News & Analysis
Comment
peter.clarke
hm
450mm Fab may be way to go for Europe. It will offer many future avenues to ...
EuroFab450 is an ambitious dream
Peter Clarke
6/6/2012 6:16 PM EDT
Without a will there is no way
I don't think common manufacturing platforms across all parties are necessary for a jointly owned and operated 450-mm wafer fab. A 450-mm More-than-Moore fab in particular would have to support a large number of processes but there might be advantages in finding common ground and common processes. The bigger obstacle is making joint ownership work. This has usually prooved problematic in the long term. There is also a lack of the risk-tasking spirit that existed at the birth of the European experiment and which drove such things as GSM cellular mobile phone technology and the Megaproject and JESSI.
I am sure the EU/EC is aware of a need to reinvigorate European manufacturing and value creation in areas like nanotechnology and the need to do something to prevent manufacturing continuing to drain east. The European companies tend to just want R&D support for their nearer term technologies on 300-mm wafers. I don't see Carlo Bozotti of STMicroelectronics, Rick Clemmer of NXP Semiconductors or the incoming Reinhard Ploss at Infineon Technologies changing that position any time soon.
And with regard to Penn's idea of a 450-mm fab to given European companies an advantage behind the leading-edge there are yet more hurdles to be overcome.
More-than-Moore technologies for the most part do not require 450-mm diameter wafers and in some cases are positively uneconomic on such large wafers. It is not practical that an entire chip product's lifetime requirement can be produced on just a few wafers. Indeed there is discussion in the MEMS community that 300-mm wafers are still way too large for any designs and that 200-mm wafers are only just becoming the sweet spot for consumer MEMS with applications that need high volumes.
We should remember that Moore's law when originally coined was an economic argument and not a technical one. The only way I can see 450-mm wafers being used for the smaller volumes and wafer area required by many behind-the-leading-edge technologies would be if there was a business model change so that multiproject wafer runs could become more mainstream, rather than a prototyping vehicle.
That could happen (see Adapteva close to sampling 28-nm, 64-core coprocessor and InvenSense opens up process to enable fabless MEMS) but again is probably not a sufficient condition. There are arguments regarding amortization of investments which mean that the newest, most expensively produced wafers should be used for the largest die with the highest value per silicon area. Older written-down fabs running smaller wafers have always been used to run older processes and chips with lower selling prices.
The best hope I see is that the bureaucrats in Brussels can call in some favors with the likes of Intel and Globalfoundries, who want to be good European citizens, to have a leading-edge volume 450-mm wafer fab be located in Dublin or Dresden.
The full report can be downloaded from the European Commission's website at http://cordis.europa.eu/fp7/ict/nanoelectronics/documents/450mm-final-report.pdf
Related links and articles:
www.futurehorizons.com
News articles:
European report considers 450-mm More-than-Moore fab
InvenSense opens up process to enable fabless MEMS
Adapteva close to sampling 28-nm, 64-core coprocessor
I don't think common manufacturing platforms across all parties are necessary for a jointly owned and operated 450-mm wafer fab. A 450-mm More-than-Moore fab in particular would have to support a large number of processes but there might be advantages in finding common ground and common processes. The bigger obstacle is making joint ownership work. This has usually prooved problematic in the long term. There is also a lack of the risk-tasking spirit that existed at the birth of the European experiment and which drove such things as GSM cellular mobile phone technology and the Megaproject and JESSI.
I am sure the EU/EC is aware of a need to reinvigorate European manufacturing and value creation in areas like nanotechnology and the need to do something to prevent manufacturing continuing to drain east. The European companies tend to just want R&D support for their nearer term technologies on 300-mm wafers. I don't see Carlo Bozotti of STMicroelectronics, Rick Clemmer of NXP Semiconductors or the incoming Reinhard Ploss at Infineon Technologies changing that position any time soon.
And with regard to Penn's idea of a 450-mm fab to given European companies an advantage behind the leading-edge there are yet more hurdles to be overcome.
More-than-Moore technologies for the most part do not require 450-mm diameter wafers and in some cases are positively uneconomic on such large wafers. It is not practical that an entire chip product's lifetime requirement can be produced on just a few wafers. Indeed there is discussion in the MEMS community that 300-mm wafers are still way too large for any designs and that 200-mm wafers are only just becoming the sweet spot for consumer MEMS with applications that need high volumes.
We should remember that Moore's law when originally coined was an economic argument and not a technical one. The only way I can see 450-mm wafers being used for the smaller volumes and wafer area required by many behind-the-leading-edge technologies would be if there was a business model change so that multiproject wafer runs could become more mainstream, rather than a prototyping vehicle.
That could happen (see Adapteva close to sampling 28-nm, 64-core coprocessor and InvenSense opens up process to enable fabless MEMS) but again is probably not a sufficient condition. There are arguments regarding amortization of investments which mean that the newest, most expensively produced wafers should be used for the largest die with the highest value per silicon area. Older written-down fabs running smaller wafers have always been used to run older processes and chips with lower selling prices.
The best hope I see is that the bureaucrats in Brussels can call in some favors with the likes of Intel and Globalfoundries, who want to be good European citizens, to have a leading-edge volume 450-mm wafer fab be located in Dublin or Dresden.
The full report can be downloaded from the European Commission's website at http://cordis.europa.eu/fp7/ict/nanoelectronics/documents/450mm-final-report.pdf
Related links and articles:
www.futurehorizons.com
News articles:
European report considers 450-mm More-than-Moore fab
InvenSense opens up process to enable fabless MEMS
Adapteva close to sampling 28-nm, 64-core coprocessor
|
|
|
|
|
Navigate to related information
the_floating_ gate
6/7/2012 12:58 AM EDT
OK
Peter - I give you some real insight
Once upon a time there was LRCX and NVLS.
Intel wants 450mm - NVLS said no.
SO Intel pushed some buttons -
and in return LRCX will get some etch business from Intel.
That's how it works in the real world.
KLA already supplying 450 mm tools to Intel .
You just way behind the curve
Sign in to Reply
peter.clarke
6/7/2012 5:56 AM EDT
Unless you mean KLA is supplying 450-mm capable tools to Intel IN LEIXLIP I don't see how your statement has a lot of relevance to this article.
I was trying to elaborate on whether there is an appetite for the cost of 450-mm wafer processing in Europe -- ever, and particularly by European-based companies.
KLA supplying 450-mm tools to Intel in Oregon and Arizona would make it more likely that Intel MIGHT produce on 450-mm wafers in Europe, some day.
Intel has been getting closer to the European Commission over the last five years. Spreading the 450-mm love around might be part of the cost of being a good European citizen.
Sign in to Reply
mifb
6/7/2012 8:28 AM EDT
I think the most interesting point in the article is that European managers and companies are not able to compete against Asian and US American companies in advanced electronic production and products, and that the result can allready be seen in the crisis of some European countries and the Euro.
When you look at TV's, mobile phones and memory products as DRAM and NAND Flash, all production more or less disappeared from Europe (Nokia is still there at the moment, that's true for the moment). TV's are gone, Siemens gave up mobile development and production, Qimonda is gone, ST gave up all NAND activities, Infineon sold its wireless to intel, ...
The question is what will be the basis for the European prosperity in the future when the European companies, countries and the EU commision does not see any value in the advanced electronic industry and does not show any activities to keep these sectors in Europe, as it was the case in the mentioned examples.
This seems to be very different in countries like Abu Dhabi in form of ATIC which spend billions of dollars to form Global Foundries.
For all this reasons I think there will be never a 450-mm Fab or even Foundry in Europe, because there are simply no capable managers nor European politicians.
Sign in to Reply
peter.clarke
6/7/2012 10:00 AM EDT
Your points are well made.
I am prepared to be corrected -- but I believe Nokia no longer has any manufacturing in Europe.
Sign in to Reply
Adele.Hars
6/8/2012 7:28 AM EDT
There's a lot of discussion of SOI in the report. In case anyone's wondering, ramping of SOI wafers (for FD-SOI, FinFETs, PD-SOI etc) at 450mm is on schedule for the industry ramp. See http://www.advancedsubstratenews.com/2011/09/ultra-thin-wafers-for-450mm-fd-soi-on-schedule/.
Sign in to Reply
hm
6/9/2012 7:48 PM EDT
450mm Fab may be way to go for Europe. It will offer many future avenues to Europe's R and D. EU should support Intel to come with 450 fab in Europe.
Sign in to Reply
peter.clarke
7/13/2012 8:49 AM EDT
STMicroelectronics, Infineon and others may disagree.
They might like to see the European Union support them with upgrades of their older fabs for More-than-Moore production.
So who should we support: U.S. companies Intel and GlobalFoundries that manufacture and create jobs in Europe..or European companies who manufacture overseas?
Sign in to Reply