For Pb-free PWB surface finishes, there's no panacea, just choices.
![Better Manufacturing Better Manufacturing](http://circuitsassembly.com/cms_bu/images/stories/column_icons/betterman.gif)
One aspect of the Pb-free conversion process is the PWB surface finish. With regard to production volumes, the contenders are HASL, pure tin, NiAu, OSP and immersion silver. There are others and others will evolve no doubt, but at this point, these are the mainstream.1
Most practitioners shy away from pure tin finishes (matte tin, bright tin and others). It seems the greater the “presence” of tin, the more opportunity and resulting propensity for Sn-whisker propagation. While most Pb-free passives are finished in matte tin, these comprise a small aggregate volume of pure tin present in most assemblies compared to assemblies with a pure tin surface finish. Thus, for most products, pure tin would not be the first choice (or second, or third, etc.).
Then there is good ol’ hot air solder leveling (HASL). Yes, it is “old”: HASL came about in the mid ’70s and was quickly adapted as “the way” to finish through-hole PCBs with a SnPb coating. Compared to hydro-squeegee, IR SnPb fusing and some other methodologies, HASL was economical, efficient and proficient. HASL was adaptable for early SMT, too. But when component lead pitches started falling below 0.025", HASL presented a hassle. In the HASL process, the PWB is immersed in a bath of molten solder and then squeeged (leveled) with hot air knives. This results in a board surface that, while flatter than reflowed SnPb, is less flat than the immersion coatings. (The vertical machines also left excessive thickness toward the board bottom, an issue horizontal machines remedied). Surface mount requires even topography of the PCB surface. There are strict specifications with regard to the coplanarity of SMT IC leads, but what happens when the pad surfaces those nice coplanar leads are being attached to are inconsistent?
The problem is compounded when HASL is accomplished with Pb-free alloys. One, with the higher temperatures (a preheat of 190-200°C, a solder bath of 230°-250°C and air knives operating at 435°-535°C) involved with said alloys, there is a higher probability of PCB warpage. Warpage presents problems in the printing process if there is poor contact between the stencil and board. Besides fine-pitch leaded IC packages, area arrays and QFN packages are also somewhat finicky with regard to PWB topography.
Pre-RoHS, HASL made up some 75% of the market. Some assemblers transitioned from HASL long before Pb-free, however, particularly those building PCMCIA cards and other thin substrates. HASL is a known process, one that can be controlled with reasonable effort (with regard to avoiding excessive intermetallic formation, delamination and exposed copper situations), and many fabricators are equipped with the required machinery. A 2006 survey by European Pb-free Soldering Network TUB found approximately 27% of the Pb-free surface finishes used in Europe are HASL. They include SnAgCu, SnNiCu, SnZnCu and a large amount of tin HASL (ugh!). HASL definitely has its fans, but for many, there are better, more coplanar ways to go.
ENIG. Early SMT emigrants from HASL adapted gold as surface finish – particularly electroless nickel immersion gold (ENIG) and immersion gold (ImAu). ENIG, with a nickel barrier thickness of 50-150 µin under 3-10 µin of gold, presents uniform thickness and planarity. It also maintains great wettability, even under long storage conditions, and good surface contact for ICT. Sounds rather idyllic.
To its detriment, gold is expensive. (Then again, what isn’t, except maybe lead?) Apparently, controlling the plating process(es) is also a bit more demanding, as is evident by the number of fabricators that do not consistently exercise such control. Black pad, whether resulting from excessive nickel corrosion, insufficient gold (as discussed in a previous column), or improper phosphorous content of the nickel (it should be a 7-12% maximum codeposit) resulting in poor gold adhesion, rewards the assembler with weak joints or even outright opens. Embrittlement results from excessive gold deposits (more than 30 µin) with grainy appearance and poor joint strength. In fact, poor joint strength has been a concern in applications that experience shock, particularly where area arrays are soldered.
We seldom see NiAu finishes in cellphones and PDAs anymore. Weak joints are somewhat systematic because the SnNi intermetallic does not form as rapidly as does SnCu. This is why many handheld devices do not use ENIG. Remember, ENIG is the only surface finish where the soldering is to nickel, not copper.
OSPs. Organic solder protectants have improved greatly during the past decade and have seen wider use. Here, an organic coating of 8-20 µin is applied over etched bare copper. During reflow, the coating dissipates where flux and heat are present, leaving the solder in contact with the copper board surface. It offers a planar board surface, is low cost, easily reworkable and widely available, as most fabricators have the means to apply OSP. On the other hand, the shelf life is relatively short. That’s not a problem for high-volume, quickturn applications, but not ideal when shelf life requirements exceed 12 months. While initially able to sustain up to three thermal excursions (reflow and/or wave solder), this might diminish toward the end of its shelf life. Also, the OSP coating is not conductive, and this must be taken into consideration when soldering test points.
ImAg. Immersion silver, with a very planar surface of 3-12 µin of silver codeposited with an organic inhibitor, presents an alternative surface finish less expensive than gold, but with a longer shelf life than OSP. Silver is an extremely conductive metal and is sometimes favored for improved signal integrity in high RF and microwave applications. But, like your silverware, it will tarnish over time. This can occur during assembly, though it is often cosmetic and does not affect solderability or joint integrity. A more serious problem is silver migration. Silver ions, under certain conditions, may migrate from cathode to anode, forming dendrites that reduce resistance (and may even, in some extreme cases, form a short). We don’t have as extensive experience with ImAg as with other finishes, so the jury is still out, but it does show a great deal of potential and has been adapted to many applications.
ImSn. Platers do not like to do large volumes of immersion tin because of the difficulties in disposing of the wastes, principally thiourea or other nasty chemicals required to get the tin to bond to copper. Moreover, ImSn has shown poor performance in multiple heat cycle tests. Throw in the prospect of tin whiskers, and ImSn becomes less attractive.
There is no panacea surface finish (oh, how we long for those simpler times of through-hole). But there are a plethora of choices and, as with everything else SMT, the selection is application-driven. Shelf life, surface planarity, conductivity and certainly cost are the key factors. Remember, we’re all in this together.
Au.: Special thanks to Jim Hall of ITM Consulting for his contributions to this column.
Phil Zarrow is president and a principal consultant at ITM Consulting (itmconsulting.org); phil_zarrow@itmconsulting.org.