|
Problems with Components and their Surface Finishes
Introduction Why has Hewlett-Packard requested an exception from the RoHS at the European Union? They want to make sure that certain fine-pitch components whose leads are ≤ 0.65 mm apart (centre to centre) can still retain lead finishes that contain a certain amount of lead. A spokesman of H-P explains the reasoning: “A content of 15 % lead prevents the creation of whiskers”. Background: On Mai 19, 1998 about 40 million ‘pagers’ suddenly did no longer work in North America. The reason for this disaster was the failure of Galaxy 4, a $ 250 million satellite. The analysis of this failure revealed that the cause probably was the growth of such a mono-crystal. Although electronics destined for space does not fall under the RoHS directive, scientists fear that even high reliability assemblies will use components from consumer technology under present cost pressure or worse that component manufacturers will phase out lead containing components altogether and thus such products may find their way into such applications. As many component makers decided on pure tin as a coating on leads such fears is not entirely unreasonable. Consumer electronics sees this problem not quite as dramatically. A short rap on the equipment usually displaces the crystal between contacts and the equipment works again. On the other hand, the current running through such a whisker during a short may also evaporate it and hence the interruption would only be temporary. Quickly hitting a satellite in space is slightly more complicated. Because of such potential problems the MIL specifications as well as requirements by NASA necessitate at least 3 % lead in coatings of contacts and leads to prevent the growth of these mono-crystals. There is no contradiction to the RoHS directive, as these applications are not even mentioned in it. To make it perfectly clear, tin is only one of the metals that is known to produce such whiskers ‚spontaneously’. Quite a number of other metals do as well if only the conditions are suitable for their creation. With all that concern in the industry about whiskers it may strike one as curious that no generally accepted testing procedure exists for the identification of surfaces that have a tendency to develop them. Even more curious may be the fact that no general classification system is available either as there are several different types of whiskers that may occur.
Problems Talks and certain publications seem to delight in stressing that certain lead-free solders show hardly any effect by a ‚slight’ contamination with lead. Perhaps this is to suggest that the life expectancy of such solder joints is not affected by ‚minor’ percentages of lead. Practically always the definition of what has to be understood as ‚slight’ or ‚minor percentage’ is lacking. Perhaps these authors would like to steer users away from cheaper alternatives such as Bi-containing solders? After all, it is well known that Bi-solder react rather catastrophically to even minor lead presence. Karl Seelig and David Suraski in one of their technical articles seem to take pleasure to show that even the highly praised SnAg4.0Cu0.5 solder does not take lightly a ‚minor’ presence of lead. When using a common test vehicle [ASTM E606] – carried out in an independent laboratory – they received the following results (10 000 cycles correspond to ‚test successfully passed'):
It is up to the electronics manufacturer to decide whether 1,000 cycles are sufficient for his product – and in many cases it actually may be and hence the presence of ‚some’ lead may be negligible. On the other hand, there may be some companies that rather see 13,400 cycles passed by their product than a mere 3,252. There are, then, two contradicting opinions that we were able to identify: one that would rather see no lead on any of the components and the other that absolutely insists on lead at least on leads and connectors. Lead on leads is not the only problem with components. We have to clearly distinguish between components that are ‚lead-free’, those that meet the RoHS requirements, and finally those that actually fulfill process needs of high-melting solders on top of it. Despite the fact that the component industry is developing lead-free and RoHS compliant components for years now, worldwide only about 40 % of component manufacturers declare that they are ready to meet the wishes of their Pb-free clients.
Higher Maximal Reflow Temperature in the Peak Zone When reflow soldering most processes will use peak temperatures in the interval between 240°C and 260°C. In order to survive such ‘extreme’ levels, many components will have to be completely redesigned. Otherwise the stress exerted to their interior by the thermal load would lead to immediate failure.
Final Surface Finishes for Components If not considered in depth a change in the final surface finish of components appears to be a rather minor matter. All one seems to be concerned with is wetting and the protection of solderable surfaces. However nothing seems to be that simple in a soldering process and as such surfaces are at least partially dissolved during soldering, the final result must be considered as well. Cautious people may at once think of all those lovely patents that have been applied for and granted. Many of them have been cleverly worded and do not only protect certain alloys as solder (disputable whether this would be fully acceptable in each and every case) but also all coatings and even solder joints that are using or even finally resulting in these patented alloys. It remains to be seen, though, whether a judge would actually rule on two or three solder joints on one assembly if the rest does not infringe on the said patent. Of much greater importance are thoughts directed towards the reliability of the ensuing solder joints. The dissolution rate of metals in SnAg solder is, for example, much higher than in traditional eutectic SnPb solders. It was possible to show that dissolved metals diffuse through the entire solder joint. Because of it even the reaction at the other side of the joint is affected. An impact on the reliability of the joint may not be excluded and one is advised that when considering the reliability of a solder joint one should include the effect of both surface finishes – the one on the component and the one on the printed circuit board – especially if those are different. The coating most often encountered so far was a tin/lead alloy with about 80 – 90 % tin. The ‚hottest’ candidates now are naturally the popular Sn/Ag or Sn/Cu alloys. But they are facing two hurdles: on the one hand there is a basic compatibility problem in the manufacturing process with these metals and on the other hand there is a lack of manufacturing facilities that could coat the lead-frames commercially in large enough quantities. What is left can be summed up in one word: pure tin coatings. They are compatible with the Pb-free solders and the component manufacturing process. But they also have – despite all the wordy promises of the component manufacturers – still not overcome their inherent problem of producing whiskers – whenever the conditions for their growth are ‚right’. The use of lead-free coatings is possible, however an important part of the electronics industry looks at it with trepidations. Adding small amounts of other metals that are known to impart similar whisker-avoiding properties to the alloy may modify pure tin surfaces. Gold and Indium are such additives. Both of them are expensive and both of them have some other negative indicators when it comes to soldering. In the eighties of the last century – to be more precise: in 1989 – Texas Instrument introduced a Ni/Pd-coating for its ICs. But despite the fact that by September 2000 more than 35 billion IC components had been sold with such Ni/Pd-surfaces, it has largely been avoided. Only seven years later Japan presented the world with Ni/Pd/Au, a coating that showed much improved wetting capability during wetting balance testing. Both these surfaces are used widely in Japan, however without the event of the RoHS directive in Europe neither Europe nor North America would have given these treatments much attention. Experiments with other additive metals are going on all the time as everyone recognizes that pure tin may be a problem. Small amounts of bismuth (Bi) or copper (Cu) or even silver (Ag) have been tried to avoid the costly and somewhat problematic gold (Au) and indium (In). Alloys with three metals as ingredients have caused difficulties during plating and many manufacturers have discarded this idea altogether. One of the more interesting coatings seems to be SnBi, which is being used in Japan (e.g. the Seiko Epson Group) and appears to be interesting to some of the European component manufacturers as well. Some of the coatings offered in the market:
A number of other possibilities are presently developed in Asia. As component manufacturers can only guess which one of the many lead-free solders his client is likely to use a decision as to the coating becomes a risky business. Any testing they might do has to take into account that not only SnCu, SnAg, and SnAgCu alloys are a possible choice but also SnZn+X or SnBi+X to name some of the more interesting lower melting alternatives. And though wetting properties are important, the real issue is the reliability of the resulting solder joint, especially after aging. Many parts manufacturers assume that the main alloy that will dominate flow soldering will be Sn0.7Cu (weight percentages). As long as the shape of the joint corresponds to ‚expectations’, neither component manufacturer nor user shows much concern about its quality. There still remains the question whether the contact lead has been wetted properly and whether the through-hole can be filled. Both these worries will remain as long as the temperatures are kept as low as they are nowadays.
BGAs and Chipscale Packages The tiny balls on BGAs must be seen as a separate topic as they still provide most of the material in any connecting joint. In cases where a Pb-free paste is applied that has a higher melting point any Sn/Pb ball on the BGA would melt prior to the melting of the paste. The low melting Sn/Pb alloys with the tin-rich paste and hence an undefined non-eutectic alloy emerges that will solidify even before the paste has melted. In this way we would create something akin to a ‘cold solder joint’ and everyone knows that such joints have a very low strength and will fail quickly on being stressed. When lead-free pastes are used, the industry is compelled to change the Sn/Pb solder balls that presently adorn BGAs. Such spheres can be produced by nearly all alloys and hence, fundamentally, the choice is free. However, changing these spheres affects the entire connecting interface structure between ball and BGA and hence the behaviour when stressing the component. Without a better understanding of this interface structure – and a substantial amount of research is directed into this area – any new development can only be tested case by case, taking into consideration the attachment to the board as well. The junior partner to the BGA topic shows similar problems. Chip Scale Packages (CSP) also exhibit deficiencies with regard to reliability. A lot of though was given to avoid early failure. Their structure is therefore quite essential to long life expectancy. Even for these parts lead-free is becoming the target and Sn/Ag/X alloys (where X is either In, Cu or Bi) as well as Sn-9Zn-1Bi-5In balls have been tested with regard to melting behaviour, phases, and solder joint quality. Sn3.5Ag8.5In as well as Sn3Ag1.0In1.0Cu have been corroded during thermal cycling. Notwithstanding these corrosions the reliability of such joints created with lead-free solder have shown better reliability under cyclical stress (from -65 to + 150ºC) than even Sn36Pb2Ag. Even FlipChip applications (and many still hope that the C4-technology will be excluded from the RoHS guidelines) require a lead-free solution. Since melting of the bumps or even a partial softening is not desirable – because of a negative influence on the reliability of the connection – alloys with a solidus under 260ºC not even considered. Changing the alloy for the bumps and the balls affects the total structure of the component. As a critical example we may just refer to the metallisation of the diffusion barrier.
Table 1: Max. Reflow Temperature for different Ball Alloys
Other Occurrences of Lead Lead is found in a few other areas than only the metallic coating of leads and terminations. Many temporary connections utilize some lead and so do crimping fixtures and IDC connectors. In plastic materials it may be used as a thermal stabilizer (PVC) or as a colouring agent (lead chromate and lead oxide).
Marking Requirements? The RoHS does not require marking of lead-free or RoHS compliant components. As a consequence quite a number of organisations have taken it upon themselves to exercise their ingenuity on this subject (JEITA, IPC/NEMI/JEDEC, Soldertec etc.) and the created mayhem looks accordingly. As could be expected there was no coordinating effort between such societies. To make matters even worse, component manufacturers have taken it upon themselves to either mark or not mark the components. If they do mark the components, they may have developed their own marking system that differs totally from that of the above organisations. To make things easy for them, they often do not mark the component but rather the package. After opening of the package, the marking sign is lost. Neither is it clear what the marking is telling us. Some declare themselves as lead-free, others refer to RoHS compliance and finally some even seem to indicate process conformity. There is a difference between being Pb-free and RoHS compliance as the lead-free component not necessarily meets the other substance restrictions of the RoHS directive. Whether such components are also suitable for higher reflow or flow soldering processes is then another question altogether. Before we see more clearly some time will have passed. As long as the supplier delivers within the Europe of the EU he will have to ensure that the items delivered meet the RoHS guideline and in Germany ensure compliance to the ElektroG. But it may do no harm to ask for confirmation in writing.
Detecting Lead Whether a termination contains lead or not can be tested relatively easily in each case under discussion. A handheld e-ray fluorescence analyser (XRF) yields results rather quickly (in about 60 seconds). Such instruments are quite easy to use and have a sensitivity of about 0.01 % for lead in a tin matrix. Testing of samples is hence a real possibility; however testing all components during placement or insertion with placement rates well below one second still remains unrealistic.
Purchasing Luckily quite a few components that meet lead-free conditions have already been placed on the market. Components that meet RoHS requirements are still somewhat less common as not all lead-free components have removed other restricted substances yet. Components that meet process characteristics are even harder to find. Before such a background one must recommend that purchasing is now cooperating much more closely with design, layout and production to ensure a minimal amount of friction in the changeover to lead-free and RoHS compliant technology. One of the typical problems we encounter on a regular basis is the ‚single-sourcing’ of components, a situation rejected by many. If only one manufacturer for a critical component can be identified, the bottlenecking in situations of short supply is pre-programmed, not to speak of the possibility of a price hijacking. Often a component is found that would meet the requirements, if only .... e.g. it would be smaller etc. A lengthening of delivery times for RoHS compliant components has been noticed and changes the purchasing strategy and may have an impact on cash flow. Finally it is still not a rare occurrence that certain components have not yet been put into the market – remember, about 40 % of component vendors declare themselves ‚not ready’. Whether one may obtain at least a temporary ‚relief’ by an exemption from RoHS is not yet clear. What remains under these circumstances is a redesign of the product, although changes of the layout or the design characteristics may open an entire box of Pandora.
Test Method for Whiskers: Most component manufacturers who test for whiskers are using the following method: Coated parts are bent and then subjected to an aging process · 6 months at room temperature · 6 months at 50 ºC · 6 months at 52 ºC and 90 % rel. humidity · 1000 thermal cycles from –40 to +85 ºC The parts are then inspected for whisker growth by SEM. The test is considered ‚successfully passed’ if no whiskers longer than 50 μm have been identified. Because of the length of such testing procedures, they must be seen as qualifiers for certain methods rather than as qualification of individual parts or products. |