Electronics Weekly Electronics Design & Components Tech News
Four years ago, research by SEMI suggested that 450mm may only reduce die cost by 8% – not worth doing at an R&D cost of $25-30 billion and the $10 billion-per-fab outlay required by device manufacturers.
However, last month Malcolm Penn CEO of Future Horizons which is heavily involved in the EU’s 450mm plans said that 450mm reduces die cost by 30%.
Nonetheless it was reported before Christmas that ASML’s effort to develop 450mm has been ‘reduced’. ASML has confirmed that this pretty much reflects the situation
Although Samsung, TSMC and Intel have not pulled out of the 450mm development programme, the programme has slowed because of an intervention by them.
The intervention could be the result of disagreement by the three companies. Samsung and TSMC don’t want to enable Intel, because Intel’s nascent aspirations in the foundry business now put Intel in competition with TSMC and Samsung.
“No one’s made the commitment to 450mm yet,” said Penn, “for the Big 3 it’s Maybe; for GloFo, Toshiba, Micron and Hynix it’s Maybe; for the equipment firms it’s No; for the Big 3 European firms it’s No.”
There was a time when people were serious about 450mm and put serious money behind it.
In mid-2012, Intel paid $3 billion for a 15% stake in ASML and said it would put another $1 billion into the R&D projects, Samsung bought a 3% stake in ASML for €503 million with a commitment to put up €276m for ASML’s R&D programmes, and TSMC paid €838 million for a 5% stake in ASML and another €276 million for the R&D projects. Intel is also funding a 450mm development effort by Nikon.
How, one wonders, can an industry worth $300 billion last year, which was worth $275 billion in 2007 and first hit $300 billion in 2010, be able to afford a $30 billion R&D effort and $10 billion apiece for factories?
The answer, of course, is that the industry can’t afford it. But a public-private effort that brings in government and consortia stretching across the industry supply chain may be able to devise a way.
What is at stake is the declining cost of electronics. This is said to have caused a massive, while happily deflationary, increase in worldwide productivity for the last 40 years.
Shrinking is one way to get declining cost but, with shrinking delivering less and nearing its limit, 450mm is the only other way to reduce die cost.
If die cost stops declining, there are ways of reducing cost-per-function like 3D transistors for processors and stacked transistors for NAND.
And sub-system cost can be reduced through packaging advances like TSV.
Even without 450mm, even without shrinking, there’s life in the old silicon dog yet.
People have to wonder if the Semicondutor Equipment manufacturers want to see another 50% reduction in the amount of equipment sold, and the doubling of NRE included in every purchase that was seen when the conversion from 200 to 300mm.
@ian : If you can point me to a fab other than the Intel Ireland and D1X fabs that is truly 10nm capable please do so. Even if you use triple patterning on current litho machines (not convinced) then quite a bit of the other equipment is not up to it and will need upgrading/replacing, along with the introduction of far more advanced metrology and CMP. Indeed as we know, Intel is struggling wuth these at their ’14nm’ (really 16nm) node..
The cost savings have been outlined in detail to the relevant bodies but the iussue you raise, and all the others I have seen mentioned over the last 2 years, were included in the 28% estimate (which got rounded up to 30% along the way)..
@Mike — you’re correct that 450mm could be introduced at 10nm, but this doesn’t fit with your earlier suggestion that since fabs can’t do 7nm with current 300mm equipment they have to re-equip anyway so might as well use 450mm.
Fabs could do 10nm on 300mm wafers with existing equipment (and triple patterning?), so switching to 450mm incurs a huge cost without advancing a process node — and then it’s back to the economics, how long before 450mm would really give lower die cost than staying with 300mm including the cost of installing the much more expensive new 450mm equipment?
Another issue with 450mm is that more and more time is being spent with wafer stepping (especially with multiple patterning needed without EUV) and the throughput per die for this is the same as 300mm, so the cost saving is reduced. I assume this is taken account of in the “30% saving” figure?
@Ian : I think we are agreeing here that it is the 7nm node that is the problem, not the wafer size. But assuming we can’t do 7nm economically then 10nm on 450mm would be a way forward to reduce die cost.
As far as I can tell lack of EUV (or some other magic like DSA) is not a technical roadblock to 7nm, but it may be an economic one — triple or even quadruple patterning will put the cost through the roof (not just more masks but also yield drop), and make design extremely difficult due to very restrictive design rules — the gates can be a regular array which can be patterned, but the metal layers can’t be.
I don’t think anyone is factoring EUV into their plans at the moment. If it appears then great, if not then it’s not a complete roadblock.
Provided the whole industry who will use 450nm hold its nerve, and the EUV fairy finally sprinkles her magic dust over the wafers after all these years… 😉
Ian, this was addressed in my recent Bridge450 presentation at Imec. It doesn’t matter if the wafer size is 300mm or 450mm, most current fab equipment won’t cut it at 7nm and hence fabs will need to be requipped. Most current R&D is addressing both wafer sizes and the incremental cost of installing 450mm is being over-stated by those against the transition, such as it appears the Micron CEO.
And at 7nm an NRE will be circa $500m on either wafer size, this simply isn’t a 450mm issue, just a node issue.
But I totally agree with you that a huge number of products will stay at 28nm, this being the future ‘More than Moore’ market.
But memory, CPU and HFSI devices will be fabbed at sub 10nm and so these markets will gain the cost advantages of the larger wafer size … provided the whole industry holds its nerve !!
@MikeBryant “In fact anybody who wants cheaper wafers from their foundry” — should be “cheaper chips” not “cheaper wafers”.
If 450mm is cheaper per mm2 *including the cost of the equipment* — which means the cost of developing everything for 450mm as well as installing it — then this will still be true. But nobody should think that 450mm equipment will cost less than 300mm equipment — even per wafer — for a long time, the cost of 300mm development has now been written off.
And since a large part of the cost of a fab is equipment, this directly affects wafer cost. And the cost of the equipment depends on installed volume, which will be small -at least initially — for 450mm, All of which casts doubts on whether 450mm will be cheaper initially — and if it’s more expensive (as I suspect), how long it takes for the cost to fall below 300mm. If this is several years then anyone using it will have *higher* costs for this time, not lower, which is not a recipe for cheaper chips.
If the fab owners think that the equipment manufacturers are going to make and ship 450mm equipment cheaply so they can make money on the wafers, I think they’re living in cloud cuckoo land. The semi equipment industry was badly bitten by 300mm, they invested a lot into development on the promise of big orders which didn’t happen, and it took years for them to even break even on these costs. I doubt that they’ll fall for the same promises again…
Saying that 450mm reduces die cost by 30% is being economical with the truth — the missing phrase is “all other things being equal”,which they aren’t…
Once the very high cost of developing and installing 450mm is paid off the cost may indeed be 30% lower — but who is going to pay back the multi-billion dollar cost of development and over what timescale? The semiconductor equipment supply industry took many years to recoup the development cost of switching to 300mm wafers, and there’s still a lot of 200mm kit out there.
If 450mm only gets installed in a few new state-of-the-art fabs (which will be at 7nm or below by then) the market share will be lower than 300mm (which will still exist, see below) so the economies of scale will be less. On top of this, price-sensitive applications (of which there are very many!) will probably stay in 28nm processes using 300mm wafers because the cost of these will be written off by the time 450mm emerges and the die will probably be cheaper than 7nm processes on 450mm.
So the only people who want 450mm are those designing/making a very small number of high-volume high-priced designs like CPU or ultra-high-volume density-driven designs like memory, where the enormous NRE cost of doing a 7nm chip on 450mm — probably something of the order of $200m-$500m depending on complexity — can be paid back by even more enormous revenue.
They’ve put in a lot more than a penny David
“One tends to wonder: Is there anyone who really wants it”
Well how about Qualcomm, NVidia, Mediatek and Xilinx to start with ? Plus Apple, HP and others who haven’t actually declared so publicly.
In fact anybody who wants cheaper wafers from their foundry.
“450mm May Never Happen, says Micron CEO” … In the words of Christine Keeler … “He would say that, wouldn’t he” 🙂
A point I made at last week’s Bridge450 workshop that people have tended to ignore is that much of the current equipment is not suitable for the 10nm and subsequent nodes and will need serious amounts of R&D to upgrade it no matter if the target wafer size is 300mm or 450mm. This is something that industry needs to build into its cost models and the investments by the big three in ASML is possibly the first step in this.
As for Micron, the day Samsung go 450mm they have a “do or die” decision as no matter what the final cost gain is, it will still blitz those who don’t transition, just as it has at every other node.