Chinese semiconductor industry

Status
Not open for further replies.

olalavn

Senior Member
Registered Member
Bro, you are misleading everyone by trivializing the complexity behind building a production worthy lithography system.

What you said may be true for an engineering or alpha tool. Conquering the major modules that everyone is familiar with from plethora of ASML marketing slide is hardly enough to get the system into high volume manufacturing.

The devil is in the details. Details, which you called "passive" components, that ASML does not reveal to the public. Having worked with ASML to bring several scanners from development to production, I can tell you some of the subtle nuisances, functionalities is what differentiate ASML from Nikon & Canon.

We are no where near releasing a HVM ready domestic immersion or EUV system.

Even the wafer production ecosystem required to support EUV lithography is still a problem for China. Even if Chinese fabs have access to EUV system, the domestic EUV reticle process flow is no where near ready. And from what I see, there has been very little activity on this front domestically. Besides EUV photoresist, EUV reticle is also a problem.
I guess 1-2 years later Shanghai Xinyang will finish EUV photoresist research
Luwei Optoelectronics will take care of the EUV mask, in addition, Changchun Institute of Optics and Mechanics has most of the EUV technologies from masks to light sources.
 

FairAndUnbiased

Brigadier
Registered Member
Bro, you are misleading everyone by trivializing the complexity behind building a production worthy lithography system.

What you said may be true for an engineering or alpha tool. Conquering the major modules that everyone is familiar with from plethora of ASML marketing slide is hardly enough to get the system into high volume manufacturing.

The devil is in the details. Details, which you called "passive" components, that ASML does not reveal to the public. Having worked with ASML to bring several scanners from development to production, I can tell you some of the subtle nuisances, functionalities is what differentiate ASML from Nikon & Canon.

We are no where near releasing a HVM ready domestic immersion or EUV system.

Even the wafer production ecosystem required to support EUV lithography is still a problem for China. Even if Chinese fabs have access to EUV system, the domestic EUV reticle process flow is no where near ready. And from what I see, there has been very little activity on this front domestically. Besides EUV photoresist, EUV reticle is also a problem.
you can't get EUV reticle process flow to work, without actually having the EUV system and the EUV photoresist. all process talk is putting cart before horse.

anyhow, I'm not an expert on EUV systems or production, but I do know thin films and deposition. maybe you can educate me on what the challenges of a multilayer mirror is if I'm missing something that the papers I cited did not mention.
 

hans_r

New Member
Registered Member
Bro, you are misleading everyone by trivializing the complexity behind building a production worthy lithography system.

What you said may be true for an engineering or alpha tool. Conquering the major modules that everyone is familiar with from plethora of ASML marketing slide is hardly enough to get the system into high volume manufacturing.

The devil is in the details. Details, which you called "passive" components, that ASML does not reveal to the public. Having worked with ASML to bring several scanners from development to production, I can tell you some of the subtle nuisances, functionalities is what differentiate ASML from Nikon & Canon.

We are no where near releasing a HVM ready domestic immersion or EUV system.

Even the wafer production ecosystem required to support EUV lithography is still a problem for China. Even if Chinese fabs have access to EUV system, the domestic EUV reticle process flow is no where near ready. And from what I see, there has been very little activity on this front domestically. Besides EUV photoresist, EUV reticle is also a problem.
Yes, this. A lot of the SMEE ArFi, does it exist/does it not, seems to be a ton of rumors - including at a hypothetical SMIC research center in Chengdu (even though SMIC's website has 0 open positions in Chengdu) - and thinly sourced references without clear performance metrics that this forum cycles through fairly regularly without any real clear resolution
 

ansy1968

Brigadier
Registered Member
Yes, this. A lot of the SMEE ArFi, does it exist/does it not, seems to be a ton of rumors - including at a hypothetical SMIC research center in Chengdu (even though SMIC's website has 0 open positions in Chengdu) - and thinly sourced references without clear performance metrics that this forum cycles through fairly regularly without any real clear resolution
Then put on your thinking cap and think again, who are the main players pursuing such research? I gave you a hint, most of them are proud members on the US entity list. :) IF you make the list you made it.;)
 

hvpc

Junior Member
Registered Member
you can't get EUV reticle process flow to work, without actually having the EUV system and the EUV photoresist. all process talk is putting cart before horse.
Come on. We can access EUV systems through IMEC. SMIC and CXMT had done so in the years past. If someone really want to work on EUV reticle process flow, they do have access to the system.

Regardless if explanation/excuse you gave is valid, this still means it will take more time for China to produce wafers with EUV long after the domestic EUV scanners become available.
anyhow, I'm not an expert on EUV systems or production, but I do know thin films and deposition. maybe you can educate me on what the challenges of a multilayer mirror is if I'm missing something that the papers I cited did not mention.
I respect your expertise and would not presume to educate you the multilayer mirrors.

I however, had been in charge of the R&D, integration, and production of litho/process that involved qualification/optimization of scanners for production. And the point I am making is there's still ways to go to reach domestic EUV industrialization after assembling all the "primary components" together. There's more to it than just having the hardware integrated.

Take for example, Nikon also have immersion system that uses the same Cymer/Gigaphoton lasers, have good hardwares, and even could match ASML specification under test conditions. So why does ASML have 90% of immersion market share while Nikon ships less than 10 systems per year? Like I said before, the devil is in the details beyond the hardware. So even if SMEE could come up with scanners that could match ASML's tool specification, they will most likely to be simply close to where Nikon is relative to ASML. Same logic applies to integrated domestic EUV system relative to ASML's HVM systems.
 

hans_r

New Member
Registered Member
Then put on your thinking cap and think again, who are the main players pursuing such research? I gave you a hint, most of them are proud members on the US entity list. :) IF you make the list you made it.;)
If everything is all part of a secret plan with decoys, that might indeed be unfalsifiable but at the same time, is fairly noncredulous.
In any case, the "Chengdu fab" that everyone references seems to be vaporware - SMIC is hiring exactly 0 people in Chengdu which makes me seriously question the veracity of the rest of the posting
 

olalavn

Senior Member
Registered Member
Yes, this. A lot of the SMEE ArFi, does it exist/does it not, seems to be a ton of rumors - including at a hypothetical SMIC research center in Chengdu (even though SMIC's website has 0 open positions in Chengdu) - and thinly sourced references without clear performance metrics that this forum cycles through fairly regularly without any real clear resolution
in this forum, people never post rumors or fake news like western media..
 

FairAndUnbiased

Brigadier
Registered Member
Come on. We can access EUV systems through IMEC. SMIC and CXMT had done so in the years past. If someone really want to work on EUV reticle process flow, they do have access to the system.

Regardless if explanation/excuse you gave if valid, this still means it will take more time for China to produce wafers with EUV long after the domestic EUV scanners become available.

I respect your expertise and would not presume to educate you the multilayer mirrors.

I however, had been in charge of the R&D, integration, and production of litho/process that involved qualification/optimization of scanners for production. And the point I am making is there's still ways to go to reach domestic EUV industrialization after assembling all the "primary components" together. There's more to it than just having the hardware integrated.

Take for example, Nikon also have immersion system that uses the same Cymer/Gigaphoton lasers, have good hardwares, and even could match ASML specification under test conditions. So why does ASML have 90% of immersion market share while Nikon ships less than 10 systems per year? Like I said before, the devil is in the details beyond the hardware. So even if SMEE could come up with scanners that could match ASML's tool specification, they will most likely to be simply close to where Nikon is relative to ASML. Same logic applies to integrated domestic EUV system relative to ASML's HVM systems.
The big difference with Nikon is that SMEE will have a captive market with government fabs, and then when they are used as the test customers to perfect applications of (first immersion ArF, then EUV later), that experience can be transferred. Nikon on the other hand had a customer in Intel, but due to shrinking Japanese chip industry, had no real captive market.

I also did not say that EUV was easy, just that among the components, the big problems are the light source and wafer stage, while the mirrors are comparatively easier than the light source and wafer stage. But only comparatively. And I only mention it because the prior post asked about mirrors. If they asked about software, I wouldn't say anything, since I don't know software.
 

hvpc

Junior Member
Registered Member
No. Even if it’s easy you need to test the replacements at the system level before you ship, which means you first need to do a business calculation of whether it’s worth the resources and time devoted to testing.

Which part are you referring to exactly?
: ) Sounds like you agree with me that it's not easy for ASML to replace US origin parts.

If I remember correctly, all ASML scanner's reticle handling, reticle stage modules, optical sensors, wafer stage positioning module, and even some optics are fabricated in their U.S. Connecticut factory. You could check out their website, I'm sure there's some introduction of their manufacturing sites around the world and what they each build.
 

hans_r

New Member
Registered Member
in this forum, people never post rumors or fake news like western media..
going to point out that the SMEE ArFi cannot be coming out in both 2021, 2022 and 2023 (all of which have been claimed by members on this forum). There's clearly rumors being posted here
 
Status
Not open for further replies.
Top