How to model cosmic ray corruption of DRAM memory?

[Bob Day]

I'd like to study how well Chipkill advanced ECC detects
multiple errors in a DRAM memory module. I've written software
that implements the Chipkill algorithm, and now I'm seeking more
detail about exactly how cosmic rays can corrupt DRAM bits.
Such as, when a cosmic ray changes a bit in DRAM, how many
adjacent bits on the average are likely to be affected as well?
What would be the probability density of the number of bits altered
per cosmic ray strike? Of course, it's likely to depend on the density
of the DRAM. I'd much appreciate any information anyone can
provide on this.

--Bob Day
http://bobday.vze.com

[Eric Gisse]

On Jan 28, 2:00Â*pm, "Bob Day" <xxx...@yyyyyy.com> wrote:
> I'd like to study how well Chipkill advanced ECC detects
> multiple errors in a DRAM memory module. Â*I've written software
> that implements the Chipkill algorithm, and now I'm seeking more
> detail about exactly how cosmic rays can corrupt DRAM bits.
> Such as, when a cosmic ray changes a bit in DRAM, how many
> adjacent bits on the average are likely to be affected as well?
> What would be the probability density of the number of bits altered
> per cosmic ray strike? Â*Of course, it's likely to depend on the density
> of the DRAM. Â*I'd much appreciate any information anyone can
> provide on this.
>
> --Bob Dayhttp://bobday.vze.com

Consider how scintillator-based particle detectors work: A particle
comes in, impacts some of the scintillator, and sends off an electron
that is then detected through some means. Depending on the energy of
the particle there will be more electrons as the particle plows
through the scintillator.

A more useful response would be "Ask NASA" or "Ask Intel", as both of
them are deeply familiar with the details of how to sheild [and most
likely, model] how cosmic ray flux breaks a semiconducting chip.

[Joseph Warner]

"Bob Day" <xxxxxx@yyyyyy.com> wrote in message
news:nE2gl.403$N5.198@nwrddc01.gnilink.net...
> I'd like to study how well Chipkill advanced ECC detects
> multiple errors in a DRAM memory module. I've written software
> that implements the Chipkill algorithm, and now I'm seeking
> more
> detail about exactly how cosmic rays can corrupt DRAM bits.
> Such as, when a cosmic ray changes a bit in DRAM, how many
> adjacent bits on the average are likely to be affected as well?
> What would be the probability density of the number of bits
> altered
> per cosmic ray strike? Of course, it's likely to depend on the
> density
> of the DRAM. I'd much appreciate any information anyone can
> provide on this.

Most of the corruption of the DRAM comes from secondary
electrons that are generated. The scanning electron microscope
people has been using Monte-Carlo routines to estimate the number
of secondary electrons generated versus electron impact energy
and the deep. This analyzes gives what looks like a water droplet
of density of generation with the least number of secondaries
generated at the surface and the most well below the surface (1/2
to 2 microns). The number and the energy of these electrons
depends on the material being bombarded. So look over in that
field will allow you to estimate the number of secondaries
generated. This analysis can be used for any charged particle.

Another event that may occur is that the cosmic particle may
directly hit the nucleus of an atom and displace it. If that
occurs that is permanent damage that will generate color centers,
local energy levels in the bandgap of the material ... This
damage usually determines how "radiation hard" the material is.
As it turns out as the thickness of the transistor becomes
thinner the "harder" the circuit is. The former is related to the
rate of single events upsets (SEU) and can be recovered from.

To do an good job of estimating the rate and range of SEU one
would begin with the secondary electrons generated. After that
one needs to know the details of the memory unit, i.e. the
voltages, the thickness of the material layers, the type of
material, the geometry, the doping levels ...

One institution that has a very good program that looks at
radiation effects on electrical circuits is Georgia-Tech
University in Atlanta, Ga. Look up articles by John Cressler

Here is a link to dissertation that you may find useful
http://smartech.library.gatech.edu/handle/1853/6952