With flash seen as a doomed technology, always predicted to run out of steam at every new generation, then miraculously finding a way to the next generation, are any of the alternatives to the traditional floating gate flash process looking promising?
"We want to make products that are non-volatile and reprogrammable, and we don't care about the process," replies John East, CEO of Actel which relies on traditional floating gate flash technology to build its products, "we have 130nm, and are moving to 65nm for use in products in 2010. Below 40nm people are looking at what's going to happen."
He points out the somewhat alarming fact that: "At 32nm the gate oxide thickness is three or four molecules."
Does he see any future in phase-change memory? "My own view is they need to make it better", replies East, "they should get a high enough ratio between ON resistance and OFF resistance. If you don't get that super resistance you get leakage problems." He concludes: "The specs don't meet the requirements."
But though he's not putting his money on any of the current proposed alternatives to traditional floating gate flash processes, he remains convinced on one thing: "What's not going to happen is the world won't quit buying non-volatile memory because the world likes non-volatile memory."
I suppose there's always the chance of using a material with smaller atoms, like carbon, but no one has yet found a way of making ICs out of carbon.
Surely flash will go to high-k insulator materials to get away from the oxide thickness/tunneling problems in the same way as logic CMOS will?
Ian
It looks like the text from "With flash seen as ..." to "... likes non-volatile memory" is duplicated.
Thank you Soren, I have taken out the duplication
"I suppose there's always the chance of using a material with bigger atoms, like carbon, but no one has yet found a way of making ICs out of carbon."
Hi David - assuming you are quoting me from the IEF lunch then carbon is a small atom, not a large one. Indeed it's the smallest that exhibits any decent semiconducting action. And there are plenty of carbon based processes currently in the labs ready to attempt the move to production at sub 18nm when they become near essential. Solutions proposed include amorphous carbon, nanotubes, diamond and graphene, as well as some quite exotic carbon based molecules - organic and inorganic. Until then there are still several ways of keeping the barrier layer insulating.
Which carbon solution will win - no idea. But I'd vote for any above PCM :-)
Oh I see, Ian, I didn't realise that would work on a memory process. Don't know why I thought that. It does sound logical (pun unintended I promise)
Thanks Mike you'reabsolutely right I should have said smaller atoms, and you're absolutely right that bit came from what you said at the lunch after IEF.
There still seem to be dulplicate paragraphs.
"We want to..." and "He points out...".
Similar problem in another recent blog too - is David having an off-day ;-)
A few off-days, I'm afraid Steve, hopefully it now reads as intended.