Wily I may certainly be wrong, but I think that 2500 wafers per month is all the FRAM production in the entire known universe.
That number seems about right (give or take a lot) for the present revenue levels (assuming 5" wafers), but the article is referring to production on the .35um process which I don't believe is ramped yet—I assume we will hear something definite from Ramtron when the .35um line goes into full production.
Which is why I think you are wrong about The Holy Grail Patent. I think TI is looking to have a test chip or samples at the end of this year with sample production in '03. There have been lots of "hints" at what's going on:
"Our program with TI provides Ramtron with a path to the Flash memory market - the largest nonvolatile memory market in the world," Staunton added. "Ramtron expects the TI program to yield an advanced FRAM memory cell and architecture that will be capable of competing with and outperforming Flash memory technology.
Ramtron's goal is to co-develop a production worthy, 0.13-micron FRAM process that can establish FRAM as the ideal nonvolatile memory solution on a density, feature and cost for benefit basis."
ramtron.com
OK, let's just assume that TI's FRAM is based on your holy grail patent. What does this mean? They have already said the cell size will be 1/4 the size of SRAM. This puts it in the 30-40F^2 range, or about half the size of the present devices coming out of Fujitsu. Such a reduction in size is probably not out of line with what is possible by adding a few levels of metal—a progressive step anticipated by Fujitsu in one of the articles you linked. There's also no claim, indication, or even any reason to think that the resistance-reading FRAM design is going to be smaller than the current designs or easier to manufacture.
It would also mean non-destructive reads. Don't you think this would be mentioned in some of the fluff literature? Or will this be a once-in-a-blue-moon event where a major milestone in tech progress is sprung upon the world totally unannounced beforehand?
Aside from the above objections, it's contrary to experience to think that a new, novel structure patent represents a state of technology ready for the real world. All kinds of functional and manufacturing issues have to be explored and resolved in the lab/fab, and you're probably going to see a phalanx of elaborations, modifications and revisions, taking several years, before the thing is product-ready—or consigned to the scrap heap somewhere along the way.
I guess my main point is that you don't NEED to assume that this patent is the basis for what TI is doing. They could be saying the same things based merely on a plan for adding extra metal layers—same thing that Fujitsu is doing.
It's certainly encouraging that they hope to have samples soon and that they are pursuing the leading edge processes.
Tyler Lowrey accuses MRAM of being hype, and admits that OUM is still several years out and not even a sure thing. FRAM is density-challenged but realer than the others. |
