Electronics Weekly Electronics Design & Components Tech News
To this end the assistance of the EU is needed.
25 years ago Cees Krijsman of Philips, Jurgen Knorr of Siemens and Pasquale Pistorio of ST were making exactly the same argument.
They won the argument and a series of EU-assisted programmes from Jessi to Medea to Eniac and Catrene followed.
Crolles was funded to pursue fundamental microelectronics research, Imec was boosted into a formidable world force in chip tech technology, and the 200-300mm wafer transition was led by Infineon with support from Germany and the EU.
The problem in trying to resurrect manufacturing in Europe now is: Who is going to do it?
Infineon gave up fab after 65nm, and NXP’s in-house capability has been decimated after it pulled out of Crolles, closed numerous fabs, and went fabless for advanced processing.
It’s not possible to just drop in and out of IC manufacturing. A period out of the chip-making game makes the barrier to re-entry huge.
ST has the experience of running a 300mm fab at advanced process geometries in Crolles.
So should the EU just be thinking about buying a new fab for ST?
It seems an inappropriate use of taxpayers’ money.
But there was once a collaboration between government and a European company which resulted in the setting up of a fab which proved of enormous benefit to the nation’s economy and hugely profitable to the company.
That was Philips’ collaboration with the government of Taiwan to set up TSMC.
If ST and the EC set up a fab which gave easy, guaranteed, low-cost access to advanced manufacturing processes for Europe’s design houses and fabless companies, it could give Europe a similar boost.
Interesting stuff, Stooriefit, so, if the EU wants European fab, it should put its money into Europe’s many smaller specialist fabs like Austria Mikro Systems. That would be a pretty smart way to go.
All very interesting – but why 450mm? If we’re going to shoot for the moon why not do it in feature size or materials rather than wafer size?
The problem with 450mm is that you need lots of big die to fill a wafer – so that is commodity volumes (low margins) vs. high complexity design (requires high margins). 450mm makes little sense for anyone but Intel, Samsung, IBM and the memory guys.
Small geometries or integrated III-V or some such on a 300mm wafer would be a much better USP, fill wafers with high value parts in moderate volumes, and not break the bank on bespoke wafer handling kit.
The real space programme stuff is in the lithography and materials science, not the mechatronics.
An excellent point Will, but whereas the US has Intel and IBM which will probably keep fabbing till the end of time, Europe has only ST with advanced fab capability and now Europe has reached the moment of decison: Should Europe let manufacturing go – in which case advanced IC manufacturing will probably be gone from for ever, or should it take a risk, develop the next generation of manufacturing technology, and keep Europe in the IC manufacturing game? It’s an interesting dilemma.
David,
Let’s look at the global perspective. Once upon a time, “Real Men had Fabs” (apologies to Jerry Sanders). A chip maker’s worth was measured by its ability to have a better process than its competitors. “My NMOS is better than your PMOS,” “My CMOS is better than your NMOS;”My CHMOS 3 is better than your CMOS 2,” etc. Remember that the major semiconductor companies were founded by and run by physicists or material science people…because the fab was the key, not the product. Think back to William Shockley, then Gordon Moore and company. Today, there are only a few IDMs (Intel as #1) but more than a thousand fabless companies (with Qualcomm #1). The fabless companies tend to be headed by electrical engineers, not physicists. That’s because their reason for being is the product design, not the fab capability. Let’s face it: TSMC leveled the process playing field and now most chip houses have access to 40nm (and soon, 22nm) fabs.
With the exception of Intel, there are no new major fabs being built in the U.S. Intel’s new Fab 42 here in Arizona has just broken ground and will employ 300mm wafers with (future) feature sizes down to 14nm. Clearly, they don’t see a need for 450mm wafers anytime soon. Should Europe choose to establish a 450nm fab, funding the supporting ecosystem involving new fab equipment that has to be developed and the associated wafer handling equipment will likely make its cost so high that break-even will probably approach the end of economic life of the facility.
Will
When Europe sets its mind to something, it can be a potent force, Philip, the MegaProject/Jessi/Medea/Eniac have shown that. We have the best car industry in the world, but in the 1980s it was nearly taken out by Japan. Political decisions saved both the European car industry and the European semiconductor industry but neither of those decisions would have accomplished anything were it not for the fact that Europe is full of well-educated, highly innovative people.
An excellent point, Robert, but what’s the alternative? Either we say ‘We can’t do fab’ or we say ‘We’ll have a go’. 25 years ago Europe plumped for the ‘Have a go’ option. Now the EC has the political will to make the same choice. If 25 years ago we’d opted for the ‘Europe can’t do fab’ option where would the European semi industry be today? Just as, in the 1980s, if we’d not used political will to stop Japanese car imports, we’d have no European car industry today. At critical moments it can take political will to make the difference.
Well Rob, Imec is doing the basic DRAM design for all the major DRAM houses – so maybe there’s a potential product.
I agree that there is no fundamental reason why a 450mm 10nm fab could not be built in Europe. The question would really have to be why? and who would use it?
At 450mm that is a big area of silicon per wafer lot and at 10nm the tooling is going to be astronomically expensive – so we need to be talking huge numbers of devices – not the type of country traditionally occupied by European design houses.
There is absolutely no doubt in my mind that Europeans could create a 450mm wafer fab with 10nm feature sizes BUT unfortunately the fab game is no longer about wizz bang “got to get it here” technology. It’s almost all about cents/sqmm at good yield. The days of advanced fab geometries driving fab gross margin died at least 5 years ago.
So the question is not can you build it, but rather will product produced by this enterprise be competitive?
Robert is acting like one of those senior managers who has read a book on management and when the business failed goes to another job with the attitude “I did everything I could”. Instead of looking for new ways to do things, it is treated as a cash cow and allowed to die slowly.
The average European engineer has much more innovation than your average Asian engineer. When these engineers are trusted and managed correctly the industry can compete with Asian industry, especially with government help to level out the massive financial input the Asian economies give their high tech companies.
You seem to be coming from the premise, Robert, that Europeans are Hottentots trying to do something hopelessly beyond them – when much of basic semiconductor science, process technology and tools development is done in Europe. If Europe decided to set up a 450mm 10nm fab they could do it. If such a fab was made accessible to European designers it would boost European IC sales.
David,
with all due respect, I believe it is you that is missing the point.
As an investment (career) which would you rather have:
– a dynamic profitable fast growing Taiwan design house (let’s call it MTK)
OR
– a dying IDM, with it’s hand continually out, always looking for a little EU money to keep the lights on, for just a little longer (lets call this option ST_EU)
The first option has a real future whereas the second option is just a slow death. Maybe you’ve reached the critical stage in life where a slow death sounds attractive, personally I hope I never get that old!
Well Yes, Robert, that makes practical sense. But if the EC is designating IC manufacturing as a ‘Key Enabling Technology’ – which means the EU is prepared to put its hand in its pocket for to fund it – it would be a bit churlish to spurn it.
Unfortunately if the Fab is not competitive than neither are the design houses, so you’re just replacing one unpleasant and uneconomic option (go to Taiwan and beg for cheap wafers)with another equally unpalatable choice (design in EU fab with ST/EU).
Maybe a better option is to bite the bullet, cut EU ties and move design to Asia.