You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!

  • ASML and IMEC EUV Progress

    Article: Shorter, better and easier PCIe and NVM Express Verification flow with advanced technologies-asml_next_euvl_tool-min-jpgDay 1 of the SPIE conference featured a number of customer updates on the status of their EUV programs. On Tuesday morning we got to hear ASML’s update on their work.

    Alberto Pirati of ASML presented EUV lithography performance for manufacturing status and update

    ASML has 8 – NXE 3300 systems in the field with a 0.33NA and currently running ~55wph. They have also shipped NXE3350B systems with ~125wph performance. The NXE3400B will ship this year and is expected to be the production workhorse ~145wph.

    NXE3300B has shown >1,000wpd with 6 systems with the 80-watt source and 2 with the 40-watt source.

    NXE3350B at the factory is running ~1,400wpd. A two week run at the factory on an NXE3300 produced ~800wpd with ~86% availability. ASML has also shown 8 days at ~950wpd and demonstrated 95wph on an NXE3350B exposing 96 fields at 20mJ/cm2. Further improvement Is expected with a 125-watt source setting.

    Over the last year over 300,000 wafers have been exposed by customers. Last year at SPIE source power was at ~100 watts, this ~200 watts.

    7 systems in the field are showing availability of 70% to 80% with a split between scheduled and unscheduled down time.

    The EUV laser system uses a CO2 laser to vaporize tin droplets to create EUV light that is then collected and steered to the exposure system by a collector mirror.

    A diagram was shown of the tin droplet generator for the EUV laser. Tin is loaded into the generator and melted, an inert gas is used to pressurize the reservoir and the tin then flows through a filter into the nozzle. Droplets are generated at a rate of 50KHz and fed to the CO2 laser. In Q3-2015 run time was ~100 hours, in Q3-2015 the second generation systems reached ~200 hours, a generation 3 generator at ASML is currently running ~900 hours (>1 month). The time to swap out a droplet generator has now been reduced from 14 hours to 8 hours.

    The collector mirror lifetime has been improving:
    • 40 watts, 100% reflectivity to 50% reflectivity after 60 gigapulses.
    • 80 watts, 100% reflectivity to 50% reflectivity after 80 gigapulses.
    • 125 watts, 100% reflectivity to 60% reflectivity after 100 gigapulses.

    A EUV 200-watt source with dose control has been shown for 1 hour with increased hydrogen flow. In 2015 a 125-watt source with >75wph, 70% availability and >1,000wpd was achieved. For 2016 a 250-watt source with >125wph, >80% availability and >1,500wpd is on track.

    Front side reticle defectivity was ~20 in 2010, 2 in 2011 and ~0.1 in 2012. The test is known not be sensitive enough but the defectivity still came down 10x per year.

    They are testing pellicles at customers with >200 exposures done towards a goal of 1,000. No reticle particle adders seen and the reticle appears to work as expected.

    Overlay appears to be good, with 5 – NXE 3350B systems qualified with 16nm resolution, CDU =<1.3nm. Overlay and focus are both performing to specification and matched for over a year at a customer. 16nm resolution was shown with 19% exposure latitude (EL), 4.4nm line width roughness (LWR) @ 18.5 mJ/cm2.

    The NXE3350B will provide 13nm resolution for 5nm logic and 13nm DRAM. 13nm ½ pitch with 17% EL, 4.2nm LWR 31 mJ/cm2 and 150nm depth of focus (DOF) has been shown. 20nm contacts holes have also been produced with 30 mJ/cm2. Photoresist are also progressing.

    All in all very solid progress over the last year.

    Christopher Wilson of IMEC presented EUV and 193i based patterning for advanced node integration
    This paper summarized work comparing EUV to argon fluoride immersion lithography, alternately abbreviated ArFi or 193i (193 is the wavelength of argon fluoride lasers).

    The comparison was done for M0/V0/M1/V1/M2 product like layout for a 24nm half pitch.
    • M1 - litho-etch-litho-etch-litho-etch (LE3) was compared to EUV single patterning (SP).
    • V0/V1 - LE2 was compared to EUV SP.
    • M2 - self aligned double patterning (SADP) + a block mask to EUV SP.

    From 2006 - 2011 IMEC had an EUV alpha demo tool, from 2011 to 2015 an NXE3100 tools and as of July 2015 an NXE3300 tool. For 193i exposure they have an NXT1950 with a Soludo duo and an NXT1970 with a TEL track.

    The LE3 process prints 42nm lines with 144nm pitch and then shrinks them with a total of 27 steps. The SADP process creates 48nm lines on a 96nm pitch and then shrinks them with a total of 18 steps. EUV creates 24nm lines in an 8 step process. Cost is LE3 > SADP > EUV.

    For 24nm lines EUV has better critical dimension uniformity (CDU) and slightly worse line width roughness (LWR). LWR improves with higher dose. LE3 etch CDU signature could be improved with more optimization. LE3 suffers from corner rounding due to the longer wavelength. EUV patterns are better than SADP + block. SADP + block has better CDU and LER/LWR than EUV but has sharp ends to the lines. Splitting the block pattern into two masks gives better line end definition.

    EUV variability is better than 193i with more line "pull back", less hot spot issues, low bridging and pinching risks. EUV has less via area variation than 193i.

    In summary EUV had the best overall performance, SADP + block was second best and LE3 the worst. The biggest issue for EUV was LER and he thinks that can be improved.

    Other articles from Scott...