6 ms·

Differences in Semiconductor Soft Errors Caused by Cosmic-Ray Muons and Neutrons

56 points by mkj 5y ago

tooltower 5y ago
From the article, what's a "thermal neutron beam"?
- TrumpsHandler 5y ago
  Probably neutrons that are slowed enough/have lost sufficient energy to break atoms or be absorbed into an atom.
- baq 5y ago
  https://en.wikipedia.org/wiki/Neutron_temperature https://en.wikipedia.org/wiki/Neutron_temperature 0.025 eV Thermal neutrons
- caf 5y ago
  It's neutrons with kinetic energy that's similar to the other particles in the environment, so they're at thermal equilibrium. Read it as "low energy neutrons".
- adrian_b 5y ago
  Thermal neutrons are slow neutrons, which have a velocity and kinetic energy similar to those of atoms that are in thermal equilibrium at the ambient temperatures. When the neutrons are generated in nuclear reactions, they have a very high kinetic energy, but after many collisions with the matter encountered in their path, they lose most of the energy. After they have become slow enough, so that the likelihoods of losing or gaining energy in collisions become similar, they are considered thermal neutrons. So a "young" neutron is a fast neutron, after some time it becomes a thermal neutron and eventually it decays into a hydrogen nucleus (proton).
- tobmlt 5y ago
  Since others have answered, for contrast high energy is sometimes denoted “ultraviolet” also high energy “hard” vs low energy “soft” loosely speaking
- Sharlin 5y ago
  Thermalizing neutrons is also what the moderator (typically water but can also be eg. graphite, like in the infamous RBMK reactor design) surrounding the fuel rods in a nuclear reactor is for. Neutrons fresh from a fission reaction are too fast to reliably induce additional fissions before escaping and being absorbed by the reactor walls.
willis936 5y ago
For some context on the company sharing this: Socionext is the name of the 2015 merger between the semiconductor companies of Fujitsu and Panasonic. I worked for a test house at the time and regularly tested Fujitsu's stuff for them. I remember being surprised that two big name companies would just scrap all of that brand recognition for a cheap sounding name.
- burnte 5y ago
  While I'd have loved to see Fujisonic or Panajitsu, they were probably prohibited from using either of the parent company names for the joint venture.
- jhgb 5y ago
  > I remember being surprised that two big name companies would just scrap all of that brand recognition for a cheap sounding name. That reminds me of Kioxia.
  - jsmith45 5y ago
    Yeah, for some reason Kioxia just screams "random Chinese company" to me. Without learning who they are from advertising, I would have lumped them them in no name Chinese components. Admittedly I don't do hardware product development involving flash memory chips. If I did I probably would have heard of this more quickly, so in that industry the rename probably did not hurt as much. But they also have a retail component with their SSD drives. Here the rename is really unfortunate. Either the old "Toshiba Memory" name, or keeping the OCZ name would have been better there.
- arbitrage 5y ago
  I suspect that, being such large name companies, they are aware of the power of name recognition and chose deliberately to stay away from it -- perhaps because it would not be a help to the new merger. Rather, it might be a hindrance to the type of business that merger is trying to do.
a3n 5y ago
Tangent: Hold on a second ... Is my brain subject to this kind of thing, and is it making similar errors? Or is my thought mechanics on a single "thought" so multi-streamed that these errors are negligible?
- chasil 5y ago
  As I understand it, the surface of the earth is commonly bombarded by muons (and lots of them). Surface muons are used for imaging the Egyptian pyramids, among other things: https://www.nature.com/articles/d41586-018-05254-2 https://www.nature.com/articles/d41586-018-05254-2 Neutrons are much more dangerous, as they will readily be absorbed by atomic nuclei when moving slowly enough, and can form unstable isotopes that might decay by alpha/beta/gamma radiation. Neutron radiation must be at much lower levels than surface muon bombardment.
- xyzzy21 5y ago
  Probably. Given that nanometer scale CMOS is and thermal neutrons trigger the primary failure mechanism, BTI. The brain operates with amazing error rates and low bandwidth because it can "refresh reality data" from expected results coded over a lifetime. Most visual illusions, logical fallacy, etc. come from the heuristic of this process. We are NOT that smart. And part of the trick is we can convince ourselves that we nominally are despite that fact!
- Paedor 5y ago
  A neuron in the brain is subject to orders of magnitude more noise, from chaotic factors like blood oxygenation, signaling molecules, aging... For our actual conscious experience though, there's some evidence that we're very noise tolerant. I can't find the paper now, but one recently just came out showing that at certain parts of the brain you won't even notice an electric shock.
  - BoxOfRain 5y ago
    >For our actual conscious experience though, there's some evidence that we're very noise tolerant. I have a disorder that's very relevant to this, it's called visual snow syndrome. It's poorly understood but the current theory is that the brain has a fair amount of internal noise which is suppressed by a filtering mechanism but in some people this filtering mechanism fails, making the brain's internal noise becomes consciously perceptible. It looks like a badly tuned analogue TV for the most part with lots of static (hence the disorder's name), although other things like haloes around objects and geometric patterns appearing in textures like bricks are common too. Less fun is the permanent headache, sensory overload issues, and often a whole host of other neurological issues. It's one of those unfortunate cases where medical professionals palmed it off for decades as patients making it up for attention. I was personally told as a teenager to "try and imagine it away" by a neurologist which of course was utterly useless advice, and I was also given antidepressants at one point which made it infinitely worse, wouldn't wish that experience on my worst enemy. Fortunately I ended up in the care of a neurologist who finally recognised what it was and gave me a formal diagnosis after doing some scans to make sure it wasn't a brain tumour or something equally unpleasant. It turns out that although nothing shows up on a normal MRI scan, if you do an fMRI scan of brains like mine you can actually see the differences that cause the brain's noise to leak into your consciousness. Weird stuff.

56 points by mkj 5y ago

tooltower 5y ago

From the article, what's a "thermal neutron beam"?

TrumpsHandler 5y ago
Probably neutrons that are slowed enough/have lost sufficient energy to break atoms or be absorbed into an atom.
baq 5y ago
https://en.wikipedia.org/wiki/Neutron_temperature https://en.wikipedia.org/wiki/Neutron_temperature 0.025 eV Thermal neutrons
caf 5y ago
It's neutrons with kinetic energy that's similar to the other particles in the environment, so they're at thermal equilibrium. Read it as "low energy neutrons".
adrian_b 5y ago
Thermal neutrons are slow neutrons, which have a velocity and kinetic energy similar to those of atoms that are in thermal equilibrium at the ambient temperatures. When the neutrons are generated in nuclear reactions, they have a very high kinetic energy, but after many collisions with the matter encountered in their path, they lose most of the energy. After they have become slow enough, so that the likelihoods of losing or gaining energy in collisions become similar, they are considered thermal neutrons. So a "young" neutron is a fast neutron, after some time it becomes a thermal neutron and eventually it decays into a hydrogen nucleus (proton).
tobmlt 5y ago
Since others have answered, for contrast high energy is sometimes denoted “ultraviolet” also high energy “hard” vs low energy “soft” loosely speaking
Sharlin 5y ago
Thermalizing neutrons is also what the moderator (typically water but can also be eg. graphite, like in the infamous RBMK reactor design) surrounding the fuel rods in a nuclear reactor is for. Neutrons fresh from a fission reaction are too fast to reliably induce additional fissions before escaping and being absorbed by the reactor walls.

willis936 5y ago

For some context on the company sharing this: Socionext is the name of the 2015 merger between the semiconductor companies of Fujitsu and Panasonic. I worked for a test house at the time and regularly tested Fujitsu's stuff for them. I remember being surprised that two big name companies would just scrap all of that brand recognition for a cheap sounding name.

burnte 5y ago
While I'd have loved to see Fujisonic or Panajitsu, they were probably prohibited from using either of the parent company names for the joint venture.
jhgb 5y ago
> I remember being surprised that two big name companies would just scrap all of that brand recognition for a cheap sounding name. That reminds me of Kioxia.
- jsmith45 5y ago
  Yeah, for some reason Kioxia just screams "random Chinese company" to me. Without learning who they are from advertising, I would have lumped them them in no name Chinese components. Admittedly I don't do hardware product development involving flash memory chips. If I did I probably would have heard of this more quickly, so in that industry the rename probably did not hurt as much. But they also have a retail component with their SSD drives. Here the rename is really unfortunate. Either the old "Toshiba Memory" name, or keeping the OCZ name would have been better there.
arbitrage 5y ago
I suspect that, being such large name companies, they are aware of the power of name recognition and chose deliberately to stay away from it -- perhaps because it would not be a help to the new merger. Rather, it might be a hindrance to the type of business that merger is trying to do.

a3n 5y ago

Tangent: Hold on a second ... Is my brain subject to this kind of thing, and is it making similar errors? Or is my thought mechanics on a single "thought" so multi-streamed that these errors are negligible?

chasil 5y ago
As I understand it, the surface of the earth is commonly bombarded by muons (and lots of them). Surface muons are used for imaging the Egyptian pyramids, among other things: https://www.nature.com/articles/d41586-018-05254-2 https://www.nature.com/articles/d41586-018-05254-2 Neutrons are much more dangerous, as they will readily be absorbed by atomic nuclei when moving slowly enough, and can form unstable isotopes that might decay by alpha/beta/gamma radiation. Neutron radiation must be at much lower levels than surface muon bombardment.
xyzzy21 5y ago
Probably. Given that nanometer scale CMOS is and thermal neutrons trigger the primary failure mechanism, BTI. The brain operates with amazing error rates and low bandwidth because it can "refresh reality data" from expected results coded over a lifetime. Most visual illusions, logical fallacy, etc. come from the heuristic of this process. We are NOT that smart. And part of the trick is we can convince ourselves that we nominally are despite that fact!
Paedor 5y ago
A neuron in the brain is subject to orders of magnitude more noise, from chaotic factors like blood oxygenation, signaling molecules, aging... For our actual conscious experience though, there's some evidence that we're very noise tolerant. I can't find the paper now, but one recently just came out showing that at certain parts of the brain you won't even notice an electric shock.
- BoxOfRain 5y ago
  >For our actual conscious experience though, there's some evidence that we're very noise tolerant. I have a disorder that's very relevant to this, it's called visual snow syndrome. It's poorly understood but the current theory is that the brain has a fair amount of internal noise which is suppressed by a filtering mechanism but in some people this filtering mechanism fails, making the brain's internal noise becomes consciously perceptible. It looks like a badly tuned analogue TV for the most part with lots of static (hence the disorder's name), although other things like haloes around objects and geometric patterns appearing in textures like bricks are common too. Less fun is the permanent headache, sensory overload issues, and often a whole host of other neurological issues. It's one of those unfortunate cases where medical professionals palmed it off for decades as patients making it up for attention. I was personally told as a teenager to "try and imagine it away" by a neurologist which of course was utterly useless advice, and I was also given antidepressants at one point which made it infinitely worse, wouldn't wish that experience on my worst enemy. Fortunately I ended up in the care of a neurologist who finally recognised what it was and gave me a formal diagnosis after doing some scans to make sure it wasn't a brain tumour or something equally unpleasant. It turns out that although nothing shows up on a normal MRI scan, if you do an fMRI scan of brains like mine you can actually see the differences that cause the brain's noise to leak into your consciousness. Weird stuff.