EUV lithography has been researched extensively, but several big issues remain unresolved. In the mean time, Intel has pushed DUV lithography down to 22 nm structures. I am now in doubt if EUV lithography has a future at all, if it will ever be competitive against DUV. What is your opinion?
I have written before and Intel's Yan Borodovsky also has mentioned about shot noise as applied to EUV lithography. This is the effect where the feature gets so small the number of photons used is subject to Poisson noise. The way to beat it is to increase the number of photons, i.e., increase the dose. This puts even more pressure to increase the EUV source power. This limitation is just as powerful as the Rayleigh diffraction limit, but hits from the shorter wavelength side (because each shorter wavelength photon carries more energy) rather than the longer wavelength.
Progress on EUV is slow, but I'm not sure it will be competitive - perhaps 'complimentary' with DUV.
I have written before and Intel's Yan Borodovsky also has mentioned about shot noise as applied to EUV lithography. This is the effect where the feature gets so small the number of photons used is subject to Poisson noise. The way to beat it is to increase the number of photons, i.e., increase the dose. This puts even more pressure to increase the EUV source power. This limitation is just as powerful as the Rayleigh diffraction limit, but hits from the shorter wavelength side (because each shorter wavelength photon carries more energy) rather than the longer wavelength.
I have to add one more point, it is also important. I will mention it at this year's SPIE (paper 8683-36). Although sufficient power can enable the higher dose with sufficient throughput, with the dose being so high, two things can happen. First, there is more outgassing. This is especially so with for example, PMMA. Second, the resist may crosslink, changing from positive tone (defining holes and trenches) to negative tone (defining islands and lines). Both are very undesirable from optics contamination and patterning points of view. Of course there are some oxide negative tone resists which can be used with such high doses, but negative tone cannot pattern back-end metal features conventionally. Plus negative tone is more flare and defect sensitive.
This might be a debateable point, but perhaps ASML can look at even shorter wavelengths, at least to open up more possibilities for consideration.
The NIST link was pretty helpful even though it seems to have kept a low profile since 2011. Thanks.
The tools can be shipped but the source power won't be enough, so they would be used for development. Also, for a given company, there aren't enough for manufacturing anyway.
http://www.eetimes.com/electronics-news/4390667/ASML-says-fast-EUV-machines-coming-by-2016
"ASML has 11 of the NXE:3300 machines on order but these are now likely to be delivered to customers in 2013, Meurice said. The final integration of the first NXE:3300 at ASML is expected in October or November he added. The NXE:3300 systems shipping in 2013 will only be used for developing manufacturing process technologies, Meurice said but will still result in 800 million euro (about $980 million) of revenue for ASML in 2013."
Information transfer from aerial image to resist film is always a challenge with higher photon energies. Partitioning of aerial and latent image densities to minimize the effects of shot noise might well be reaching its limit, as dependence on chemical amplification to define the latent image becomes a risky proposition, as a function of pattern feature size and density. EUVL might well be optical lithography's last gasp, for this reason, but there has been so much investment in making EUV work, that you'll see it used on ultra-critical layers, at least for a time. It is interesting to note that the best electron beam resist materials are, in fact, not chemically amplified. The push for other patterning processes, such as selective etching block copolymers, appears to be genuine.
If EUV is only going to be used for sub-wavelength nodes, it doesn't look promising.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science
- Thin Films and Nanotechnology