I believe there are two challenges to be faced: a. Effective and efficient FDSOI wafer supply. 160K/month would be not enough in production phase.
b.$500 per wafer is too expensive to be adopted. For production, $250-300 would be affordable target.
Article mentions "... already produced tape-outs of a dozen system-on-chip devices for mobile, ..."
any idea which mobile SoC is referred to here? Is it same NovaThor SoC?
FDSOI helps in multiple power saving techniques like dynamic body biasing, on chip regulator, sub-clock power gating etc., how these feature compares to HKMG process technology?
thank you Adele for a clear and comprehensive explanation...BTW, would you have any interested in editing a small book on SOI? (I have been recently approached by one of the publishers I work with)...Kris
Hi Kris -- yes, it's been a confusing issue for everyone! Basically the wafer mfgs quote top Si thickness of the wafers they ship, while the chipmakers quote the Si thickness once they're done processing the device -- which is always less.
Bruce Doris of IBM explained it nicely in an ASN article from 2010 http://www.advancedsubstratenews.com/2010/07/etsoi-substrates-what-we-need/: "The top silicon of the starting wafer (currently 12nm) has to be thicker than the final target thickness of the channel since some Si is used up in the process flow prior to final channel thickness definition. To reach full industrialization, we need a high-volume supply of wafers with thickness variation of less than +/-0.5nm with corresponding wafer-to-wafer uniformity." So now Soitec's shipping wafers for FD-SOI with Angstrom-level uniformity, meeting all the requirements. SEH (the world's biggest wafer mfg), recently told SemiMD that they are, too (http://semimd.com/blog/tag/seh/ )
BTW, back in 2010, IBM was getting post-processed Si thickness of 3.5nm, which could meet the requirements of the 11nm node, so you can imagine they've made all kinds of progress since then!
And as Soitec's pointed out on their website: "Uniformity of the top silicon layer of Soitec FD-2D wafers [[their wafers for FD-SOI]] is guaranteed to within +/-5Å at all points on all wafers. This uniformity is equivalent to 5 mm over 3,000 km, which corresponds to approximately 0.2 inches over the distance between Chicago and San Francisco. For planar fully depleted SOI (FDSOI) CMOS technology, uniform thickness is crucial to controlling transistor Vt variations." Their current product spec sheet http://www.soitec.com/pdf/Soitec-FD-2D-product_brief.pdf indicates top Si of the FD-SOI wafer (this is pre-processed, as indicated above) down to 10nm.
Hope that helps!
ST is doing FD-SOI -- make no mistake about it -- it's a game changer. See today's interview with ST's JM Chery http://www.advancedsubstratenews.com/2012/10/exclusive-asn-interview-sts-jean-marc-chery-on-fd-soi-manufacturing/ He said: " ST/Crolles fab is now working to bring yield at production levels and complete the qualification of the technology, as done for any other. Phase-in of the technology at GlobalFoundries is planned to start Q1 2013, with process qualified and with production level yield foreseen for Q4 2013."
Interestingly, he also says they're moving right to 14nm with it.
@Michigan -- Soitec has no trouble with the wafers. SEH is also a supplier. Don't confuse the starting thickness with the finished thickness. Starting wafers have slightly higher top Si thicknesses, and some is etched away during manufacturing, hence what looks like a discrepancy is not.
Also, re: Samsung, Chery recently told David Manners that " And we can use Samsung for SOI if we need to". (see http://www.electronicsweekly.com/Articles/04/09/2012/54458/st-a-competitive-follower-in-process.htm)
I think ST will remain committed to analog, mixed-siganl and MEMS manufacturing. But it could be about to announce the withdrawal from digital IC production?
As you point out ST has spoken of several dropped days of production at Crolles and Catania in Q4.
And then disclosed that an updated strategic plan for the company is due to be presented in December.
It will be interesting to see what that contains.
Blog Doing Math in FPGAs Tom Burke 14 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...