Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/intel-euv-for-7nm.5911/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021770
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Intel EUV for 7nm

Intel can afford to adopt EUV even if it's more expensive than multiple patterning if it helps performance, they push the technology harder than anyone else (see their 14nm CPP/MP) since their big market is premium performance-driven high-margin large-die CPUs and their process is targeted at this -- higher cost per wafer (or per gate) is OK for them if it delivers more performance and/or lower power and their CPU customers are willing to pay for it.

This Rolls-Royce (or BMW) process model doesn't work for the foundries who are now driven by the mobile AP market, they won't adopt EUV in mass production in a big way until it offers cost savings as well, which means higher throughput/lower cost than Intel will accept.
 
IanD said:
...................

This Rolls-Royce (or BMW) process model doesn't work for the foundries who are now driven by the mobile AP market, they won't adopt EUV in mass production in a big way until it offers cost savings as well, which means higher throughput/lower cost than Intel will accept.
Click to expand...

Hmm..wasn't it TSMC that in November 2014 placed the first 2 real orders for the production EUV tool NXE3350B and
with 2 upgraded NXE3300-> NXE3350B effectively placed orders for 4 EUV NXE3350 production tools?

LONDON, 24 November 2014 - At its Investor Day, ASML Holding N.V. (ASML) will today outline its expected opportunity to grow net sales to about EUR 10 billion and to triple earnings per share by 2020.
ASML also announces that Taiwan Semiconductor Manufacturing Company Ltd. (TSMC) has ordered two NXE:3350B EUV systems for delivery in 2015 with the intention to use those systems in production. In addition, two NXE:3300B systems already delivered to TSMC will be upgraded to NXE:3350B performance.
ASML: Press - Press Releases - ASML Outlines Long-Term Growth Opportunity at Investor Day; Receives First Order for EUV Production Systems


I would think TSMC in this way announced their HVM production EUV plans (possible mid 10 nm node insertion and 7 nm node) in November 2014 versus INTEL (? large US customer in ASML press release) that announced their HVM EUV plans only yesterday as was announced as follows by ASML:
ASML: Press - Press Releases - ASML reaches agreement for delivery of minimum of 15 EUV lithography systems


It seems EUV is also well within reach of foundries, now we have to wait for the `big` Samsung orders for EUV NXE3350...

With the >15 EUV 3350 tool order by INTEL (?) one wonders if NIKON is just `kept alive` by the semifactories to keep market
regulators not messing around with the effective litho monopoly in EUV (100%) and iArF (90%) of ASML?

As long as ASML `behaves and delivers` I guess all (?) the fab companies are happy and won`t complain to market regulators?


User nl
 
In mobile I see Intel integrating more and more functions onto fewer die. This is their core competency and these machines will help with this.
 
Chipper said:
Do you think these are these dev systems or production? 15 is a lot.
Click to expand...

ASML: Press - Press Releases - ASML reaches agreement for delivery of minimum of 15 EUV lithography systems

VELDHOVEN, the Netherlands, 22 April 2015 - ASML Holding NV (ASML) today announces that it has signed an agreement with one of its major US customers to deliver a minimum of 15 ASML EUV lithography systems to support increased development activity and pilot production of future-generation manufacturing processes. The customer intends to use EUV lithography for multiple processing steps in future process technology nodes. The delivery of the first two NXE:3350B EUV systems is expected before the end of 2015. The new systems will be in addition to the existing EUV development systems already at the customer. Financial terms were not disclosed.
 
Arthur Hanson said:
ASML Shares Advance on Biggest Order for New Chip Machine - Bloomberg Business

Intel has clearly thrown down the gauntlet for EUV for 7nm. Follow the money/orders for everything else is just rumor.
Click to expand...

ASML: Press - Press Releases - ASML reaches agreement for delivery of minimum of 15 EUV lithography systems

Note it is "a minimum of 15 ASML EUV lithography systems to support increased development activity and pilot production"

"EUV is now approaching volume introduction. Long-term EUV planning and EUV ecosystem preparation is greatly supported by this commitment to EUV, kick-starting a new round of innovation in the semiconductor industry. The commitment extends the planning horizon and increases the confidence in EUV," ASML President and Chief Executive Officer Peter Wennink said.
 
user nl said:
Hmm..wasn't it TSMC that in November 2014 placed the first 2 real orders for the production EUV tool NXE3350B and
with 2 upgraded NXE3300-> NXE3350B effectively placed orders for 4 EUV NXE3350 production tools?

LONDON, 24 November 2014 - At its Investor Day, ASML Holding N.V. (ASML) will today outline its expected opportunity to grow net sales to about EUR 10 billion and to triple earnings per share by 2020.
ASML also announces that Taiwan Semiconductor Manufacturing Company Ltd. (TSMC) has ordered two NXE:3350B EUV systems for delivery in 2015 with the intention to use those systems in production. In addition, two NXE:3300B systems already delivered to TSMC will be upgraded to NXE:3350B performance.
ASML: Press - Press Releases - ASML Outlines Long-Term Growth Opportunity at Investor Day; Receives First Order for EUV Production Systems


I would think TSMC in this way announced their HVM production EUV plans (possible mid 10 nm node insertion and 7 nm node) in November 2014 versus INTEL (? large US customer in ASML press release) that announced their HVM EUV plans only yesterday as was announced as follows by ASML:
ASML: Press - Press Releases - ASML reaches agreement for delivery of minimum of 15 EUV lithography systems


It seems EUV is also well within reach of foundries, now we have to wait for the `big` Samsung orders for EUV NXE3350...

With the >15 EUV 3350 tool order by INTEL (?) one wonders if NIKON is just `kept alive` by the semifactories to keep market
regulators not messing around with the effective litho monopoly in EUV (100%) and iArF (90%) of ASML?

As long as ASML `behaves and delivers` I guess all (?) the fab companies are happy and won`t complain to market regulators?


User nl
Click to expand...

Actually with such a big order, it could tie ASML's hands, preventing other customers from getting more. This lockout strategy is pretty smart but ruthless, and has been played before, with ASML's immersion tools. Taiwan's smaller companies couldn't get enough in time as a result.
 
user_nl, TSMC ordering 4 EUV steppers is enough for prototyping/pipecleaning/10nm trialling on critical layers, it's not enough for mass production in the sense of being used for all chips at a given node. According to the ASML press release, even the big Intel order for 15 is for pilot production only.
 
IanD said:
user_nl, TSMC ordering 4 EUV steppers is enough for prototyping/pipecleaning/10nm trialling on critical layers, it's not enough for mass production in the sense of being used for all chips at a given node. According to the ASML press release, even the big Intel order for 15 is for pilot production only.
Click to expand...

IanD and others: as I read the information provided by ASML is that they are limited in their EUV production capacity. The numbers
quoted in the Table that I made (included below) are from the info CEO Peter Wennink provided when releasing the Q1 2015 results last week. As you see they plan on ramping production up to 50-60 EUV tools/year around 2019 and further.

Intel, Samsung and TSMC all bought a share in the ASML company of 15%, 5% and 3.5% respectively, and they provided R&D finances for EUV tool development. So it seems `fair` they get the first few years the bulk of the EUV tools as there is limited production capacity at ASML while they extend their production buildings, supply chain capacity, production manpower, etc. I get the feeling building a single NXE3350B tool presently takes about 6 months or so, taking into account the finite capacity of Zeiss to built the EUV reflective optics and other things.

I think the announcement by ASML of the big US order (>15 units) is meant to signal to all litho customers that they need to talk with
ASML seriously as production capacity is limited the coming years, and if ASML needs to ramp up capacity sooner, ASML need to know
as soon as possible to provide the supply chain with that info. Remember that ASML has a real monopoly of the Litho tool market (see also
the table provided by CFO Nickl where they publicly state their market share numbers as 100%, 80% in EUV and iArF respectively.

So, ASML has to keep all (!) their litho customers happy, they have to `behave and deliver` all the advanced semi litho tools the coming decade.

Just my two cents...User nl.


Table 1 (click on figure)
View attachment 14077
 
I wouldn't argue with any of that. The fact remains that Intel will dive into EUV before anyone else because their economics are different, they make money by selling the high-margin end product (CPUs) not wafers, and they have a lot more DP layers at finer pitch even at "14nm" than anyone else (8@52nm SADP for Intel, 3@64nm LELE for TSMC/Samsung?) so the benefit of EUV is bigger for them -- they can spend more to get more return even if EUV isn't competitive on cost with multi-patterning yet.

All of this assumes that the other teeny little problems like pellicles, mask inspection/repair, throughput, source cleaning, mask costs/manufacture time and so on are solved in time. ASML are confident, but then they have been for years, EUV was originally proposed for 65nm IIRC...

The other problem is that even assuming EUV makes it in time for 7nm -- which some people are doubting for true mass production -- it'll probably need double patterning by then, which puts the cost back up. Maybe still cheaper than octuple patterning with immersion, though :)
 
Last edited:
The fact that Intel ordered 15 EUV machines doesn't change the fact that the technology still is a long way to go. According to TSMC EUV is at an average few 100 wafers per day now, but it needs to be a minimum of 1000 for it to be economical.
 
Intel would not have jumped into it if the product was not ready enough to add value in volume to their process. It also tells me that they have process that can leverage EUV which means that they have a 10nm process that will yield. And I don't for one second believe that some of this production won't will find its way into mobile.
 
They are definitely keeping their options open for 7 nm. Switching to EUV is not merely a matter of ordering different ASML tools, but changing the entire infrastructure, including mask inspection, pellicles, buried defect handling, etc., which requires many other vendors to get involved. The ASML tools themselves are also slow to come, this order won't be filled until 2017, maybe later. Also, multiple patterning cannot be avoided, for continuation of EUV (5 nm, maybe even 7 nm). Since MP will be forced to become more cost-effective, the benefit extends to continuing 193 nm immersion, which is currently much faster.
 
Last edited:
Agree, this is a multi faceted effort but this is what Intel does when it comes to processors. TSMC may fab more logic and Samsung more memory but Intel will fab more bleeding edge processors by dint of all three of its end markets(Servers, PC, mobile).

My prediction is that by 2017 or 18 Intel will be # 1 in mobil chip sets.
 
Fred P, many people have predicted in the past than Intel will win in mobile by 20xx, and they've always been wrong. It's not a case of having the most advanced technology, it's having the right technology and the right IP.

Intel's processes are optimised for their core market which is high-end x86 CPUs, TSMC/Samsung processes are optimised for their core market which is mobile AP, and the two are very different (transistors, number of masks, number/pitch of metal layers, wafer cost, design rule complexity, design time/effort). So it's difficult to see how Intel can win in the mobile space with their technology -- unless they do one optimised for this like the foundries, in which case they'll be later and probably still more expensive due to "gold-plating".

Another issue is that due to the very different CPU-optimised process, a design using Intel is not only single-source but is in no way portable to any other process, in comparison the foundry processes are much more similar with much better (still not free!) portability -- not surprising since they're all targeting the same (mobile) market.

Most of the world (except Microsoft) does not want x86 CPUs in mobile, they want ARM. Most of the world does not want a Intel standard product in mobile, they want their own customised ARM-based AP. For this they want/need the big range of IP that the foundries support (through the fabless ecosystem) but Intel doesn't.

So Intel's "process and IP advantage" (finest CPP/MP, most DP metal layers, highest speed, excellent x86 cores) is not actually an advantage for mobile, it's a disadvantage...
 
“This is a clear sign that most probably Intel is moving to EUV for 7 nanometers,” Robin van den Broek, an analyst at ING Bank NV, said by phone.
Click to expand...
"for 7"? :rolleyes:

My humble guess would be 10nm production.
 
IanD, very strong points but let me give you my thinking.

First, unlike past predictions about Intel’s success in mobile Intel now has parts working in well received phones running android.

Second, my reading of the various papers published by Intel tells me they have some very interesting designs slated for the next nodes that will allow them to leap frog the ARM/Foundry power advantage. Remember in the mid 90's when RISC was going to keep Intel on the server sidelines? Intel’s tightly integrated process and design overcame the advantages inherent in RISC and I think we are in for a repeat.

Third, let’s think about the market. With only two foundries Samsung has been able to maintain a good book of business despite the strategic risks of companies like Apple funding their competitor. But Samsung can’t have it both ways especially with the Chinese phone suppliers. Chinese phone companies will be extremely reluctant to fund Samsung. Yes if GF works they could be a player but GF has a lot more to proven than Intel.

So for many companies the choice will be between Intel and ARM/TSMC. Yes, ARM/TSMC gives them more design flexibility but to be successful, in mobile means devoting resources to creating an ecosystem. I think many of the newer Chinese players are going to be happy with Intel RD on the chip side so that they can devote precious resources to the ecosystem.

So, if I’m correct we will have Apple at TSMC, Samsung with Samsung and Intel and TSMC/ARM with the rest. Let’s say Apple get 15-20% and Samsung the same. That 70% up for grabs and I think Intel wins the majority of this.

Again lots of assumptions especially on the design side so I could be way off but I remember in the mid 90's when RISC was going to keep Intel on the server sidelines. Intel’s tightly integrated process and design overcame the advantages inherent in RISC and I think we are in for a repeat.

I’m long Intel and Apple.
 
You must log in or register to reply here.
Back
Top