Chinese semiconductor industry
- Thread starter Hendrik_2000
- Start date
- Status
Interesting comments in there about how long it takes to go from R&D to production for 3 critical materials (photoresist/CMP polishing solution/KOH developer) It looks like R&D takes at least a year to 2. Verification takes a minimum of 2 and ramping up production takes another 1 to 2 years. The verification for KOH developer with Grenada Chem seems to be extremely long. They've been verifying for 2 years and still cannot achieve production. Must verify with photoresist.
Also other interesting part from that piece.
It's hard to get verified for 14nm process because they can only work with SMIC and SMIC's 14nm production is all taken so there just aren't a lot of opportunity to validate new domestic material.
So if we are wondering why it takes such a long time to have full domestic process, tools verification or even lithography machine is not the components holding things back. Seems like materials are.
My translation: SMIC’s 14nm production line (only one in China) is fully utilized so there is no idle machines available to test the new materials.
They need to innovate a more accelerated verification process.
They need to buy a secondhand lithography machine like other photoresist companies.
35,000 * 12 * 500 = 210,000,000 CPUs
Or they need new 14 nm line expansions from one of the fabs.
So how to break this vicious circle？SMIC's 14nm will always be running at full capacity in the foreseeable
Get the US to sanction photoresists
Well, no better time than to do it now when demand is down somewhat. It's also unclear how much advanced process capacity they do have right? They have to find tools somehow given that AMAT/Lam haven't been able to sell to SMSC for a while. Maybe they find some on the secondary market or qualify new people. But give it another year or two this will be less of an issue.
At some point, they just need to suck it up and do it.
It does make me wonder though, I would think the photoresist for 28nm should be the photoresist for 14/7nm process. Any reason why KOH developer would be different? Or is Greenada just making an excuse here? Why can't their product be verified on 28nm or even HLMC's 22nm process.
In other news.
Phytium CPUs continue to win orders from major SOEs for critical infrastructure. It's too bad Loongson did not develop LoongArch earlier when it was obvious MIPS was at a dead end. They have a better product than Phytium, but missed out on about 1 to 2 years of ecosystem buildup and domestic procurement drive.
- Status