Differences in Semiconductor Soft Errors Caused by Cosmic-Ray Muons and Neutrons | Hacker News Reader

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

socionextus.com

56 pointsmkj21 comments

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.

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.

> 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.

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.

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.

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?

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

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.

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!

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.

>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.

>geometric patterns appearing in textures

Is it similar to the kinds of visuals a person using psychedelics might experience?

Yes, in fact visual snow has quite an overlap in symptoms with HPPD experienced by some chronic users of psychedelics! It was even sometimes misdiagnosed as HPPD in the past, although that theory was discounted as the vast majority of visual snow patients don't have a history of psychedelic use (I was far too much of a nerd to be dropping acid aged 14/15 when I contracted the disorder!). For me the appearance of these visuals is triggered by closely repeating patterns such as bricks, tiles, striped or chequered shirts, chessboards, that sort of thing. It makes reading a bit difficult sometimes too, I really struggle with dense typography like newsprint for example so I usually read on a Kindle with the line spacing and font size set quite high.

I definitely think psychedelics have a role to play in understanding these dysfunctions of the visual system, although unfortunately my country (the UK) has a very unenlightened view when it comes to banning things and firing chief scientific advisors when they point out that there's no real scientific justification for prohibitionism. Vernon Dursley is apparently the target voter for both major parties!

Yes - phosphenes can be caused by cosmic rays and other sufficiently energetic sources. Astronauts see such sparkles when they're not shielded. And such cosmic ray interactions are a mutagenic factor in changes to DNA, resulting in cancer and evolution.

Visit some museeum, exhibition, laboratory, hack space with a https://en.wikipedia.org/wiki/Cloud_chamber and make up your own mind? :-)

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

Probably neutrons that are slowed enough/have lost sufficient energy to break atoms or be absorbed into an atom.

https://en.wikipedia.org/wiki/Neutron_temperature

    0.025 eV  Thermal neutrons

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".

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).

Since others have answered, for contrast high energy is sometimes denoted “ultraviolet” also high energy “hard” vs low energy “soft” loosely speaking

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.

Reply on news.ycombinator.com