Characterizing copper erosion in soldering remains elusive.

As the industry prepared to convert processes to Pb-free, we used our collective knowledge to predict where the greatest challenges would lie and prepared accordingly. In some cases our concerns were justified; in others, we were a bit overly worried.

And then there were those situations for which we were not fully prepared. Copper erosion is a prime example of the great, unanticipated issue that took many by surprise. Oh sure, a few pundits issued warnings, but many of us were preoccupied with what were thought to be the mainstream showstoppers: dealing with those enormous ΔTs on heavy server boards, wondering how low peak temperatures could go without introducing reliability or warranty concerns, navigating the extinction of SnPb component finishes. In retrospect, some of these problems seem puny compared to the troubles copper erosion presented.

Wow! Copper erosion in wave soldering turned out to be one of those funky little complications of Pb-free that didn’t really make our radar in the preparation phase of the transition. But when our ICT techs began showing traces, and annular rings were missing off the bottom sides of boards, we sat up and took notice. Our technical community appreciated this potential showstopper right away. Top engineers launched some really great studies on the topic – and published them too. In years past, such studies may have been considered competitively advantageous and remained proprietary, but in this case, many of the details have been disclosed and introduced into public domain. At this juncture, it’s now generally globally accepted that Pb-free (SAC) solder alloys with higher silver contents (3-4%) erode the copper features on boards much faster than do alloys with lower silver contents (<1%) or alloys with small amounts (0.05%) of nickel. How much faster is this erosion? By some calculations, about 10 times faster between best and worst. When it comes to washing thin copper connections off the board bottom, that’s an enormous difference.

Despite definitive data showing erosion rates depend on key process variables (alloy type, alloy temperature, dwell time and flow rate of solder), a nagging question remained: What part does the PCB itself play in the erosion equation? Plenty of cases were cited – maybe not photo-documented and published, but mentioned verbally – that claimed an identical PWB design from different board shops could demonstrate vastly different erosion behaviors in the same soldering process. Several associates and I wondered if this were true, or an aberration rooted in a completely different issue. Perhaps this was a premature conclusion, the result of a few unsubstantiated observations that snowballed and gained credence as the rumor circulated? I recall a lot of crazy rumors during the transition years; I’ve even addressed a number of them in this space. Although it makes sense that the copper plays a part in the erosion process, hypotheses without corroboration are still just hypotheses.

A team of engineers and scientists from Jabil, Nihon Superior and Cookson Electronics convened and enlisted the support of some academic institutions and other suppliers to collect hard data on whether the PWB has some influence. In cooperation with a laminator and eight fabricators, we compared erosion rates of coppers plated in 10 different baths. As it turned out, there was a big difference in erosion rates among the various electrodeposited materials – some eroded 50-75% faster than others. Even allowing some latitude for experimental error, there was no denying the copper plays a key role in the erosion process.

Our original hypothesis – that the erodibility may be dependent on grain structure and perhaps indicated by tensile properties – was not substantiated by the data. Although we did identify significant differences in the grain morphology and the behaviors of different electrodeposited coppers, we unfortunately found no correlations whatsoever with grain structure, ultimate tensile strength or elongation at break. Last October, we published our first report on the erosion experiment at SMTA International. At that time, we had data on the erosion properties of the 10 copper samples in SAC 305 alloy only. Now we have a full data set that includes three more alloys: low-Ag SAC, SnCu with nickel, and low-Ag SAC with nickel. It’s no surprise the coppers with the highest erosion rates in SAC 305 were also the highest with the other alloys; likewise with the lowest erosion rates. Within each alloy, the difference between the highest and lowest erosion rates ranged from 1.4 to 1.7 times.

This month’s lesson learned? There’s something about the copper. I wish I knew exactly what it was about copper that influences its erosion rate, but I don’t. However, the differences are well documented, and I am certain the technical community will eventually get to the bottom of it. When we first published the study last fall, we received some excellent feedback on particular characterization methods that might help reveal key properties of the copper material. Now that we have more data and see some consistent trends in erosion rates, we’re hoping to again receive input on possible characterization techniques.

The current report was published last month at SMTAI, and will be available in an upcoming issue of Circuits Assembly. If you’d like a copy, or have a question or comment, please send me a note (at my new email address) and I’ll be happy to share it with the research team.

Chrys Shea has 20 years’ experience in electronics manufacturing and is founder of Shea Engineering; chrys@sheaengineering.com.