10 observations from SPIE Litho

步從容

10 observations from SPIE Litho

Mark LaPedus, Dylan McGrath EE Times (02/26/2010 1:34 PM EST) SAN JOSE, Calif. -- The themes of this year's SPIE Advanced Lithography event were clear: ''D and D.'' That stands for delays and double-patterning. Indeed, EUV is delayed. So is maskless. And nano-imprint is still stuck in R&D. So, chip makers must continue to ride the 193-nm immersion wave. They must also look at dreaded double-patterning techniques. In any case, here's what EE Times observed at SPIE: 1. EUV woes. Extreme ultraviolet (EUV) lithography is delayed--again. Now, the industry faces dreaded double-patterning or some variation of the technology to extend 193-nm immersion. That article can be read here. 2. Self-assembly grabs the spotlight. If there is a darling of this year's SPIE, it just might be directed self-assembly, a technology that emerged in recent years to land on the International Technology Roadmap for Semiconductors (ITRS) and is considered a potential candidate to extend optical lithography. That article can be read here. 3. Direct-write litho still facing uphill climb. At least according to one prominent lithography researcher, multi-beam production tools are still a minimum of five years away. That article can be read here. 4. Nano-imprint still stuck in R&D. While nano-imprint has not cracked mainstream production in semiconductor fabs, as some had hoped at one time, the technology has been delayed in perhaps its biggest potential market: hard disk drives (HDDs). That article can be read here. 5. Tool vendors are cautious despite the upturn. The fab tool market is ''hot'' and litho vendors are seeing lead times stretch out to 15 months for new orders, said G. Dan Hutcheson, CEO of market research firm VLSI Technology Inc. Still, vendors are worried. Why? ''The market is recovering. Last year, we were at the bottom and it was scary. It's not quite time to celebrate,'' said Hamid Zarringhalam, vice president of technical sales and marketing at Nikon Precision Inc. ''Things are looking better. We are seeing growth. But it will take time to fully recover.'' Another tool vendor sees a ''double dip'' coming. Others still see more technology buys verses capacity buys. 6. Where's the booths? The SPIE event itself seemed smaller than last year, but attendance was OK. Missing at the event were the booths of the big name players. Applied, ASML, Canon and Nikon did not have booths. So foot traffic in the exhibition hall was light. OK. It was a ghost town. 7. Let's laugh a little. Let's face it: SPIE is sometimes dry and boring. However, TSMC tried to change the mood. During a presentation, TSMC attempted to convince the audience that EUV is cheaper than double-patterning. In a chart, it proved its point. The trouble: TSMC neglected to add EUV mask costs to the equation, meaning its data was wrong and pointless. It did draw some much needed laughter during the dry event. Thanks for the memories, TSMC. 8. M&A in the air? We get the feeling it's a make-or-break year for some litho vendors. IMS, Mapper, MII and others must show more progress--or else. Consolidation appears to be in the air. Our feeling: ASML could enter the nano-imprint field, which could shake up the landscape. 9. We want to party! There were fewer receptions this year. Nikon had its event. Canon did not have its party. Brion, now part of ASML, did not hold its annual event. ASML had a party for customers, but it was not open to the media. To be honest, SPIE was a gloomy event. 10. No time to party. On the other hand, there's too much to cover at SPIE (i.e. EUV, ML2, nano-imprint, self-assembly, etc.). So we have little or no time for the receptions or parties. (Ok. Maybe one or two).

步從容

EUV metrology R&D doesn't add up
Mark LaPedus
EE Times
(02/24/2010 5:52 PM EST)
SAN JOSE, Calif. -- The numbers for the development of future EUV mask and metrology tools are too costly and do not add up, according to experts at the SPIE Advanced Lithography event here.
As reported, chip-making consortium Sematech has recently launched a consortium to develop metrology tools for detecting defects in advanced masks needed for extreme ultraviolet (EUV) lithography. Producing prototypes of these EUV metrology tools is expected to cost an estimated $200 million or more.

The problem is that there are no viable EUV metrology tools in the market today. If you can't inspect EUV-enabled masks--which must be defect free--EUV will not succeed in the market.


Current machines are not sensitive enough to detect defects in EUV masks. So many believe the industry must develop actinic-based tools, which have an equivalent 13-nm wavelength as EUV. This is a costly and monumental effort.


''The problem has been known for a long time,'' said Kurt Ronse, director of the lithography department at IMEC, an R&D organization, but the issue has been a ''chicken and egg'' situation.


The industry was looking for the metrology tool makers--such as Applied Materials, KLA-Tencor and others--to foot the bill and develop EUV machines. But on the other hand, the tool makers have been slow to devise these types of machines.


This is due to the associated costs involved and lack of return-on-investment, said Chris Mack, a consultant and gentlemen scientist. All told, it could take ''four years'' before the EUV metrology tools are ready, Mack said.

This implies EUV could be delayed again--or not happen at all. This is not a surprise to most lithographers, but EUV lithography is delayed--again. EUV is now targeted for chip production at the 16-nm half-pitch node.

Here's more bad news: Now, the industry faces dreaded double-patterning or some variation of the technology. Double-patterning adds cost and complexity to chip production.

There was some hope that EUV would be ready for the latter part of the 22-nm half-pitch node, which is slated for the 2011 timeframe. But once again, EUV has been delayed due to lack of power sources, resists, defect-free masks and metrology tools.

Regarding the metrology tools, the numbers do not add up. It could take up to $100 million (or more) to devise an actinic-based metrology tool. That's for the R&D alone, according to one expert.


The bill-of-material costs are around $5-to-$10 million. What's more, there are few customers for EUV metrology: Only six or so chip makers--if that--will buy EUV scanners in the first place.


So, a tool maker must charge a premium: Some believe that a tool maker must charge $40 million or more for a single machine.


That does not include the EUV scanner itself, which could go for $100 million per system if or when they appear. Others are throwing out numbers like $130 million for a EUV scanner.