Weebit - 2023 and beyond, page-787

Hi guys

A while back I posted some musings on the possible 'secret sauce' that WBT seem to be using in the structure of the ReRAM cell. I got quite a few likes, but on reflection I think I set you all crook. I've now done virtually a 180 in my thinking. So before I get into it, here is a snip of the that earlier post (check out the SP!):



I believe now, that either Coby misspoke, or much more likely, I misheard (I do have a bad habit of trying to do 2 things at once and so often listen to these interviews while sitting at my computer doing other things). I've tried looking for these interviews (there were at least two that I remember) but can't find them again. If anyone else can, I would be grateful if you could post the link.

Anyway, so what I believe he said (or meant) was not metal or metallic filament, but metalloid filament. For a definition of metalloid you can look here: https://www.chemistrylearner.com/metalloids

Thus I believe Coby was simply referring to the silicon filament created by oxygen depletion of the SiO_x (I notice now also that I used the symbol for sulphur in my previous post - d'oh!).

Regarding what I wrote about the possible migration of metal atoms into the filament region during the forming process. I now think that if this actually occurred the metal atoms would be a serious contaminant. I doubt it would be possible to make the process consistent across all cells. As I said in my original post, "I little would go a long way", so any variation in the small number of metal atoms in the filament region would cause large variation in filament conductivity. In this technology the name of the game is repeatability, and anything that interferes with that is bad (in a bad way).

So, what of this possibility that metal atoms could migrate into the filament region. Well, maybe a way of guarding against it could be to introduce a very thin layer (just 2 or 3 molecules thick), in between the metal layer and the SiOx (active) layer, of an oxide of the same metal that is used for the metal layer. Such an oxide layer is what is described in the patent posted by @Flectional, so perhaps this is the purpose of it - to act as a barrier between the metal layer, and the active layer, to prevent metal atom migration into the filament (either at forming, or over time during write cycles).

Another good reason for having such a barrier layer could be that direct contact of the metal with the SiO_x may cause oxygen depletion of the SiO_x at the interface, due to metal atoms 'stealing' oxygen atoms from the SiO_x. This could occur during metal deposition (temperatures are high). After etching of the metal layer (to leave the pattern of conducting tracks), if any of this oxygen depleted layer remains, it would be a conductive path between the tracks, thus shorting out adjacent rows of cells.

Either of these possibilities would only need to affect a tiny fraction of cells to be a really big problem.

Well, after all that I could still be way wide of the mark. I hope I didn't lose too many of you along the way. My thanks, congrats and sympathies to all who made it this far. The take-home I think is that this is mind-boggling technology, being developed at atomic-scales right at the edge of what is possible. When I think of the 8 bit microprocessors and 7 segment LCDs I used to work with in a former life, the industry has come a very long way. It left me behind decades ago.