A surprising number of metals are capable of spontaneously growing whiskers. They do it under ordinary room conditions. The whiskers are thin, but up to millimetres long and highly conductive. They are death to electronics. We've known about this since 1946, and found a way to eliminate metallic whiskers, but we are now inviting them back.
At least aluminum, cadmium, gold, indium, lead, silver, tin, and zinc will grow whiskers. All of these metals have been important in electronics, though the use of cadmium is rare now that we understand its toxicity. Metallic whiskers are typically 0.5um to 10um in width, so they are difficult to see: compare this to human hair, at about 50um diameter. Their length is impressive; a specimen 23 mm long is the current record holder. With a growth rate as fast as 15um/s (that's a millimetre in 6.7 minutes) they can sneak up on you fast. It may not take more than a few minutes to produce a short-circuit between IC pins on 0.5mm pitch, or under a 0.4mm pitch BGA. NASA has published some beautiful photos of whiskers by Lyudmyla Panashchenko [1].
Tin whiskers are the ones that seem to get the most publicity, probably because they grow in modern solders. Back in the bad old days, solder was mostly tin, and trouble was avoided only because conductor spacing was mostly more than the several centimetres a whisker can span. As miniaturisation proceeded, whiskers were seen as a source of failures. In the 1960s Bell Labs did a good deal of work on whisker suppression in tin solder, and found that as little as 0.5% to 1% of lead did the job. Since then it has been common to add lead to solder. A eutectic alloy of 63% tin, 37% lead passes directly from liquid to solid, with no intermediate pasty stage, and was very popular. It has the lowest melting point available in tin/lead alloys, 183°C, compared to 232°C for pure tin. Problem solved.
Fast forward to the 21st century. Politicians with no engineering knowledge decree that all but 0.1% of the lead must be removed from solder. The makers of car batteries, who use about 88% of the lead in the US, are exempt, having $omehow convinced the politicians that they "need" lead. But the makers of electronics, who use 0.5% of the lead (exclusively in solder) must remove it [2]. Solution aborted.
Needless to say, there has been a huge search for other additives which will stop whiskers. Regrettably, our understanding of whiskers is slight [4]. We know which materials grow beards, but the spread in whisker density is eight orders of magnitude - and nobody seems to know what affects it. We have no idea how to assess the whisker-proneness of a coating, how to mitigate whisker growth without lead, how to model the impact of whiskers on reliability, or even how many unexplained failures are due to undetected whiskers - single whiskers can be the very devil to spot. We know that some conformal coatings help, but they are not the magic bullet: some whiskers can penetrate, and not all parts of the board are coated (consider the underside of a BGA).
Worst of all, we may be fooling ourselves. Most of the alternatives to lead in solder are more toxic, consume more energy, and cause more acidification and global warming [2]. In the search for acceptable solders without lead, only an alloy of 99.3% tin, 0.7% copper is level with 63% tin, 37% lead on these environmental metrics. With a melting point of only 227°C, it looks attractive - but small amounts of copper seem to encourage the growth of tin whiskers.
Other alloys involve bismuth, silver and antimony. Oddly, the highly toxic antimony is quite alright under RoHS regulations. Bismuth and siver, ironically, are extracted as a by-product of lead production. Both are non-toxic in themselves, but their extraction involves a lot of energy. Solder manufacturers are generally dismissive rather than helpful; there is no mention of whiskers on the Kester website, and Indium minimises the problem [7]. Quebec-based AIM seems to be doing research, and will present results later in 2014.
The military and aerospace industries would like to have nothing to do with all this. After all, the military has some very unfriendly uses for lead [3]. Manufacturers are closing down production of components containing lead, because the cellphone market is much larger than the military one. Increasingly the military, at only 1% of the semiconductor market, must accept lead-free components if they wish to keep up with modern developments in electronics.
What it all boils down to is that we no longer have a way to assess the reliability of our electronics. It may be no big deal if your cellphone dies a premature death. Lots of people change 'phones every year or two anyway, just to get new features. Consider the implications of reliability in industrial controls [5], fly-by-wire systems in a passenger airliner, your car [6], or the impending driverless cars.
Why are we doing this? To improve the environment, of course. Is moving away from small quantities of lead, in its metallic and hence non-toxic form, really better than moving to more toxic materials with greater environmental impact? What are the environmental consequences of added e-waste due to premature falure? If politics is the art of the possible, should we be allowed 1% lead in solder, to make electronics reliability possible?
References
1. Lyudmyla Panashchenko has wonderful whisker photos, both optical and using a scanning electron microscope, along with practical tips on how to see the beasties (and how easy they are to miss).
2. Lead-free Electronics Reliability - an update by A.D Kostic is a recent (2011) presentation, generally pessimistic about the reliability of lead-free electronics, and the environmental impact of lead removal.
3. Environmental Scientists in the Wild West shows why the military are unconcerned about lead.
4. A History of Tin Whisker Theory: 1946 to 2004 is a thoughful history of our knowledge of whiskers, showing that we know a lot of the science but little of prediction or prevention.
5. A nuclear reactor shutdown caused by tin whiskers.
6. Accelerator failure in Toyota cars may have been caused by tin whiskers. Toyota deny this, but have come up with no credible alternative explanation.
7. Dr. Ron Lasky's blog, in which he seems to say that failures in electronics don't matter, unless the equipment is "mission-critical".
At least aluminum, cadmium, gold, indium, lead, silver, tin, and zinc will grow whiskers. All of these metals have been important in electronics, though the use of cadmium is rare now that we understand its toxicity. Metallic whiskers are typically 0.5um to 10um in width, so they are difficult to see: compare this to human hair, at about 50um diameter. Their length is impressive; a specimen 23 mm long is the current record holder. With a growth rate as fast as 15um/s (that's a millimetre in 6.7 minutes) they can sneak up on you fast. It may not take more than a few minutes to produce a short-circuit between IC pins on 0.5mm pitch, or under a 0.4mm pitch BGA. NASA has published some beautiful photos of whiskers by Lyudmyla Panashchenko [1].
Tin whiskers are the ones that seem to get the most publicity, probably because they grow in modern solders. Back in the bad old days, solder was mostly tin, and trouble was avoided only because conductor spacing was mostly more than the several centimetres a whisker can span. As miniaturisation proceeded, whiskers were seen as a source of failures. In the 1960s Bell Labs did a good deal of work on whisker suppression in tin solder, and found that as little as 0.5% to 1% of lead did the job. Since then it has been common to add lead to solder. A eutectic alloy of 63% tin, 37% lead passes directly from liquid to solid, with no intermediate pasty stage, and was very popular. It has the lowest melting point available in tin/lead alloys, 183°C, compared to 232°C for pure tin. Problem solved.
Fast forward to the 21st century. Politicians with no engineering knowledge decree that all but 0.1% of the lead must be removed from solder. The makers of car batteries, who use about 88% of the lead in the US, are exempt, having $omehow convinced the politicians that they "need" lead. But the makers of electronics, who use 0.5% of the lead (exclusively in solder) must remove it [2]. Solution aborted.
Needless to say, there has been a huge search for other additives which will stop whiskers. Regrettably, our understanding of whiskers is slight [4]. We know which materials grow beards, but the spread in whisker density is eight orders of magnitude - and nobody seems to know what affects it. We have no idea how to assess the whisker-proneness of a coating, how to mitigate whisker growth without lead, how to model the impact of whiskers on reliability, or even how many unexplained failures are due to undetected whiskers - single whiskers can be the very devil to spot. We know that some conformal coatings help, but they are not the magic bullet: some whiskers can penetrate, and not all parts of the board are coated (consider the underside of a BGA).
Worst of all, we may be fooling ourselves. Most of the alternatives to lead in solder are more toxic, consume more energy, and cause more acidification and global warming [2]. In the search for acceptable solders without lead, only an alloy of 99.3% tin, 0.7% copper is level with 63% tin, 37% lead on these environmental metrics. With a melting point of only 227°C, it looks attractive - but small amounts of copper seem to encourage the growth of tin whiskers.
Other alloys involve bismuth, silver and antimony. Oddly, the highly toxic antimony is quite alright under RoHS regulations. Bismuth and siver, ironically, are extracted as a by-product of lead production. Both are non-toxic in themselves, but their extraction involves a lot of energy. Solder manufacturers are generally dismissive rather than helpful; there is no mention of whiskers on the Kester website, and Indium minimises the problem [7]. Quebec-based AIM seems to be doing research, and will present results later in 2014.
The military and aerospace industries would like to have nothing to do with all this. After all, the military has some very unfriendly uses for lead [3]. Manufacturers are closing down production of components containing lead, because the cellphone market is much larger than the military one. Increasingly the military, at only 1% of the semiconductor market, must accept lead-free components if they wish to keep up with modern developments in electronics.
What it all boils down to is that we no longer have a way to assess the reliability of our electronics. It may be no big deal if your cellphone dies a premature death. Lots of people change 'phones every year or two anyway, just to get new features. Consider the implications of reliability in industrial controls [5], fly-by-wire systems in a passenger airliner, your car [6], or the impending driverless cars.
Why are we doing this? To improve the environment, of course. Is moving away from small quantities of lead, in its metallic and hence non-toxic form, really better than moving to more toxic materials with greater environmental impact? What are the environmental consequences of added e-waste due to premature falure? If politics is the art of the possible, should we be allowed 1% lead in solder, to make electronics reliability possible?
References
1. Lyudmyla Panashchenko has wonderful whisker photos, both optical and using a scanning electron microscope, along with practical tips on how to see the beasties (and how easy they are to miss).
2. Lead-free Electronics Reliability - an update by A.D Kostic is a recent (2011) presentation, generally pessimistic about the reliability of lead-free electronics, and the environmental impact of lead removal.
3. Environmental Scientists in the Wild West shows why the military are unconcerned about lead.
4. A History of Tin Whisker Theory: 1946 to 2004 is a thoughful history of our knowledge of whiskers, showing that we know a lot of the science but little of prediction or prevention.
5. A nuclear reactor shutdown caused by tin whiskers.
6. Accelerator failure in Toyota cars may have been caused by tin whiskers. Toyota deny this, but have come up with no credible alternative explanation.
7. Dr. Ron Lasky's blog, in which he seems to say that failures in electronics don't matter, unless the equipment is "mission-critical".