The eBeam Initiative’s 11th annual Luminaries survey in 2022 reported EUV fueling development of the semiconductor photomask business whereas a panel of specialists cited quite a lot of issues in shifting to Excessive-NA EUV throughout an occasion co-located with the SPIE Photomask Expertise Convention in late September. Business luminaries representing 44 corporations from throughout the semiconductor ecosystem participated on this yr’s survey. Aki Fujimura, CEO of D2S, Inc., the managing firm sponsor of the eBeam Initiative, moderated the panel dialogue about this yr’s survey outcomes. He was joined by Naoya Hayashi, DNP Honorary Fellow; Harry Levinson, Principal Lithographer at HJL Lithography; Ezequiel Russell, Senior Director of Masks Expertise at Micron Expertise; and Noriaki Nakayamada, Senior Expertise Skilled at NuFlare Expertise. That is half one of many dialogue.

Picture left to proper: Aki Fujimura, Naoya Hayashi, Ezequiel Russell, Harry Levinson, Noriaki Nakayamada

Fujimura: The outcomes of this yr’s survey present a really brilliant outlook for multi-beam masks writing and EUV. Let’s take a look at EUV first. 78% of these surveyed stated EUV masks would make a optimistic contribution to 2022 photomask revenues, proven in determine 1. Harry, EUV presently includes single patterning of the crucial layers versus a number of patterning producing a number of masks for optical lithography. So why doesn’t that find yourself lowering masks revenues?

Fig. 1: Luminaries surveyed view EUV masks as a optimistic for masks revenues.

Levinson: Every little thing about EUV is costlier. There are many causes EUV masks are costlier beginning with the masks clean. I additionally assume with out pellicles, folks make a number of masks. So though it’s single patterning, you might have a number of masks. However the EUV advantages are price it – yield, cycle time, turnaround time (TAT). Suppose that we change 10 layers triple-patterned with optical lithography with 10 single publicity EUV lithography steps. With 20 fewer lithography operations, discount of skinny movie depositions, etches, and cleans, at 1.5 days between masks steps, whole course of time is decreased by practically one month through the use of EUV lithography! Mainly, EUV is “costly however price it.”

Russell: Whereas EUV masks are costlier, the extra decision that EUV brings makes the ROI optimistic for lots of corporations at the vanguard. EUV masks lifetimes are shorter, so that you order extra of the identical masks which will increase the masks spend.

Hayashi-san: EUV masks manufacturing supplies and course of prices are fairly excessive, contributing to price of EUV masks. Masks blanks are costly and EUV-specific high quality assurance instruments are wanted.

Nakayamada-san: Defect mitigation provides to price. You want extra EUV masks blanks which will increase the entire masks income.

Levinson: So, EUV simply prices extra, and it’s price it due to enchancment in wafer high quality.

Fujimura: What concerning the EUV masks challenges – have they been solved for 0.33 NA? Let’s speak about pellicles first. In determine 2, luminaries surveyed assume EUV pellicle utilization is trending up.

Fig. 2: Luminaries surveyed see growing use of EUV pellicles.

Hayashi-san: Pellicles are presently not an issue as a result of we already know EUV pellicles are in high-volume manufacturing. Okay for now however after all enhancements are all the time wanted.

Levinson: Whether or not to make use of an EUV pellicle is a call every firm should make, buying and selling off between lack of transmission and yield. Michael Lercel from ASML introduced an excellent paper on this subject a couple of years in the past. For recollections, you may take the danger of defects however for logic, pellicles are factor. Folks appear to be following Lercel’s logic.

Fujimura: EUV masks inspection has been a problem, however the Luminaries stay assured once more this yr in EUV actinic inspection with 69% agreeing that will probably be utilized in 2023 for HVM, proven in determine 3. What modified from final yr’s survey is the settlement on eBeam masks inspection, which is right down to 24% from 42% final yr.

Fig. 3: Luminaries surveyed have faith in actinic EUV inspection.

Hayashi-san: I believe actinic could possibly be taking consideration away from eBeam inspection. In future, Excessive-NA EUV might have extra decision. Even actinic inspection might attain a decision restrict. eBeam inspection could possibly be one of many candidates for Excessive-NA EUV.

Fujimura: Let’s speak about Excessive-NA EUV. New questions on this survey point out that whereas Excessive-NA EUV first utilization could possibly be by 2026, it’ll take till 2027 and past for broad HVM by a couple of firm (determine 4). Harry, you’re on the IEEE Worldwide Roadmap for Units and Programs (IRDS). What’s the IRDS roadmap relying on?

Fig. 4: Luminaries answered new questions on Excessive-NA EUV.

Levinson: Insertion is basically depending on ASML’s timeline. Can’t be quicker than ASML’s timeline which is early 2025. However for high-volume manufacturing, it’s a must to have a number of instruments up and operational. If you take a look at a number of corporations, it will be tough to see a couple of firm by 2026 however by 2027, I believe we might see loads of exercise in wafer fabs.

Fujimura: Ezequiel, I think about you’re saying “we’re busy with .33” however any ideas on Excessive-NA EUV? When did you say it’d be broadly utilized by a couple of firm?

Russell: For reminiscence purposes it could be slightly totally different than logic. First, we are going to push .33 EUV to the decision restrict. For Excessive-NA, it’s not clear it’ll present sufficient ROI for insertion due to different scaling difficulties. It could require a unique structure to beat. So Excessive-NA EUV might not be as enticing for reminiscence makers. It’s going to rely on when it’s accessible and when it’s price efficient for reminiscence makers to transition.

Fujimura: What are the masks challenges for Excessive-NA EUV?

Hayashi-san: The massive distinction for Excessive-NA masks is the anamorphic optics which is 8x magnification within the Y path and 4x within the X path. This choice was made explicitly in order that the present masks infrastructure works. Perhaps a problem is the extra crucial specs within the 4x path, or each instructions.

Fujimura: What concerning the perspective from a masks author, Nakayamada-san?

Nakayamada-san: There’s a good alternative to align the writers to match a looser spec within the Y path. Although the spec doesn’t change, it’ll nonetheless work.

Fujimura: What about round masks? It’s associated to Excessive-NA. What’s the fundamental concept, Harry?

Levinson: If you go to the anamorphic optics with 8x and 4x magnification, and you retain the masks the identical measurement, it’s problematic on the wafer as the sector measurement is just half of what we’re used to. Then you need to make the most of methods like stitching which isn’t a trivial factor. What you actually need is an even bigger masks. I consider Intel made the statement in the event you might make your masks on 300mm spherical substrates, it’s a well-known process. There are loads of issues that will must be checked with round masks. Would it’s a must to modify the reticle stage within the scanner? In any other case, I believe it’s proposal and value taking a look at much more intently.

Hayashi-san: We tried round-shaped masks previously, a very long time in the past. Can the masks author deal with it? Writing uniformly throughout an even bigger discipline might be a problem. Expertise-wise, round masks are a good suggestion however undecided it’s enterprise concept. A service provider masks home doesn’t have a 300mm course of software in order that’s type of an enormous problem for us!

Fujimura: Nakayamada-san, can the masks writers do this? Isn’t {that a} huge change?

Nakayamada-san: The issue could be if a buyer requests an interchangeable stage writing a sq. six-inch masks on the similar time they wish to write a round 300mm masks.

Russell: I agree with the panel that from a expertise viewpoint, it’s possible and it is smart since we have already got the 300mm machines within the wafer fabs. However I see this as an issue for masks retailers. They don’t wish to preserve twin tools for normal six-inch masks in addition to for round masks. I don’t know what number of nodes Excessive-NA EUV will assist and what number of layers per node, so we could also be speaking a couple of small variety of masks. Altering the infrastructure for only a few nodes might not make financial sense.

Source link


Please enter your comment!
Please enter your name here