By this stage, you should have made substantial progress in procuring components (Task 2), and have decided on the implications of your intended process for the board specification (Task 1). This unit relates to the requirement for you to report on the potential impact of lead-free implementation on the reliability and performance of the whole of the range of AMS’s products, leading to your outlining the recommended evaluation procedure for the company to follow.
During your exploration of the materials options, the way that the lead-free joint is made, and the implications for assembly and board fabrication, you should have come across a number of areas of uncertainty about the reliability of assemblies made with lead-free solder. This is not surprising, given that the industry has almost a century of experience with tin-lead solders, and a great deal of basic research has been carried out during the past 40 years in particular, whereas our experience with lead-free materials is rarely more than 12 years old.
In considering solder joint reliability, there is an amazing amount of information available: a simple Google search yielded:
"solder joint reliability" (3,720 hits!)
"solder joint reliability" +"lead-free" (900 hits)
We have therefore taken the approach of suggesting specific resources. This shouldn’t, however, dissuade you from dipping into the wider range of material available. We found http://www.smtinfocus.com/leadfree.html a generally useful source for this topic.
We have also made use of a number of reports produced by the solder materials team at the National Physical Laboratory, which are freely downloadable as PDFs from NPL at http://www.npl.co.uk/ei/publications/, or you can search at http://libsvr.npl.co.uk/npl_web/search.htm. For most reports there is a useful abstract, and the full text can be accessed (apparently without limit) once you have filled in your personal data. [The form will auto-complete once you have accessed your first report; if you have a fast web connection, it might save you time to browse the archive and download at a single session all the titles you think you might need]
A word of caution! Your files on this topic can “grow like Topsy”, so it’s well worth putting a ‘filter’ on your review of the material by asking the question “What impact will the uncertainty about the reliability of the lead-free system have on my customers?” Risk Management is one of the most important aspects of lead-free implementation, and this will influence both the type of evaluation to be carried out and the level of resources that need to be applied to process validation.
Assemblies fail for many different reasons, and both the nature of the failure and its consequences depend on the reason for failure and on the application of the product. An intermittent failure that might be acceptable in a domestic environment may be totally unacceptable in a telecommunications or mission-critical application. So we need to understand the environment faced by our products and the expectations of the end-customer as to what constitutes reliability. This is not a module about failure, but we have included some links to material in the AMI topic area in case you wish to read further.
For the rest of this unit we will be focusing on failure modes that are related in some way to the lead-free process, but it is important to keep in mind that these are only some of the potential causes of malfunction. For example, we haven't yet mentioned failure induced by humidity, not because it isn’t important, but because you will meet those effects later.
We suggest that you start by looking at some general articles that we hope will illuminate your studies and in particular give you some thoughts about the reliability and compatibility issues involved in implementing lead-free.
Global Trends in Lead-free Soldering, John Lau and Katrina Liu, Advanced Packaging, January and February 2004 (two parts)
NPL’s Lead-free soldering page
Managing lead-free compatibility, Dongkai Shangguan, Circuits Assembly, November 2003
Our experience with tin-lead solder has been that the material gives joints that are, broadly speaking, over-designed and more than adequately strong for purpose, although the margin has reduced in recent years with the introduction of finer pitch surface mount devices, particularly area arrays with small solder balls. In such cases, the CTE differences between component and substrate can result in joint fracture after moderate amounts of temperature cycling. That apart, the general consensus is that, if a joint looks right, then it will be reliable. Certainly for practical applications, where changes of temperature are relatively slow, and the homologous temperature of the solder relatively high, the ability of solder to ‘stress relieve’ by mechanisms that involve creep means that brittle fracture is relatively rare.
Did that paragraph make sense? If not, now might be the time to step aside from the main unit text and review some material on the strength of solder and the way joints fail.
Read our brief on Failure in solder joints carefully, so that you really understand the types of failure that can occur. Note that there is a subsidiary section on stress caused by thermal mismatch.
If you find this too daunting, we suggest three sources of additional information:
In this context, ‘modelling’ aims at creating a simplified view of the real world from which we can derive equations for the time-dependence of failure, enabling us to accelerate the testing and validation process. Our preliminary reading is intended to expose the issues and indicate the type of work carried out in reliability evaluation exercises.
The following are worth browsing through:
Results of Comparative Reliability Tests on Lead-free Solder Alloys, Grossmann et al, Electronic Components & Technology Conference, San Diego, 2002
Lead-free Soldering for CSP, Gordon Gray, IPC Lead-free Conference, March 2004
HALT (highly accelerated life test) for Lead Free Soldering, Howell B. Schwartz and Phil Conde, IPC Lead-free Conference, March 2004
HDPUG’s Design for Lead-Free Solder Joint Reliability of High-Density Packages, John Lau et al, Apex 2003
For solder joints we have to consider separately each of the main failure modes – the overall failure rate will have a contribution for each of these, though the situation may be complicated if any of the failure modes interact. Bill Plumbridge identifies four main failure modes:
Which of these failure modes is dominant will depend on the application and on the properties of the material, which are determined by the basic alloy, by its heat history and by any manufacturing defects. In the case of a hand-held product, for example, fracture caused by dropping will probably be a more significant contributor to failure that changes in the temperature of its surroundings.
Can we create a model for any of these failure modes? Much theoretical work is still being done. For example, Plumbridge and Kariya have recently applied fracture mechanics to try to relate the crack growth resistance of solders to their life expectancy. They reported that the mathematical form of the model depended on the nature of the stress conditions, but that silver-containing lead-free solder performed either the same as or better than eutectic tin-lead.
As another example, creep has been explained by Martin Rist of the Open University as having three phases: a primary phase, in which creep increases; a linear, secondary phase; a tertiary phase, accelerating to failure. His model is of hard particles in a soft matrix, where defining a term for matrix creep and a “hardening parameter” allows an effective model to be built for primary and secondary creep. In order to produce a model that also deals with the tertiary phase, one needs to develop a “damage parameter”, which helps model the dependence of creep on the history of the part under test.
The details of his presentation are on the SMART Group website (25 March 2004 workshop): we haven’t put in a link because such mathematical modelling is well outside the scope of this module. But what you do need to understand is that:
In many brittle materials, such as glasses, a great deal of their strength derives from the surface. Given that small joints have relatively more surface, one might question whether they would be fundamentally weaker (or stronger) than their larger counterparts. However, informed opinion is that, given the relatively low strength of the matrix, failure takes place in the bulk of the joint, so that surface and interface conditions are not important.
If we have an understanding of each failure mode, and can devise a model of how reliability changes with the severity of the test, then we can carry out “short, sharp tests” that are related to real-life experience by a known acceleration factor.
Papers on joint failure models often make mention of Coffin and Manson, early workers in the field who devised a formula that relates cycles to failure to the shear strain in the material, and Arrhenius, whose ideas on “activation energy” form the basis of many predictions of how time-to-fail will vary with temperature. Quite a lot of statistics lie behind implementing this approach, but there is a simple explanation at this link.
As you will have seen from some of the typical evaluation exercises referenced above, it is common to apply test conditions that are broadly representative of hard use, rather than to attempt to initiate specific different types of failure. Typical results of this pragmatic approach are test schedules that combine thermal cycling with drop and vibration. These are readily understood as demonstrating a module’s fitness for the intended purpose, but are usually not standardized, with the result that it can be difficult to compare results from different workers. It can also be difficult to compare the results for lead-free materials with those for eutectic tin-lead, and the need for “base line metrics” has itself resulted in several studies.
Even a casual review of test results will show attempts made to draw conclusions about the behaviour of the joints from the (often fairly limited) data collected. Unfortunately, extrapolating from limited data is dangerous, particularly when one takes account of the finding that the acceleration factors for tin-lead and SAC are different; for accurate predictions, we need an understanding of the failure mechanisms and not just an understanding of the mathematics and statistics behind the empirical approaches.
“Reliability is getting more difficult to achieve and there are no fixed goal posts – the trend is towards higher reliability under worse conditions. In order to prove that assemblies can meet these requires longer tests, and we are heading towards a bottleneck where purely empirical tests will take too long, so we need more understanding of the fundamental causes of failure.”
Bill Plumbridge at SMART Group Workshop, 25 March 2004
The holistic nature of the requirement is seen by what NPL describe as their ‘tool kit’ – look at the diagram at this link and at the complexity of the task of building a coherent body of knowledge that will relate practical observation to underpinning theory.
Reliability and Evaluation of Lead-Free Terminations, Terry Glascock, IPC Lead-free Conference, March 2004
Surface Mount Assembly Evaluations with Lead-Free Solder Pastes, Jasbir Bath and Emmanuelle Crombez (Solectron)
Materials and process considerations for lead-free electronics assembly, Karl Seelig and David Suraski (AIM)
Are Lead-Free Solder Joints Reliable? John Sohn, Circuits Assembly, June 2002
Converting from eutectic tin-lead to lead-free solder involves moving to a material that is less dense, requires a higher process temperature, and generally has a less attractive finish. These factors unavoidably lead one to anticipate that the change will have an adverse impact on reliability. During the 1990s this concern stimulated a number of broadly-based investigations. Their focus was on evaluating a range of different materials with a view to settle on a single material that would be as nearly as possible an equivalent to eutectic tin-lead, and then to establish appropriate process conditions and verify the integrity of the resulting joint. A summary of the work undertaken is available at this link, from where there are further links to descriptions of the projects collated by Bob Willis.
The recommendations from early projects are well summarised at the Indium web site:
“The group comprising tin-silver-copper alloys is considered the mainstream alloy system that will replace tin-lead. This family of close composition solder alloys is near-eutectic, with acceptable thermal fatigue properties, strength and wettability.
“The tin-silver eutectic alloy has a history of use, but has a higher melting point and exhibits poorer wetting than the SAC group of solder alloys.
“The tin-copper eutectic will find application in wave soldering due to its lower cost (contains no silver). However this alloy has the disadvantages of a higher melting point than the SAC alloys and may corrode iron containing solder pots.
“The tin-silver-bismuth composition was found by NEMI and NIST to have exceptional thermal fatigue performance, better wetting and a lower melting point than the SAC group of alloys. However, if any lead is present on component terminations or PC board pads, a low melting ternary tin-lead-bismuth phase can form which has a melting point of 96°C. Therefore NEMI has recommended that general use of this alloy should be avoided until it is assured lead on component terminations and board pads has been completely phased out, perhaps 7–10 years.”
There was thus reasonable consensus that lead-free joints were reliable under normal working conditions, though this might not be the case at elevated temperatures or if an otherwise lead-free joint had been contaminated by lead from components. The research also found that:
The mention of creep resistance improvement by adding silver and copper totally ignores the enormous changes that take place as the temperature of the alloy is altered. For example, there is a ×100 difference in creep on going from –10°C to +75°C.
This is consistent with the fact that solders have a high homologous temperature. This means that their mechanical properties vary according to temperature, strain rate, microstructure and thermal history. This is the reason why, when we want to carry out reliability modelling or life prediction, we need to use values of the mechanical properties that are appropriate under the service conditions.
The differences between lead-free and lead-containing materials are well illustrated in the sequence of graphs in Figure 1. Note that, the higher the temperature, the higher the creep strain rate difference.
Source: John Lau
In parallel with the increase in creep as the temperature increases, one observes the expected reduction in strength (Figure 2). However, there is also a reduction with strain rate, as much as by a factor of ×2, a value of 10MPa at 10–6 strain rate. However, ductility is very little changed by temperature, the typical value of 20% for a solder being generally regarded as acceptable.
Which lead-free solders are the strongest2 depends on the application conditions, but the usual ranking is SnAgCu stronger than SnAg stronger than SnCu, and the last of these is approximately the same strength as SnPb.
Viewed from the perspective of the important material properties of a solder, its melting point and conductivity are the go/no go parameters. However, a range of mechanical properties are becoming increasingly important, especially for small surface mount joints. These include the solder’s strength, ductility, impact, toughness, fatigue, creep and resistance to thermal fatigue. Which of those properties is most important for reliability will of course depend on the service conditions of the joint, and this means that different alloys may be chosen for different critical applications.
When it comes to assessing their usability, we need to take into account other aspects of the alloys, as well as these mechanical considerations. IPC reported3 in 2003 on a series of “round robin” tests on lead-free alloys that involved a number of companies. They focused on three SAC alloys with varying copper and silver contents and characterized them in terms of their melting behaviour, their wetting balance responses (IPC-TM-650 Method 126.96.36.199) and their solder spread (IPC-TM-650, Method 2.4.46). Interestingly, analysis of the results indicated that the differences between alloys were small as compared with the variability between different test locations (at all seven sites).
As well as the assessment of properties affecting the process, the IPC Council developed a long-term reliability test programme involving monitored assembly and test of two separate test vehicles and the cycling to failure of the assemblies using both fast and slow thermal cycles. This work is ongoing.
The first generation of lead-free materials, introduced as a result of work such as the IDEALS project, produced three ‘favourite’ materials, SnAgCu, SnAg and SnCu. It is these, and their many minor variants, that are forming the basis of the industry’s initial compliance with the EC Directive. The exceptions are mostly commercial products originating in the Far East, where lower melting materials containing bismuth or zinc (only suitable for less critical applications) are preferred on the grounds of their lower melting point and cost. The search for a lower melting point has also generated a number of more complex and expensive (often quaternary) alloys, as was mentioned in Unit 4.
But is it possible to create new materials that have improved characteristics? Fuelled by observations such as the major changes in tin-copper for wave soldering that are produced by only a minute amount of nickel, there is a hope that a greater understanding of microstructure may make it possible to ‘tailor’ new types of solder.
Although there is no report of volume product being made by the technology, the team from Timken4 showed that adding trace amounts of nanometer-sized particles of copper, nickel and iron to eutectic Sn3.5Ag solder altered the solidification process and influenced the microstructure, forming second phases in the cooled composite and increasing its microhardness. Similar work was carried out by Jeff Sigelco5 with a range of alloying elements that included reinforcement with particles of the Cu6Sn5 intermetallic.
Work that is similarly radical, although following different paths, is being attempted under the European COST531 programme on ‘second generation’ alloys. This work is theoretically based, rather than empirical, and materials are being researched from first principles. However, the outcome of these studies is reported to be “a few years yet” away.
Most of the basic metallurgical experiments have used large specimens, in order to be able to make meaningful tests on sections of metal with a controlled geometry. Real joints of course don’t look like test pieces! But not only do joints not behave like bulk solder, they are rarely perfect. Look at this link for a reminder of some typical soldering problems.
The reasons for imperfection include problems with solderability and wetting, many apparent component failures being in reality failures to make an adequate solder joint. Both the surface finish of a component and the conditions under which it is stored may affect the wettability of the component and hence its fitness for purpose in the assembly; board and component solderability is extremely variable (Figure 3), and damp and heat combined will do much to reduce the wettability of even the more robust surface.
The situation becomes worse when using lead-free soldering systems. Higher temperature is one factor that may degrade solderability; a second is the known reduction in wetting performance compared with tin-lead eutectic. Both these make an impact on the joint profile and solder volume. Of course, processes can be adapted to prevent a reduction in yield, and correct flux selection is important.
The higher temperature and longer time can also affect the reliability of the joint because too thick an intermetallic layer is formed, though enough time has to be allowed for wetting to take place – it’s all a question of getting the right balance.
Kester’s Flux and solder paste considerations for lead-free soldering contains a salutary reminder of the typical defects that can increase when moving to lead-free assembly.
For some reflow guidelines, read Reflow profiling: Time above liquidus by David Suraski of AIM.
Lead-free soldering causes reliability risks for systems with harsh environments by Hans Danielsson shows how, in an automotive application, thicker intermetallics can have a negative effect on long-term reliability.
As we mentioned earlier, the visual appearance of the lead-free joint is typically slightly different from its lead-containing counterpart, although the extent of the difference will vary according to the process and the fluxes used. Certainly, when examined closely, there is likely to be evidence in the lead-free joint of its different structure, that is of particles in a tin matrix rather than the alternating lamellae created when tin-lead solder cools.
Some defects that make joints less than perfect come from the printing process, which can affect solder volume. Some of these changes result from the trend towards smaller components and pads, for which assemblers have tended to specify thinner stencils. Unless stepped stencils are used, or pads for critical components can be printed oversize, this inevitably means less solder in the joint. The conclusion from NPL, reported informally by Milos Dusek, is that the change in reliability is around 50% for one-third of the thickness.
We remarked in an earlier unit that the slower wetting of lead-free solders often resulted in less than complete pad coverage (Figure 4), giving rise to concerns about reliability, particularly with OSP finishes, where the corners of the pads will be of a different colour, and hence easily visible. You might like to explore this issue, and come to your own conclusion as to whether or not this is a reliability issue.
Does incomplete coverage of the pad by solder reduce the reliability of a joint? Be creative in your choice of search terms!
The differences in wetting characteristics also give rise to differences in defects such as solder wicking and solder extrusion (also referred to as solder beading when associated with chip components). These have already been remarked on in Steve Dowd’s presentation.
A more severe problem, and one that has been extensively investigated, is that associated with the formation in the joint of areas containing a low-melting material. This was first noticed in 1994 with the occurrence of fillet lifting (Figure 5), a phenomenon that is particularly noticeable when bismuth-containing solders are used, but occurs with many lead-free solders when used with lead-containing terminations. You should already have researched this topic in Unit 6; if not, follow this link.
The use of lead-free solder in combination with lead-containing terminations also exacerbates problems of ‘secondary reflow’. This is where a previously-reflowed joint melts for a second time when the assembly is exposed to soldering temperatures, for example when wave soldering a product that has already been reflowed on the top side. Secondary reflow occurs with standard tin-lead materials – in the past, it has been more common when wave soldering in nitrogen, because the closed environment leads to greater internal heating – but the problem is potentially much worse at the higher temperatures used by lead-free materials.
With double-sided reflowed assemblies made with any type of solder, some degree of secondary reflow can be anticipated, because the whole board experiences reflow temperature. It was for this reason that underside components used to be glued, until it was discovered that there was sufficient surface tension in the solder to retain even quite large components. However, this secondary reflow can become a problem leading to open-circuit joints (Figure 6) if a board is flexed or vibrated, or if the flux becomes ineffective following its first exposure. Secondary reflow can also occur during the rework of double-sided products, a by-product of using sufficient background heat to get good joints. Unintended reflow is especially likely when the components being reworked have a large thermal mass.
Provided that the solder joint is homogenous, made of a single material, secondary reflow is not important. However, the effect can give rise to defects where there is any low-melting component of the solder due to contamination, especially where this occurs at the interface with the termination. Unfortunately, as described by Angus Westwater in this slide, as a joint cools, any low-melting material moves preferentially6 to the areas that are the last to cool, typically ending up in the region between joint and board.
Contamination of the solder is a somewhat contentious issue. Some believe that ta lead-contaminated joint will be less reliable (but against what standard?), others that a small amount of lead may even enhance reliability. Read some of the evidence and draw your own conclusions. If you make critical products, you’ll probably feel as we do, that, in order to gain real confidence about the integrity of the joints for your application, a substantial amount of specific evaluation will be required.
Lead contamination in lead-free electronics assembly, Karl Seelig and David Suraski (AIM)
Lead-Free and Mixed Assembly Solder Joint Reliability Trends, Jean-Paul Clech (EPSI), APEX 2004
Reliability of Lead-Free Chip Resistor Solder Joints Assembled on Boards with Different Finishes Using Different Reflow Cooling Rates, A.R. Zbrzeznya et al, IPC Lead-free Conference, March 2004
Lead is not the only possible contaminant. We made mention of the copper curve in Unit 4 (Figure 3), and a build-up of copper will certainly make the joint look grainier, as well as significantly raising the melting point of the solder. Fortunately there is a way of restoring an acceptable copper level as David Suraski points out in this 2004 interview (video; transcript).
So far in this section we have moved from defects that are easy to spot, to ones such as fillet lifting and secondary reflow that are less apparent. Voiding is another way in which a joint can be less than perfect, but which is impossible to spot except by X-rays (Figure 7). Voids are created by the entrapment of volatiles produced from fluxes during the soldering process; their incidence depends critically on the materials and choice of process (see New No-Lead Solder Pastes and Reflow Techniques, Jim Raby and David Heller), and they can only be totally eliminated by using a vacuum-soldering regime.
So the question is not whether voiding is allowable, but what extent of voiding is acceptable with no impact on joint reliability. Much work has been done in this area, and the consensus seems to be that a degree of voiding may even be beneficial, whereas large voids, especially in the small joints associated with BGAs, can make interconnections less reliable. More information at this link.
During your study of this module you will already have seen the effects of over-temperature on board and components. Our list of components that are affected includes:
Those with liquid or gel contents, such as electrolytic capacitors, where internal pressure can build up and destroy the component
Those where reflow is above the heat deflection temperature of the casing, so that it either cracks (Figure 8) or becomes distorted.
With boards, exposure to soldering temperatures can result in the warping of structures with uneven lay-up or copper, or the sagging of thin laminates that are not supported during processing. In all cases, if the board is allowed to cool whilst in its distorted state, a permanent bow or twist is likely to result (Figure 9).
Over-temperature can also result in discolouration, both of boards and metal work, though this is usually purely a cosmetic issue7. Over-temperature can also harden flux residues, creating problems with probing, so that a perfectly good assembly may appear defective.
Temperature changes will of course have an effect on other board materials apart from the solder, so temperature cycling (either during life or as a deliberate test) may produce structural failures in the board, and not just destroy the joint. The mechanisms by which failure occurs are described at this link.
Temperature on its own can result in problems; combined with moisture, it can destroy components. The reason is that most polymer systems absorb moisture, the extent of this depending on the temperature, the humidity and the resin type. Whilst the absorption process is reversible, a problem occurs when the moist resin system is suddenly heated, as during the soldering operation; the result can be delamination of a printed circuit board (Figure 10) or ‘popcorning’ in a semiconductor device. More information on humidity absorption is at this link.
The problem is well understood, but becomes more severe at the higher temperatures of lead-free reflow. Some components are more sensitive than others, and a moisture sensitivity classification system has been developed for both reflow and wave soldering, with an associated standard for the dry storage of sensitive components and marking packaging in order to warn of the hazard. The specifications are all freely downloadable (IPC now charge for this information) from http://www.jedec.org/ and are:
Reflow soldering – J-STD-020C Moisture/Reflow Sensitivity Classification for Non-Hermetic Solid State Surface Mount Devices
Wave soldering – JESD22-A111
Storage/marking – J-STD-033B Handling, Packing, Shipping and Use of Moisture/Reflow Sensitive Surface Mount Devices
Revisions to the J-STD specifications are in the ‘working draft’ stage, are not expected until the end of 2004.
“The increased soldering heat associated with lead-free soldering has proven to highlight manufacturer storage systems that are poor or marginal with respect to humidity ingression control, that is they experienced no package humidity expansion problems because they were lucky, are now seeing some issues with the increased lead-free solder heat and duration.
“You should ensure supplier lead-free part number MSL levels (which may have been modified because of lead-free) and store as per IPC standards. I always say there is no real difficulty to comply with IPC ESD and MSL levels other than good housekeeping resulting in confident product reliability.
“Just another turning of the screw, reducing the process window to avoid failures.”
Angus Westwater on SMART-e-link 10 August 2004
Common wisdom has it that most packaged devices will need to be reclassified by a grade or two on the moisture sensitivity scale. However, as you will see from this example, reclassification may not always be necessary.
We have already discussed the changes needed to the board metallization; to make a reliable contact the component terminations also need to be lead-free. The choice of termination materials will depend on the application and on the preferences of the component manufacturer. Angus Westwater’s presentation lists the materials and discusses how his company are tackling the challenge of converting.
But the plating material itself has an impact on reliability. For example, the plating may be brittle and unable to withstand the lead-forming process. This has already been reported in relation to a lead-frame formed post-plating during component manufacture, but similar problems might affect processes carried out by the assembler. At the same time, where the plating is not intended to be soldered, but forms part of a contact system, the contact resistance of the plating needs to be verified; the contact resistances of different plating materials have different initial values and will change in different ways on exposure to atmospheric, depending on the underlying mechanisms of deterioration of the surface. Finally, for plating designed to be solderable, there are usual issues of initial solderability and the changes that takes place with time.
A final problem in this area relates directly to the fact that lead-free solders are generally more rigid. As a result, more stress can be transmitted to chip devices. Monolithic ceramic chip capacitors are particularly prone to stress fracture, as explained at this link, but such problems can affect all parts with a chip format (Figure 11).
|Has the electronics industry missed the boat on Pb-free? - Failures in ceramic capacitors with Pb-free solder interconnects||Nathan Blattau and Craig Hillman,||ISBN Number||IPC Lead-free Conference||March 2004|
We have already seen in Unit 7 that changing the board finish to ENIG gives the potential for black pad failures. These are batch-dependent, intermittent in their occurrence, and not easy to see (Figure 12).
And changing the metallisation materials for both boards and component terminations may give rise to other problems. For example, makers of equipment for mission-critical applications in moist environments have concerns about silver’s propensity for electromigration; more about this effect at this link. However, as our page on immersion silver linked in Unit 7 concludes, the whole issue is currently being debated, and engineers are likely to be reassured by the thinness of the silver layer involved and the fact that this is totally dissolved in the solder.
As the section on alternative finishes in Unit 7 indicates, major players are proposing the use of an immersion tin finish. Cheaper than silver and more resistant to tarnishing, tin provides an acceptable solderable coating, but has two reported reliability concerns.
The first of these arises because tin exists as two different allotropes. The normal form that is stable at room temperatures is ductile, but at low temperatures it can transform, with an accompanying 27% volume change, into a grey material with no physical strength. Referred to as “tin pest”, this is fortunately extremely rare except in non-electronic applications involving comparatively large masses of tin.
The second reliability issue concerns the growth of tin whiskers, a topic that has been attracting much more attention in recent years, because the short-circuits created can destroy low-current circuitry. Figure 13 shows a typical whisker; this one grew from a copper lead-frame, and was produced during an NPL project on different plating technologies
Source: National Physical Laboratory
Read our description of tin whiskers at this link, and then search the web for the latest information on this topic.
See if you agree with our conclusions that:
Conductive Anodic Filamentation (CAF) is a specific form of electromigration that occurs within the bulk of fibre-reinforced laminates, caused by copper salts being deposited inside the fibre bundle, forming a conductive path. Though not a new problem, CAF has grown in importance in recent years with the drive to smaller board geometries and the increasing use of electronics in demanding environments. Additional focus has come from the conversion to lead-free, as this affects both the laminates that are specified and their conditions of use.
CAF occurs when the fibres separate from each other and the resin, allowing the ingress of moisture, after which filament growth takes place by the mechanism described at this link. Figure 14 shows glass fibre bundles in a multilayer board separating, with CAF visible between the bundles.
Source: National Physical Laboratory
The conclusions from NPL’s January 2004 study on the susceptibility of laminates to CAF failure confirmed that smaller geometries significantly reduce CAF resistance, and higher voltages reduce the time to failure. They also found that
Evaluating both laminate materials and suppliers is important:
Note that CAF is distinct from the surface phenomena of dendritic growth and corrosion, both of which influence the reliability of an assembly exposed to environmental challenges that include humidity. More information at the links on dendritic growth and corrosion.
|Developing a test to characterise internal stress in tin coatings: phase 1||M Wickham, A T Fry, C Hunt,||NPL Report MATC(A)148||October 2003|
|Susceptibility of glass-reinforced epoxy laminate to conductive anodic filamentation||A Brewin, L Zou, C Hunt||NPL Report MATC(A)155||January 2004|
|Meeting the Challenges of Environmental Leadership: Lead-Free and Beyond ,||Dongkai Shangguan, IPC Lead-free Conference ,||March 2004|
|Lead-free manufacturing issues and transition to lead-free soldering ,||Jasbir Bath (Solectron)|
Soldering lead-containing finishes with tin-lead eutectic solder produces a reliable joint; soldering a surface without lead with a lead-free solder equally produces a sound joint, though there will be some differences in microstructure and mechanical performance. Unfortunately, the practicalities of re-engineering, process development and component supply mean that, of the three possible transition routes shown in Figure 15, the conversion of components and solder shown as Route A is unlikely to be available for most applications. [Lau suggests that this route will be confined to products such as network servers that are exempt from lead-free status until 2010]
Source: John Lau
But what happens if we mix lead-free and lead-containing? Which of the two routes with an intermediate stage will create a reliable joint?
Re-read carefully the section on transition in Global Trends in Lead-free Soldering by John Lau and Katrina Liu, Advanced Packaging, February 2004 (Part 2).
In the light of your knowledge of solder joint reliability, do you agree with the comments made in that article, and would you choose Route B or Route C?
Making the move to lead-free has significant reliability implications, especially when it comes to mounting area arrays with small solder connections; standard BGAs present a challenge, but the situation gets worse with µBGAs and flip-chips. In responding to the lead-free consultation, EIA specifically asked for exemptions for flip-chip use (see p11 of their submission).
For BGAs there is some consensus that an SAC ball can be used for Sn-Pb solder reflow, but this needs a minimum peak reflow temperature of 225°C to ensure that the ball is totally molten. As a result of the reduced wetting and spreading and higher surface tension, there will be less self-alignment, so the devices need to be placed more accurately. Also there are likely to be more voids in the solder joints. And the measured reliability on temperature cycling depends crucially on the temperature range and the peak temperature. Altogether a topic to be watched carefully . . .
Making a reliable product is all about getting every aspect right; the right solder, the right process and the right components. So the task involves many ‘nitty-gritty’ activities, including:
Also, given that it is likely that most companies will have to run lead-containing and lead-free processes in parallel for some time, careful attention has to be given to ‘error-proofing’ operating procedures wherever possible.
All these are issues that we will need to address during our implementation phase.
To see how other people have tackled the challenges, read:
|Visteon Experience on Lead Free||Stephen Wong||May 2002|
|Reliability of lead free solder joints on manufacturing conditions ,||Th. Herzog et al, SMTA International Conference||2001|