3

Answers to Why is the James Webb Space Telescope's data storage space so small? have got me thinking.

Solid State Hard Drives (SSDs) are usually made from FLASH memory though alternatives like magnetic memory also have their place in space.

The unit cell of FLASH memory is a very well isolated capacitor on which charge is introduced by tunneling at high(ish) voltage.

The charge can stay there for years or centuries potentially, we rely upon this vanishingly-low discharge rate when we put something important on an SSD or USB stick and expect it to stay there indefinitely. We sometimes even casually toss flash drives from spacecraft and assume they'll get to Earth with data intact!

enter image description here click for larger

Symbolic jettisoning of the video capsule containing messages from last year’s Victory Day celebration. Photo: NASA TV

Flash memory gets smaller and denser by making the devices smaller, and as a consequence, the oxide gaps and voltages and amount of charge on each capacitor get smaller as well.

It gets even smaller by going analog sort-of, by storing eight different levels of charge on one capacitor, so that the difference between one value and the next is a small fraction of the total charge.

My concern:

Ionizing radiation in a space environment will pass through electronic devices and leave trails of charge; electrons separated from their atoms. My concern is that this process can eventually change the amount level in a FLASH memory bit's capacitor enough to change the logical value it is assigned upon readout at some later date.

As SSDs are built larger in storage and smaller in size, stored data integrity may become more susceptible to ionizing radiation in space.

Question: Are Solid State Hard Drives in space made from radiation-hardened FLASH memory? Or do they just use older technology with higher voltage and charge rating that is less susceptible to ionizing radiation? Or is something else done, or is this not actually a problem at all?

uhoh
  • 148,791
  • 53
  • 476
  • 1,473
  • 3
    It would be possible to use FLASH memory chips with the 8 level technology and use only 1 bit for data and the other 2 bits for an error correcting code. Just a different firmware for the solid state drives. The reliability may be even better than possible with one level technology. But less bits for ECC will do it too if multiple error correction is possible. ECC is used for hard drives anyway see, so a space version should just use more bits for ECC than the ground version. – Uwe Feb 01 '21 at 10:13
  • 1
    There are so many aspects of the influence of radiation to semiconductors - the active change of charge in a capacitor due to ionization is one of the least issues. There's no way to describe all the aspects in an answer to this question. just to list a few buzz words: Bulk damage, doping changes, parasitic diodes, trapped charges... – asdfex Feb 01 '21 at 13:07
  • 1
    Devices not built to meet rad hard requirements can (and do) have highly varied responses to dose. One small process change at the fab can completely alter their sensitivity. Aside from student cube sats, anybody relying on the flash actually working will use a qualified rad hard chip. – Jon Custer Feb 01 '21 at 15:42
  • @asdfex you've mischaracterized the question, take another look. It can be answered with a boolean; are they, or aren't they? Your question would be off-topic here which is why I didn't ask that. You can try it in Electronics SE though. – uhoh Feb 01 '21 at 18:39
  • 2
    My comment was meant to say that there are so many factors preventing the use of standard flash cells, it just doesn't matter if charge loss due to introduced charges plays a role or not. – asdfex Feb 01 '21 at 19:10
  • @asdfex yep, I have a hunch that bit flips in FLASH will happen relatively frequently considering the lifetime of an important file stored on SSD, compare to catastrophic failures to do radiation damage in the bulk. Maybe though, redundancy and error correction routines in modern hard drive implementations/architectures make SSD so tolerant to these that it's a non-issue. That may in fact be the answer to the question. – uhoh Feb 01 '21 at 19:25
  • 1
    Another possibility here is using redundant storage like RAID 1, where bits can be constantly checked for “bit flips” and fixed within seconds. – CourageousPotato Feb 04 '21 at 17:21
  • 1
    Wind's last two magnetic tape drives are still working just fine after 26+ years. Voyager's failed a while ago (I think this is so but regardless, they became pointless due to telemetry limitations) so they are in a transmit-only mode. I know Parker Solar Probe has a few solid state recorders and know they have redundancy encoding, but I am not sure what type they happen to be. Yes, they have to be radiation-hardened and have redundancy checks. Van Allen Probes had to have extra requirements on this front for obvious reasons. – honeste_vivere Feb 24 '21 at 15:49
  • Partial answer here - How does the Perseverance Rover store/process large quanties of images? https://space.stackexchange.com/a/50291/9092 – Kozuch Mar 26 '21 at 14:04
  • @Kozuch I disagree; I don't see what part of my question here is actually answered there at all, and in fact I've linked to this question under the answer there already. I don't see any mention of rad-hard vs non-rad-hard there. – uhoh Mar 26 '21 at 14:42
  • 1
    @uhoh Yes you are right, the answer mentions COTS computer but not the SSD. I did not read the original article though. – Kozuch Mar 26 '21 at 14:48

0 Answers0