TIL computers can sometimes grow crystals
TIL computers can sometimes grow crystals
Tin in solder or some other meals can form spiky crystals when under stress. These whiskers can form short circuits if not properly insulated or not alloyed with other metals.
You're viewing a single thread.
This can happen inside ICs and has been a known failure mode for high frequency processors for many years. I work in chip design, and we use software tools to simulate it. It's called electromigration.
This statement is not fully accurate. Whiskers in OP’s case are about (usually) tin whiskers that grow, often visibly, and then can connect (short) to unintended areas.
Electromigration is effectively when a large potential difference encourages ions to relocate to reduce the potential difference.
Big Whiskers have two methods of formation. The first way is that tin ions are able to move by becoming soluble in some form of water so they’re mobile. The other way whiskers can form is from stress alone. (Stress being force per area that compresses or tensions the metal in question, applied through a multitude of ways) Whiskers can be directed by electromigration so they form tendrils to a differing potential, basically purposefully ruining stuff instead of randomly shorting things.
Now in integrated circuits (ICs), there are extremely high currents running through extremely small regions. Electromigration in ICs is caused by electrons getting yeeted at extremely fast speeds, giving them significant momentum. They collide with ions in their path and dislodge the ions from their matrix. This can result in voids of ions preventing current from flowing (open circuits) or tendrils of ions making a path to an unintended area and connecting to it (shorting it). The tendrils here are also called whiskers, but are generated in a very different way (e.g., no water solubility or inherent stresses required) and on a significantly smaller scale. And probably not in tin.
The more you know!
The mechanism behind metal whisker growth is not well understood, but seems to be encouraged by compressive mechanical stresses. According to Wikipedia.
Electrons in metal always move the same speed, and potential differences in modern high perf applications are never above 3.3V. There are mechanical stresses in ICs introduced during manufacturing. So these cases aren't as different as you let on.
Anyway, point is, metal moves, we have some ideas why and can model some of them. From an engineering perspective these are both tin whiskers. We call whiskers made of copper and aluminum tin whiskers. You're describing a distinction without a difference.
The metal moves due to very different reasons. I would not say whiskers due to mechanical/residual stresses are due to "electromigration" - electromigration isn't even there since the wiki definition is "transport of material caused by the gradual movement of the ions in a conductor due to the momentum transfer between conducting electrons and diffusing metal atoms". You build stresses and strains into semiconductors for better mobility profiles, and I'm sure that can cause whiskers - but again, it's not electromigration.
Electromigration, as noted, plays a role in the form of encouraging stress whiskers to grow in a direction (with a very relaxed definition).
But in ICs, with their very unique extremely small scales, electromigration can directly form whiskers by moving individual ions via electron collisions. But the generation mechanism for those whiskers shares nothing with Big Whiskers generation mechanism. That's my point.
Electrons in metal do not always move at the same speed; they move at v=mu*E where v is the velocity, mu is the electron mobility, and E is the electric field. Crank the E, you go faster. At very high E fields you reach the electron saturation velocity where slowing factors limit the maximum speed - I assume in your IC world you're basically always there due to the extremely small regions (E = V/m; any V with m at nanometers is big E) which is why you claim that. But even then the electrons are accelerating due to the E field, smashing into ions and losing their momentum (mass static, so it's just velocity), and then re-accelerating. The saturation velocity is the average bulk motion of electrons but it's not a smooth highway, it's LA traffic (constant crashes).
Electrons can gain significant momentum, which is just their static mass times their velocity. Limited at velocity by the saturation velocity, current density is important for significant momentum exchange. Luckily ICs are so tiny that the currents they drive are massive current densities.
What you said originally is correct; it's just in ICs electromigration can cause whiskers. In the Big World it can't. But it can influence Big Whiskers to grow to the worst places and fuck up things optimally if you take an extremely relaxed view of electromigration that defines it as "movement of ions encouraged by an electric field".
You're misunderstanding me.
For instance, electrons always move the same speed in a given metal. Which of couse isn't even 'true' because temperature affects mobility.
There are multiple mechanisms for metal to migrate, grow whiskers, or whatever you like to call the individual growth on an object. I mentioned that in the case if ICs, we are concerned with one we call electromigration. I'm not saying all metal migration is due to electromigration.
You're being pedantic when all I'm saying is, I deal with these sorts of concerns in my job.
Tiger I think you're being pedantic, they linked to Whiskers (metallurgy) not Whiskers (electromigration). There is a difference! But it's not super clear cut, which is why I took the time to write about it.
Electrons do not always move at the same speed in a given metal. A lot of things affects mobility, but the E field is very important too. Both things combine so that electrons do not always move at the same speed in a given metal. But you can simplify in an IC world because there you're riding the saturation velocity basically always, which is why I assume you keep claiming that.
I want you to know that your experiences from your education and job are valid - you do deal with whiskers in ICs, not denying that; the fact is that whiskers due to stresses and strains aren't called electromigration which is what the original comment says.
"A similar thing also called whiskers can happen inside ICs and has been a known failure mode for high frequency processors for many years. I work in chip design, and we use software tools to simulate it. It’s due to electromigration and doesn't rely on stresses but instead high current densities."
Compliments for a excellent example of constructively having a discussion with some minor disagreements. Lemmy is better for your contributions!
Metal whiskering is a phenomenon that occurs in electrical devices when metals form long whisker-like projections over time.
That's what the article says. EM is a subset of metal whiskering. It's not a similar thing, it's an example of it.
Tiger, you’re very similar to many of the semiconductor EEs I know :) and I mean that in a teasing-but-you-know-cause-you-work-in-the-industry way. Yeah, we only really care about whiskering in the context of electrical devices. That’s what it’s saying. Read the “Mechanics” section, it tells you nothing about actual electromigration doing it; they describe an E field encouraging metal ions in a fluid to make a reaching whisker and link to electromigration because it technically is “electromigration” making the targeted whisker occur. But IC-style electromigration is not causing the whisker, clearly cause no currents are flowing, which is why I took the time to write the explanation in the first place.
But just because the semiconductor community called it whiskers so it shares the name with the Big Whiskers, does not make the process anywhere close to similar. The current densities that cause absolutely not present for the stress ones, which the wiki article is about.
Electromigration is not the same as metal whiskering.
Does electromigration move the dopant?
Electromigration in ICs typically occurs in metal interconnect, so there are no dopants there to move. Dopants are added to the silicon substrate.
I'll have to look this up cause now I have like three more questions. Thanks!
Check out https://en.wikipedia.org/wiki/Hot-carrier_injection hot carrier degredation, it's in the vicinity of electron mobility but in a semiconductor setting. Key link is it's electrons with momentum doing the work. In this case electrons (much hotter than in electron mobility, which are limited by the saturation velocity) smash into the gate dielectric, making it a worse dielectric. Hot carrier injection doesn't have to end in damage to the dielectric, but when it does it's hot carrier degredation. There's a lot going on though, semiconductors are really complex - like electron tunneling also exists.
This is like saying an SUV isn't a car.
OP: Look at my car
Me: I have one too. It's an SUV
You: SUV's aren't cars.
If you understand "car" as "hardware degradation" there is something to it, despite calling it "electromigration".
You said it (= whiskers) can be simulated and that it's called electromigration. From what I understood, this statement is wrong, since they are both different in both cause and effect. Metal whiskering can be simulated to a certain extent, yes. But that's vastly different to what electromigration is and how it works.
You got a degree in semi conductor physics? Lol
Like i told the other guy, you're being pedantic. Engineers will call all these things whiskers. And I just mentioned I deal with one of them in my work. I'm not saying the photo in the thumbnail is an example of electromigration.
You got a degree in semi conductor physics? Lol
Man, that "lol" really annoys me and comes accross condescending. If you've got no arguments, there is no need for an academic dick measuring contest. You can just leave it. To answer your question:
In parts, yes. Not my specialisation though, but enough to be able to distinguish electromigration from whiskering.being pedantic. Engineers will call all these things whiskers.
Being pedantic is part of the job of an engineer. I'm an engineer working in research. I don't call electromigration "metal whiskering" or vice versa.
Besides, as I've mentioned, it wouldn't even be pedantic to distinguish them that way as the differences are not miniscule. They are formed differently and look differently.I’m not saying the photo in the thumbnail is an example of electromigration.
Yes, to the post which is titled "TIL computers can sometimes grow crystals" you said:
This can happen inside ICs [...] It’s called electromigration.
Which is still wrong. We can observe electromigration in ICs, or in metallic conductors in general, but this is a different phenomenon than whiskering, which can look like those crystals while conductors affected by electromigration form voids and protrusions out of material build-ups which usually can't even be seen by the bare eye.
But maybe that was a misleading expression and you didn't mean to equate those two.You got a degree in semi conductor physics? Lol
Appeals to authority are not a good look
From where do the gentleman's statement gain their credibility. Site your sources, please. We are having an argument on the internet folks, the consequences are dire.
This is Saturday night and this is staring to remind me of Monday morning meetings.
Have an nice weekend.
Why is it an argument at all? Why can't it be just as a constructive discussion where people learn from each other, and have fun with new ideas