They already have the tools for DUV.
They are very close to getting the tools for EUV.
Involves increasing the precision of a tool.
Chinese semiconductor industry
- Thread starter Hendrik_2000
- Start date
- Status
They are very close to getting the tools for EUV.
Involves increasing the precision of a tool.
Due to vulnerability to US sanctions China really has no choice than to develop advanced semiconductor and IC chip manufacturing equipment. In the previous prevailing paradigm Chines companies largely refused to give market to Chinese made lithographic equipment. Now, they must.
None of this is correct because semiconductor equipment is not a competitive marketplace either, it is literally filled with legal collision.
There is no test "every day" either because these equipments have 10+ year lifetimes and contracts are signed with 3-6 month lead times.
Last edited:
10 years gap is pre-national security crisis. Expect it to narrow significantly now that is a national security threat by US, and China is putting it's best brains, best private companies, and nearly unlimited funds at Manhattan-project level urgency. A breakthrough will happen, just takes time. US has done China a favor in the long-term.
Seems like we have to go back to this point a lot but the only really technological limitation for Chinese EUV is the light generation mechanism. To be fair this has also always been the technical hurdle for everyone else who can do advanced lithography. The nanometer precision mirrors are also hard but most countries with advanced machining capabilities can handle that component pretty readily.
What's special about this light source? Ultra-short wave laser?
Depending on your optical system you can actually bypass most machining for grazing incidence optics using multilayer mirrors like the Mo/Si mirror I mentioned. You don't even need glass, just an atomically flat wafer substrate.
All EUV instruments need to use reflective optics made with materials like MoSi rather than refractive optics, because EUV photons are so high energy they get absorbed by typical refractive materials. This effectively means you must have the capability to make atomically flat wafer substrates, but it does *not* mean you don't need nanometer precision machining. Generally in practice the vapor deposition process you use to grow your substrate is not going to have the level of finish you need for effective reflection of EUV photons. This is because the EUV photons you illuminate on the mirror, by function of having wavelengths around ~10nm, will be absorbed or misdirected if your surface finish has a root mean square roughness that is the same or larger size as the wavelength of your photons. You need an RMS roughness that is smaller than the size of your emissions wavelength. The finer your roughness the better, as the smaller your surface features on your mirror the less photon loss you get from reflection and the higher your reflection efficiency, which then lowers the output requirements on your light source. The effective goal is to deliver as many photons from generation to wafer as you can.
EUV photons are basically soft x-rays, which means for the same amount of power input you generate fewer photons as output (more energy per unit of photons) compared to longer wavelength light sources. The total quantity of photons emitted, not the total power output of the light, is what ultimately determines your resolving power for photochemical exposure of the wafer and your wafer scanning speed. There are relatively straightforward methods for generating a small quantity of photons at such short wavelengths, but for effective patterning at industrial scales and speed you need mechanisms with much higher generation of photon emissions than those straightforward methods can deliver. The most well developed approach for generating that quantity of EUV photons has been through laser excited plasma emissions from materials like Tin or Xenon, which involves some pretty challenging physics and engineering. Even if you can build a light source that generates sufficient photon output, the extreme physics involved makes developing a machine that is reliable enough to be used for industrial production yet another set of challenges. And then to buttress all that, because of how expensive the engineering for this stuff is, you need a light source powerful enough to let you scan wafers at a fast enough rate to be economically viable for industry.
Actually I would disagree with you -- at this stage, given the level of advancement of China's ability to fabricate semiconductors, the "customers" that one might hear about for the processes they can produce for are not "mainstream" or obvious enough for English language media to report on them or recognize them. That is also affected by the fact that they are only recently getting a move on with domestic production lines (and will naturally see a bit of gap between receiving a product and having it enter large scale commercial production).
All of which underlies my point -- which is asking "just how accurate/useful are most English language reports/videos/news articles on the Chinese semiconductor industry"?
As others have mentioned, the Chinese marketplace for semiconductors is not a free market given US technological blockade efforts.
Furthermore, the overall discussion around achieving processes and various DUVL, EUVL machines, to my understanding, is all made with the appreciation that there obviously is a delay between attaining a given threshold and commercializing it.
- Status