Revisiting Rework
Excess copper dissolution remains a big problem for assemblers.
No one, save for a few equipment suppliers, likes rework, but it’s an ugly fact of life. At best, it is performed properly and costs only time and money. At worst, it results in a product that leaves the factory with a latent defect destined to fail in service. In between, all kinds of things can go wrong that result in more extensive repairs or even scrap.
There has always been an inherent risk to rework. Lifted pads, ripped barrels, torn traces – they all come with the territory. Pb-free ratchets up that risk a few notches, especially with plated through-hole components. Pb-free PTH rework provides numerous opportunities to ship product with hidden defects caused by excess copper dissolution. We’ve seen plenty of evidence a PTH solder fillet can, from the outside, be visually acceptable, but on the inside lack the knee to connect the barrel to the annular ring. We’ve seen thin spots in barrel plating consumed by solder, permitting laminate to outgas and create giant voids inside joints that have good looking top- and bottom-side fillets. And we’ve seen 0.005" traces reduced 60 to 96% with 30 sec. exposure to flowing Pb-free solders. As an industry, we don’t really know the full ramifications of these hidden defects, and I don’t think we’re going to find out anytime soon.
With the potential issues that can arise from Pb-free PTH rework, we should ask why we perform the risky operation in the first place. If a component is defective or a pin bent, then the process clearly is unavoidable. But if performed because of insufficient hole fill during the primary attach process, we need to revisit the basis for the rework. Missing topside solder fillets is not a condition that should necessitate repair. Even IPC’s Class 3 workmanship standards dictate 75% hole fill on signal connections and 50% on ground plane connections. So why do so many assemblers rework PTH devices that lack topside fillets? Because QA wants to see them. For 50 years, we have validated SnPb wave processes by looking at topside solder fillets. Intellectually, most assemblers know they are not necessary for performance, but emotionally, they want to see them. This is a classic example of the road to hell being paved with good intentions: The person who requests topside solder fillets thinks they will make the joint stronger and more reliable, not realizing the joint can actually wind up weaker and less reliable if too much copper is dissolved in the process.
I don’t think we should stop at simply reconsidering our stipulations for topside solder fillets. I believe we should rethink hole fill specifications in general. If a 50-pin connector has 49 pins that meet workmanship standards and one that falls a little short, do we risk all 50 joints for the sake of that one? Which scenario renders the connector less reliable: one barrel of 50 that is short on solder fill, or 50 barrels that have all endured rework? This is where the decision-making process can become a bit imprecise. To avoid ambiguity on the shop floor, quality metrics need to be based on specific, measurable quantities that have no room for interpretation. But in the case of hole fill – especially on PWBs 0.093" or thicker – strict enforcement of rigid metrics has the potential to do more harm than good. Perhaps a better direction for production personnel is to seek the advice of engineering or quality professionals, or relegate the assembly in question to the material review team for disposition.
When determining the rework candidacy of an assembly, a number of factors should be considered. They include the end-product, its performance requirements, the degree of the shortfall and some specifics of the rework process to which it will be subjected. The process itself is a critical factor in the decision, as mini-wave soldering often can be much harsher on the PWB than the original primary attachment process. Consider the number of solder contact cycles associated with a mini-wave repair: The first cycle melts the existing solder (often without preheat) so the connector can be removed; the second cycle re-melts residual solder in the barrels so the holes can be cleared; the third one re-solders the connector (again often without benefit of preheat), and subsequent cycles may be applied to remove bridges. The cumulative exposure times add up quickly. They can exceed 60 sec. on a minimally challenging design and 120 sec. on a demanding one. Compare that with the typical 3-9 sec. contact time on a wave-solder machine and it’s no wonder the copper is disappearing.
Numerous factors influence erosion rates; they include alloy type, solder temperature and flow rate, dwell time, and the quality of the copper that’s on the PWB. The copper’s plating process has been identified as a considerable factor in dissolution rates. It’s influence has only recently been quantified, and the reasons why are not yet fully understood, but even if they were, they would likely be out of the assembler’s control anyway. Probably the single most influential factor the assembler can control is the dwell time on the wave. There are a couple of ways to effectively do that:
Use preheat. Although preheaters are available for miniwaves, many processes still don’t use them. Instead, they rely on the flowing solder as the sole heat source for the process. That practice might have been okay in the world of SnPb, but not so with Pb-free solders. Heating a board to liquidus temperature from room temperature requires three times more solder exposure than heating it from a preconditioned 300°F/150°C. The entire time the flowing solder is transferring the heat, it’s simultaneously washing away copper from the bottom of the PWB. Three times as much solder contact for at least two cycles: the removal and the reattachment. That’s an awful lot of extra dwell time the copper might not be able to tolerate.
Perform component removal and barrel clearing with hot air. This process, popularized by Bob Farrell at Benchmark Electronics and Greg Morose of the University of Massachusetts at Lowell, exposes the PWB to flowing solder only when it is actually needed – for the purpose of attaching the component.1 The process artfully avoids unnecessary dwell time and therefore limits risk of excessive copper loss. It uses a BGA rework system to melt solder in the barrels with forced convection. When the solder is fully liquefied, the component is removed manually. To clear the barrels, the board is preheated and the barrels vacuumed by the automatic site redressing system on the same BGA rework station. This process has shown great success, even on large components like 240-pin DIMM connectors. Farrell provides two caveats when employing this method: Always preheat the entire PWB assembly, and use only automated site redressing equipment, if possible. Similar to BGA redressing concerns, handheld solder vacuum pens can do a great deal of damage to a PWB’s surface if they fall into the wrong hands.
Limiting the dwell time is probably the easiest way to combat copper erosion during rework, but not the only way. Alloy selection can have a profound effect on erosion performance. Research indicates alloys with lower (or no) silver content and/or small amounts of nickel dissolve the PWB copper more slowly than SAC 305 or 405, which were selected as SnPb replacements based primarily on SMT considerations. Due to the many wave solder-related issues identified with the 305/405 family, like copper dissolution, shrinkage tears, pot corrosion and material cost, alternative alloy options for wave soldering continue to proliferate, and at a relatively rapid rate. None of these newer alloys boasts as much reliability data as SAC 305/405 systems, and early products could not meet the SAC 305 benchmark for debridging and hole fill, but that’s not necessarily the case anymore. Some recent market entrants show a lot of promise in rework and primary attach processes, and the overall performance of alternative alloys will get better as researchers continue to improve them.
Other tactics to mitigate copper erosion include optimizing the solder’s temperature and flow dynamics. These methods probably will have less impact on erosion rates than limiting dwell times or switching alloys, but in a process that has the potential to ruin an assembly in 30 sec. or less, every little bit of optimization helps. It’s scary to think while we are fixing one defect we might be creating another, and it’s even scarier when we consider the new problems are often undetectable. Some industry leaders would like to see development of a standard test that can check a copper’s sustainability, similar to the solder float test, but adapted to address the current issues. It would provide assemblers with some assurance the PWB could withstand a certain minimum level of exposure.
Unfortunately, we can’t avoid the need for rework altogether, but we can rethink our criteria and perform it only when it is more likely to help a product’s performance than hurt it. We still have a lot to learn about the new material sets and manufacturing processes. What we do know is that when it comes to PTH rework, our best defense against excess copper dissolution is limiting the board’s exposure to the flowing solder. In short, we need to fix the defect, not dwell on it.
References
1. R. Farrell and G. Morose, “Assembly Rework and Lead-Free Impact,” CALCE Symposium on Part Reprocessing, Tin Whisker Mitigation, and Assembly Rework, November 2008.
Chrys Shea has 20 years’ experience in electronics manufacturing and is founder of Shea Engineering; chrys@sheaengineering.com.
