Jan 28

3D Xpoint, the other shoe drops (maybe)

Duck shoe

Duck Shoe….

At last week’s Semi conference in Half Moon Bay in California, Guy Blalock Co-CEO of IM Flash Technologies (the Intel/Micron memory joint-venture, Blalock is the Micron half of the CEO duo), gave what appears to be, frankly, a rather extraordinary talk. He described a few more details on the memory materials and technology embedded in 3D Xpoint (announced to much fanfare last July) along with the manufacturing challenges.

I wasn’t at the talk but a (relatively) sober account of it can be found here. A rather more (provocative?) interesting take on it can be found here. (I really liked the picture and have included my homage here).

The bottom line is that 3D Xpoint is chalcogenide based both in the memory cell and (probably) switch. Of course, in the July announcement, The Intel/Micron team strongly implied the technology was not phase change memory (PCM) but that was then and this is now…. And maybe it isn’t phase change in the sense of the products that Micron briefly manufactured based on the technology and materials developed by Ovonyx/Numonyx. But it certainly seems similar.

Plenty of commentators are now in ‘told you so’ mode and I suppose I could make a claim to join them. (For me the most convincing evidence was the patent trail.) While one should never underestimate the sheer bloody mindedness and creativeness of engineers and scientists trying to develop a new technology, PCM has/had some real challenges in terms of scaling, materials, materials processing and endurance. I remember dreading that the project paying my wages at the time was going to be redirected in a PCM direction some time ago!

The rest of the talk seems to be the sort of presentation that you would make if you wanted to try and persuade your boss that undertaking a given project was a really, really bad idea. In no particular order, Guy pointed out that that 3D Xpoint would require:

  • 3x to 5x the usual node to node capital expenditure to get started in the fab
  • Similarly large transition costs for subsequent generations
  • A large increase in fab space required
  • 2-3x productivity gains (up from traditional 20-30%) in wet processes and dry etch
  • Process complexity resulting in a 15% reduction in fab throughput

Image: IM Flash

Image: IM Flash

At this point I would have expected my job would have been done and the bosses would have told me to go back to what I had been doing. But just in case, I would have a couple of slides in reserve making these additional points that Dr Blalock made in his Semi talk. 3D Xpoint would also require:

  • 100’s of new materials in the manufacturing process
  • Some of the materials only have one supplier located in one geographical location

100’s of new materials is an extraordinary challenge in itself (and unprecedented, IMO). Even only some of them end up in the actual product, they all have to be introduced into the Fab and evaluated meticulously for potential contamination and compatibility issues with existing processes and technology.

In any case, all ye naysayers, Sampling is “right around the corner… give the R&D guys a little more time to work out the kinks”….
Good luck to all you guys and gals!

Christie Marrian, ReRAM Forum, Director

1 comment

  1. Andy Walker

    Great analysis Christie!

    – Andy

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>