Info about cosmic rays affecting microchips

[f95toli]

Sorry for the vague title; I couldn't think of a better one.

I have been looking for information about how frequently one could expect cosmic rays (of all types) to hit and be absorbed by a piece of silicone (say 5x5x0.5 mm3) at the surface of the Earth. But haven't had much success.

I know this is something computer chip manufacturers worry about since such events can e.g. cause errors in microprocessors so I thought this would be easy to find; but there is very little actual information out there (and quite a bit seems be classified).

I'm interested because I'm involved in experiments where we look at very weak signals in quantum circuits. Our samples are sitting at the bottom of a dilution fridge at 10mK and our measurements are sensitive enough to worry about individual IR photons. I know -anecdotally- that people who work at even lower temperatures (experiments on helium at microkelvin temperatures) occasionally (daily) see events they say is probably cosmic rays (these events literally heat their samples) so I've started to wonder about how frequent these event really are and if we might be seeing them as "noise" in our experiments.

Does anyone have a good reference or even some good keywords I can use in Google scholar?

 

[mfb]

What do you mean by "absorbed"?
If it is horizontal, it will be hit by one muon every few minutes, and once in a while by other particles. Most of these muons will go through the detector and leave a trail of ionizations behind. "Cosmic ray flux" should lead to various useful papers.

If you are sensitive to these events, put another detector as muon veto above it or go underground.

 

[phyzguy]

As mfb said, muons are the most common events. In general, computer chips are not sensitive to cosmic ray muons, because they transfer too little charge. Most errors in computer chips come from either alpha particles emitted from the materials in which the chips are packaged, or from cosmic ray neutrons. These events are quite rare, but in large computer systems the errors are common enough to require error correcting schemes. For estimates of the computer chip upset rates, try googling either "soft error rate" or "single event upset".

 

[f95toli]

What do you mean by "absorbed"?.

Basically some interaction with the lattice that causes local heating. Or, alternatively some interaction that temporarily creates free electrons.
A third possibility is the an interaction that actually breaks Cooper pairs in the superconducting film on top of the device. However, I would assume(?) that this is less likely since the film is only ~100nm thick.

The samples I am working on are not meant as detectors and are -in theory- sitting in a well shielded (from EM) environment. However, the technology we use (thin film superconducting devices) is more or less the same as is used for detectors in e.g. radio astronomy (e.g. kinetic inductance detectors, KIDs or TES). There are papers on the effect on CR on KIDs (e.g. https://arxiv.org/ftp/arxiv/papers/1505/1505.01647.pdf) but these are all considering space-based missions where the CR flux would be much higher.

What would be a "detection event" for a purpose-made detector would in our case possibly look like low frequency noise; which is why I am trying to understand how frequent these events really are at sea level. It is -as far as I am aware- not a source of "noise" anyone else has considered for the type of work I do; but the techniques we are using in our measurement s are much more sensitive than what you would find in an "average" lab (low frequency noise measurements is something of a specialty)

 

[f95toli]

As mfb said, muons are the most common events. In general, computer chips are not sensitive to cosmic ray muons, because they transfer too little charge. Most errors in computer chips come from either alpha particles emitted from the materials in which the chips are packaged, or from cosmic ray neutrons. These events are quite rare, but in large computer systems the errors are common enough to require error correcting schemes. For estimates of the computer chip upset rates, try googling either "soft error rate" or "single event upset".

Thanks. Those search terms gave me lots of useful hits