EUV lithography faces short-term gains and long-term challenges in resist, actinic inspection, mask and mask pellicles and light sources.
A leading chip maker will start inserting extreme ultraviolet lithography (EUVL) into commercial production this year. However, there are still some pending issues which will impact how quickly remaining chip makers start using EUVL in fabs–scanner uptime (mostly related to source), lack of commercial actinic patterned mask inspection tools and readiness of EUV mask pellicles.
The 250 W sources and corresponding mask pellicles are almost ready for 125 wafers per hour throughput scanners. This year, I expect to see progress toward 90 percent uptime for high volume manufacturing with EUVL scanners. All of this should speed adoption for remaining leading-edge chip makers.
Readiness of commercial resists will be a leading challenge for EUVL in future nodes. EUV resists use secondary electrons chemistry, a different approach than current resists. We need to figure out all the knobs to address random printing failures and micro bridging, which were reported last year.
At EUV wavelength, stochastic effects become important. In addition to better understanding of secondary electrons dynamics, we need to address material inhomogeneities at the nanoscale for producing production level EUV resists. For over five years, we have seen some great results in new types of resists from labs, but we need to see the same from fabs.
Commercial tools for actinic patterned mask inspection is a big gap that needs to be closed. Wafer inspection is now increasingly being used to detect defects on masks, but it is costly and inefficient.
Wafer inspection itself needs to evolve and move beyond 193nm to smaller wavelengths in order to offer better resolution for coming generations. New types of EUV sources and corresponding optics are expected to play a part here.
I expect alternate suppliers for commercial actinic patterned EUV mask inspection to emerge soon. These tools will be powered by 13.5 nm plasma sources or HHG.
I am not sure if current pellicle designs and materials are extendable all the way to 500 W. I believe that pellicle designs must evolve. I now see that source power may be able to get to 500 W in the future, but I am not sure what technology will provide power greater than 500 W – Sn LPP or FEL.
Pellicle designs will evolve with new materials. The challenge will be to test and integrate the best options into the scanners. E-beam inspection will gain, but optical inspection will remain the primary workhorse as work continues to take it beyond 193 nm to improve its resolution.
ASML shipped 10 scanners in 2017 and about twice that number are expected to ship in 2018. Some people ask me if scanner optics or mask blanks supply will be the bottleneck as the number of scanners delivered are expected to double every year from 2018 onward. I expect some challenges in the supply chain as EUVL ramps up, but nothing glaring has come into sight yet.
Among the players, GlobalFoundries has emerged with manufacturing strength, investments and progress in EUVL technology development. They have a good technical and manufacturing team, and I expect them to make steady gains in the share of foundry business, as well as in getting EUVL ready for their fabs.
—Vivek Bakshi, president of EUV Litho, Inc., will release a new EUV textbook at the 2018 SPIE Advanced Lithography Conference in San Jose. He also has published with colleagues a special section in SPIE JM3 that includes papers on challenges for EUVL at the 3 nm node and beyond.