above 232°C both will melt and form a good solder joint, but even if the temperature was less than 232°C, say 225°C, the tin would melt. Can you explain this phenomenon?

Richard,

Thanks for this question, which can be interpreted two ways. The first would be that, in a reflow oven at temperatures above the melting point of both metals, the one with higher melting temperature prevents the metal with a lower melting temperature from melting it. This is not true, since both metals would come near to the temperature of the air in the reflow oven and melt.

The other perspective would be that the temperature in the reflow oven is above the melting temperature of SAC 305, but below that of tin. So, how can the tin melt?  To consider this situation let’s say the oven is at 228°C. Will the tin on the lead or pad finish melt? The answer is yes. But, let’s try to understand the phenomenon with gold and tin first.

Metals that have extreme melting point differences often dissolve in each other. As you stated, tin melts at 232°C, whereas gold melts at 1064°C.

This phase diagram can be found here.

Figure 1. The gold tin phase diagram

To make a gold-tin solder, all one has to do is have a bath of tin at some moderate temperature, say 350°C. Insert the gold and the gold will melt and flow into the molten tin. This is true even though the gold melts at 1064°C. This effect can be shown experimentally. A similar phenomenon exists with gold and mercury. Mercury reacts with gold at ambient temperatures. The phenomenon can be used to extract tiny gold particles from soil and is commonly used today in artisanal gold mining. Unfortunately this use of mercury is often toxic to the miners and pollutes the environment.

Figure 2a

Figure 2b

Figure 2c

Figure 2d

Figure 2e

Figure 2f

Cheers,

Dr. Ron

# Service Life

My company makes an electronic product that requires a 40- year shelf life. We assemble with tin-lead solder on FR-4 PWBs. The product is to replace older technology (i.e. 1960-70s), but has some newer components such as BGAs, SOICs, and PQFPs. The product will be stored in dry nitrogen at 70F.  We take great care in manufacturing, by cleaning, inspecting, and testing the end product.

My question is, Do you know of any studies that would discuss the reliability of products stored or in use for 40 years?

My sense is that our reader will be successful, but his question is profound and hard to answer with confidence. The military would like their electronics to perform for that long, but realistically much of it is replaced every 10 years or so. If you look at something like the B-52 bomber, which debuted in 1952, the electronics have been upgraded regularly. So there isn’t as much 40-year electronics experience as one might think. An exception being the IBM AP-101 computer. This computer was kept in service for over 30 years, because it served its function and had survived the rigorous and expensive military qualification testing.

However, anecdotal data might support optimism for 40-year shelf life. In a class I teach at Dartmouth, The Technology of Everyday Things, I have sought out some old transistor radios from the late 1960s and early 70s to show the class how this old technology works. Anytime I have every found an old device like this, they always work, unless the batteries have leaked inside the radio.

This question raises an interesting thought. Although those of us in electronics assembly are concerned with tin-lead and lead-free solder joint life, what about the modern devices inside the components? Forty years is a long time. How will the 3D-22 nanometer copper circuit lines in a modern microprocessor hold up over this amount of time? These circuit lines lines are so fine that the 22nm width is only about 70 atoms.  In addition, copper integrated circuits are still a relatively new technology. I’m sure much accelerated life testing has been done on such circuits, but would such testing confirm 40 years of shelf or service life?

I would appreciate any thoughts that readers have on these questions.

Cheers,

Dr. Ron

# On Pb-Free Reliability and its Doubters

I was at SMTAI (Surface Mount Technology Association International) in late September. As mentioned, I chaired a session on Alternative Alloys. At this session, Greg Henshall presented a paper on the Low Silver BGA Sphere Metallurgy Project. This paper was a collaborative effort of six companies. In addition, Richard Coyle presented an overview of the work of three companies titled “The Effect of Silver Content on the Solder Joint Reliability of a Pb-free PBGA Package.” Both projects evaluated lead-free thermal cycle reliability as a function of silver content and compared the results to SnPb reliability.

Both papers concluded that as far as thermal cycle reliability is concerned

SnPb < SAC105 < SAC305 < SAC405

Coyle’s paper summed it best:

Each of the SAC alloys outperformed the SnPb eutectic alloy in every test, including the long, 60 min. dwell time test. This tends to diminish the argument that SAC is less reliable than SnPb. (See Coyle’s figure. Data curves to the right are more reliable.)

Henshall’s paper also showed that the addition of dopants, to improve shock resistance, in SAC105 does not reduce thermal cycle life.

So, it appears, at this time, that, from a thermal cycle and drop shock perspective, it is looking more and more like SAC-based solders out perform SnPb solders in these two reliability arenas.