Hacker News new | ask | show | jobs
by robomartin 1569 days ago
I have dealt with the tin whisker problem in the context of aerospace applications (both space-borne and terrestrial flight), including extensive consulting with subject matter experts from NASA.

The bottom line is quite simple:

Tin whisker growth onset is a stochastic process. We cannot predict when it will start and we cannot prevent it.

Once they start growing it is almost impossible to contain them. They will poke through conformal coatings such as parylene and arathane. If they don't, they will buckle (coil-up) under the coating.

While buckling sounds like a desirable outcome, this could lead to shorting of adjacent contacts in todays fine pitch integrated circuits and components.

Growth rate can be in the order of 10 mm per year. This means that adjacent leads of something as mundane as a SOIC-16 package can be shorted by a tin whisker in 28 days or less.

The take away is: There's nothing we can do about tin whiskers that is 100% guaranteed to prevent growth or slow it down by a non-trivial amount. The only path that prevents their growth is to use lead-based solder. This is why, as an example, we would do such things as send out BGA's with RoHS compliant solder balls to be re-balled with leaded solder.

Time for a bit of a rant: All my work in this area led me to look at the RoHS initiative as yet another example of something that, while well intentioned, it will likely have precisely the opposite effect from what was intended.

The fact that lead-free solder is susceptible to tin whisker growth means that 100% of all consumer electronic products are ticking time bombs when it comes to failures. This means that all kinds of consumer, commercial and industrial electronic products will fail over time in ways we might not be able to explain. The reason for this is that nobody does deep forensics when products fail. There is no reporting from the likes of Apple, Samsung, LG, Visio, Sony and myriad other manufacturers on failure rates and causes. In fact, they might not even have this data as consumer, commercial and industrial users simply replace the devices as they fail and move on.

In other words, it is likely RoHS has caused --or will cause-- massively more garbage in landfills. As a simple data point, my 40 year old HP-41 calculator still works perfectly fine. It is impossible to imagine a RoHS-compliant calculator not ending up in a landfill way earlier than 40 years.

There was a bit of a movement to roll back RoHS around the time it was being enacted. Going up against many nations and politicians using "save the planet" to get elected proved impossible for those who rightly brought-up that the transition to lead-free solder required far more research before we fully understood the potential consequences.

It wasn't about not wanting to go lead-free, it was about making the move when the science and math indicated that it would not create the massive problem we now likely have on our hands. The data on electronics waste due to tin whiskers is probably impossible to find. It might not even exist. Which is a tragedy.

If you want to learn more about this, here are a couple of good links:

https://nepp.nasa.gov/whisker./background/index.htm

https://nepp.nasa.gov/whisker/reference/tech_papers/kadesch2...

https://www.google.com/search?q=tin+whisker&hl=en&tbm=isch

https://web.calce.umd.edu/tin-whiskers/

8 comments

> In other words, it is likely RoHS has caused --or will cause-- massively more garbage in landfills. As a simple data point, my 40 year old HP-41 calculator still works perfectly fine. It is impossible to imagine a RoHS-compliant calculator not ending up in a landfill way earlier than 40 years.

Environmentalists can only wish that people were disposing of their electronics because of tin whiskers. Long lived consumer electronics needs a cultural overhaul more than it needs leaded solder.

I did not know these whiskers had been implicated in the Toyota unintended acceleration scandal: https://nepp.nasa.gov/whisker/reference/tech_papers/2011-NAS...
The Restriction of Hazardous Substances Directive (RoHS) is not about preventing garbage in landfills, and frankly it's been 18 years and the sky is not falling.
The sky is not falling in large part because the transition to lead-free solder was simultaneous with the transition to electronic devices that are not repaired and which frequently have a lifetime not much longer than their warranty time.

Most current electronic devices are dumped much earlier than when they would fail due to the tin whiskers.

Many consumer electronic devices made 50 years ago are still usable without any problems caused by the aging of the soldering or of the semiconductor devices (but old electrolytic capacitors may have to be replaced). The electronic devices that are made now do not have any chances of such a long lifetime, with the exception of a few devices made for special requirements, e.g. military/aerospace.

Consumer electronics used to break down all the time 50 years ago. Metal whiskers are not even close to being relevant when other factors impact reliability and durability to a far greater extent.
Consumer electronics used to break down all the time, but in almost all cases that was due to manufacturing defects, which were much more frequent, because many operations that are now automated were still done manually then.

The consumer devices which survived infant mortality, because they were free of manufacturing defects, had a negligible aging rate after that.

Modern electronic devices have far fewer initial manufacturing defects, due to automated production, but all age much quicker, due to very small component sizes, lower safety factors, surface semiconductor devices (MOS transistors) instead of bulk semiconductor devices (bipolar transistors), lead-free soldering and other similar changes in technologies.

They used to break down and they could and were repaired. Nowadays if something breaks it goes in the garbage can. I had an extension cable which died staying in the basement for a year. When i opened it to check the reason i was shocked. It looked like a spider net made of dust but this was metal.
> When i opened it to check the reason i was shocked.

Sounds like you should unplug it before you open it up.

It was not working. I opened it to check why.
Funny!
> Consumer electronics used to break down all the time 50 years ago

But they could be easily repaired.

My mid-1980s digital alarm clock came with a full schematic. Can you imagine that today?
Actually, yes I could imagine that quite easily. Just buy "hackable" products which do that as a matter of course. A digital multi-purpose device with an alarm-clock form factor would not even be especially complicated to make.
Once things started going to multi-layer PCBs it was the end of reparability. It’s too bad because I have fond memories of fixing broken components on PC hardware and game consoles, even as someone who’s not an expert and simply a hobbyist.
> Once things started going to multi-layer PCBs it was the end of reparability.

Very true, however having full documentation would help nonetheless. My water heater electronic board cost over €250 to replace, although it contains less than €20 parts; a preprogrammed uC makes it impossible to replicate it. If it had public hardware and firmware documentation, someone could repurpose a similar but cheaper board or replicate the functions using a different rugged enough uC board, which would also likely bring down the retail price of the original spare part.

My understanding is that the new manufacturing techniques stop many people but usually not skilled technicians. It's component serialization, keeping schematics secret, and exclusive supply chains that are the real problem.
"Consumer electronics used to break down all the time 50 years ago"

Citation required

Yes but the parent is not stating that. They are stating that a "save the environment" effort is not very environmentally friendly if it makes more waste.
Saving the environment by reducing actively, acutely hazardous materials and saving the environment via reducing landfill/carbon emissions are completely different goals.

As others have pointed out in this thread, reducing landfill waste from electronics is a much more complex problem and just adding leaded solder will not solve it.

That assumes waste reduction is the goal, rather than lead reduction. With toxic, intelligence reducing [1], materials like lead, maybe some extra waste is a perfectly good trade off.

1. https://news.ycombinator.com/item?id=30600539

Last i checked tin was not an edible material. There is a small difference between having tin dust, which you cannot contain, and having electronics containing lead which you can colect and store in a warehouse ( asuming you want to address the problem in the first place).
Are you saying that electronics devices continuously let out tin dust so we end up breathing in this stuff?
They are assuming that these devices would be used long enough for tin whiskers to become a problem. I seriously doubt that RoHS will be causing more waste - because at the point that devices become unusuable, they are on in the trash anyways for completely unrelated reasons (think: bezel to large to be popular, plastic backshell instead of metal or glass, device is too thick, device is too heavy, ...).
Tin whiskers continue to grow even after you device is in the landfill. You basically have a dust of tin which goes into landfil.
Uhm, that is exactly the point that I'm making? That the goal was never save the environment, but rather reduce the exposure to toxic-at-any-concentration things like lead?

Hence why I quoted the program name.

Devices are disposed of (becomes waste) long before they become broken from whiskers. That makes RoHS a net-win by reducing toxic materials from landfills full of phones with broken screens and kitschy doodads with broken plastic.
Do you happen to know if tin whiskers have anything to do with passing current? (in other words, will a device that's in constant use develop them faster than a physically identical device that's switched off and in storage)
> will a device that's in constant use develop them faster than a physically identical device that's switched off and in storage

Storage was one of the areas I researched extensively. It was like that scene in the movie "Clear and Present Danger" [0] in that the answer was always the same frustrating one regardless of conditions:

Stochastic growth onset; 0 to 3 years; can't predict growth start or rate; can't stop them; they will penetrate almost anything practical you can put on a board.

[0] https://www.youtube.com/watch?v=PqtjbWJPIgQ

Passing current means thermal cycling...

And thermal cycling definitely accelerates whisker formation.

Yes. Voltage has a big effect on whiskers.
Are there any easy ways to clean up a PCB that's developed whiskers? And once a whisker erupts on a given PCB, does that generally indicate that others are likely to form on that board in short order?
All else being equal, once growth starts it is likely to start everywhere on that board. This is a probabilistic assumption based on the likelihood of all of the solder on that board being from the same batch and having been applied with the same process parameters. The same cannot be said of the device leads, where each manufacturer and batch could very well be different.

It's quite a nightmare, particularly when you are trying to figure out if this stuff can kill people you want to send into space. The only real mitigation is lead-based solder and coatings on components.

Cleaning? That can be both dangerous and highly ineffective. The whiskers are very strong due to their molecular scale. Mechanical brushing might fracture longer whiskers. Then you have the problem of ensuring that they don't go under devices or in-between contacts. The process would likely have to be repeated many times and include both manual and automated optical inspection as well as x-ray imaging (which might not be able to detect fine whiskers). And then there's the reality that you probably don't want to inhale these things at all.

So, off to the landfill we go. It is likely better to build a new board than to try to clean one. I can't even begin to compute the delta in carbon footprint between making a board with lead-based solder that will last decades and the "clean/green" RoHS board that is sure to end-up in a landfill (cleaning/fixing it is bound to have a massively larger carbon footprint that making a new board).

Is this why space electronics often use wire-wound connections? Maybe welding the contacts together instead of soldering?

I assume for the really critical components, you'd need to avoid solder completely.

Unfortunately, all consumer electronics companies have this fetish for making products ever smaller and thinner. It dovetails with their profit motive: make things less reliable so we all have to buy more frequently.
Well actually, there is a way, but you might kill said PCB.

Heat everything up in an oven, the solder will reflow, and you might temporarily fix the board. It’s a similar idea to the Towel/Xbox 360 fix. I can attest to having successfully saved lots of random electronics this way.

All of this started with the eco-friendly alternatives to lead solder, I have a lot of old computer hardware and motherboards, and the hardware from the early 2000s is the least reliable, whereas most game consoles, motherboards, etc. from the 80s and 90s works flawlessly. To this day I swear by the leaded stuff for personal use, it flows better, doesn’t crack, and is superior in every way.

Miniaturization trend definitely doesn’t help here either…
RoHS does not just restrict lead. It also restricts use of mercury, cadmium, and several toxic compounds.
Does wisker form inside semiconductor also or do they use metals immune to that?
Yes they do but slower. In the past gold was used for wires from terminal to pads. Gold as far as i know is not so subceptible to form whiskers. But gold is expensive and now they use copper instead of gold.
Unfortunately, gold is also susceptible: https://nepp.nasa.gov/whisker/other_whisker/gold/index.htm