Chinese semiconductor industry

Status
Not open for further replies.

caudaceus

Senior Member
Registered Member
I really don't care what they have to say. China has the authority to deny the expansion and they aren't denying it. There's near infinite demand for 28nm wafers in China and an additional 20kwpm from TSMC isn't going to significantly alter market dynamics or make entry prohibitively expensive for SMIC. And FDI ends up improving technical knowledge
Can Chinese government "suggest" TSMC to use indigenous equipment for manufacturing 28nm wafers?
 

gelgoog

Lieutenant General
Registered Member
...
This lithography machine is 193nm ArF immersion deep UV lithography machine, which can be used for 28nm chip production. Multi-exposure can produce 14nm, 7nm, which can meet most of the chip manufacturing needs.
...
the prototype is ready to be put into mass-production. prototype is in a commercial state, will not be placed in the corner to eat dust.

Initial yield during the beginning testing stages is lower, about 30%. Time waits for no one, so production companies should be subsidized. Low yield rate can be improved in production. If production is established, improvement is not a problem.

Thus, the main equipment and materials used for 28nm chip production have been localized. Before the end of this year, all-domestic 28nm production lines could be realized, and testing/development of 14nm and 7nm production may be carried out next year.
...
The development of EUV lithography machine a military directive, and is expected to be off the assembly line by 2023, the progress is apparently greatly ahead of schedule. By 2025, 80% of the chip localization may be realized."

There are way too many details in that story for it not to be true. For what is worth @HybridHypothesis 30% yield is really low. It is usually considered a fail. When the initial production gets to 50-60% that is considered good. In mass production the yield should be like 80% or more.

I think this is good news but 30% yield will not be commercially competitive. For a company like Huawei which cannot get any chips otherwise and who need limited production, they can make this work for them. For general purpose the yields will need to increase. 30% is good enough for limited production, like up to hundreds of thousands of chips at best, but the per chip costs will be high.

I hope they get top people in Chinese academia, industry, and technical staff at the semiconductor fabs working on fixing this. The Chinese government needs to setup a dedicated facility for the industry to collaborate on getting these details right.
 
D

Deleted member 15887

Guest
There are way too many details in that story for it not to be true. For what is worth @HybridHypothesis 30% yield is really low. It is usually considered a fail. When the initial production gets to 50-60% that is considered good. In mass production the yield should be like 80% or more.

I think this is good news but 30% yield will not be commercially competitive. For a company like Huawei which cannot get any chips otherwise and who need limited production, they can make this work for them. For general purpose the yields will need to increase. 30% is good enough for limited production, like up to hundreds of thousands of chips at best, but the per chip costs will be high.

I hope they get top people in Chinese academia, industry, and technical staff at the semiconductor fabs working on fixing this. The Chinese government needs to setup a dedicated facility for the industry to collaborate on getting these details right.
Having said that, the author of this story is "Big Entertainment Family", which apparently has the reputation of being a gossip and rumor publisher, so it not greatly credible. That's why I would appreciate if @Oldschool or @WTAN could provide verification for this new (such as a cjdby post or whatever)
 

gelgoog

Lieutenant General
Registered Member
Having said that, the author of this story is "Big Entertainment Family", which apparently has the reputation of being a gossip and rumor publisher, so it not greatly credible. That's why I would appreciate if @Oldschool or @WTAN could provide verification for this new (such as a cjdby post or whatever)

Well whoever wrote it seems to have a proper understanding or at least they selected quotes which make sense for the most part.
Saying that the machine can produce chips at 30% yield is relative anyway. 30% yield of what? Is it a highly regular thing like an SRAM chip (which is what is typically tried out first) or a more complex logic chip? What is the size of the chip and the diameter of the wafers? Since this is supposed to be used for 28nm production I assume 300mm wafers. But there is no way to know really.
If you go below 28nm then you need to use multiple patterning. In other words you need to do multiple passes to manufacture a chip. If your yield is crap with a single pass you can guess what will happen if you need to do multiple passes to make a chip.

What I think might be more likely than being wrong is that some or all the information might be out of date by the time it was posted.
 
D

Deleted member 15887

Guest
Well whoever wrote it seems to have a proper understanding or at least they selected quotes which make sense for the most part.
Saying that the machine can produce chips at 30% yield is relative anyway. 30% yield of what? Is it a highly regular thing like an SRAM chip (which is what is typically tried out first) or a more complex logic chip? What is the size of the chip and the diameter of the wafers? Since this is supposed to be used for 28nm production I assume 300mm wafers. But there is no way to know really.
If you go below 28nm then you need to use multiple patterning. In other words you need to do multiple passes to manufacture a chip. If your yield is crap with a single pass you can guess what will happen if you need to do multiple passes to make a chip.

What I think might be more likely than being wrong is that some or all the information might be out of date by the time it was posted.
If only someone like havok posted on here, we could gain a better insight
 

gelgoog

Lieutenant General
Registered Member
It seems China can manufacture 300 mm silicon wafers using domestic equipment already. I found this today:
...

That article says a lot of things but the most important things for commercial viability, like the speed they can grow the crystal, are not there. For what it's worth I remember a similar article several months ago.

Is that due to normal under-investment from companies or is there some bottleneck that China still depends on US IP to build more older processes Fabs?

If we are talking about manufacturing at 55nm none of that is sanctioned. So the right answer to your question is yes, it is under-investment. I doubt when they say 55nm that it really is 55nm, more likely it is 55nm or lower. Because 55nm is the best you can manufacture with 200mm wafers semiconductor fabs with ArF dry lithography tools.
 

Bellum_Romanum

Brigadier
Registered Member
I mean they can find ways to it man. Its not strange.

Just say that is a "national security issue" and the R&D centers will go poof. Put some propaganda about "stealing IP" and thats it.

The US does this like every week with sanctioning Chinese companies
Canada ain't the U.S. even if it's a vassal of the U.S. I know because I lived in Canada for 20 years was a member of the current ruling party because of Pierre Trudeau his son is an idiot.
 

Bellum_Romanum

Brigadier
Registered Member
Small countries are different.

Those 900 well paying jobs at Huawei are important. That is is the extent of Canadian leadership in 5G!

No one is going to do what Huawei is doing for Canada. That is kind of end of story. Unless Canada does not want these high tech research jobs.

That is why the white Canadian is so butt hurt over this. They want the 900 high tech jobs, but rather prefer it not being that Huawei is the provider of those jobs.

That is a small country. The strong do what they can. The weak suffer what they must. And that is the butt hurt.

:p
Exactly. It ain't that simple. If the world and the U.S. was this superbeating team it wouldn't be scrambling everything in it's power to disrupt, deter, and ultimately try to destroy Chinese inevitable technological superiority. You can't fight historical trend. China ain't Japan (which is a p..yfied shell of it's former war criminal self) and China ain't Jai Hind India either that does nothing but brags about it's non existent greatness 24 hrs 7 days a week, 365 days a year. For sure China will face immense challenges and slow down with it's current all encompassing struggle with the U.S. and his merry band of court jesters but given it's history and how far the country has come along these 40 years led by the CPC I have no doubt they will achieve the desired results it's aiming for. But if they don't and the failure somehow brings about the collapse of China then so be it. CPC will become just another set of rulers that will have to be set aside for another system since the failure(s) must be shouldered by them and no one else.
 

Bellum_Romanum

Brigadier
Registered Member
Because it is the essence of China. Losing yourself to win is losing in the end. Plus, China and the US have different advantages and styles. America has far less people per capital of resource and its people are lazy, not driven to STEM majors so its natural play would be to draw talent from all over the world with an easier lifestyle. The flaw with this strategy is that they are dependent on imported talent and imported talent is far more likely to sell out. China has 4x the population with a huge emphasis on STEM (more than the entire developed world combined), to the point where China is supersaturated. Bringing in foreign talent (by the droves as opposed to on special occasions from need) will sometimes serve to displace Chinese scientists and engineers. Chinese people trained in China are obviously more loyal on average than some foreigner to either China or the US so that is another advantage to covet over America. So basically, different countries with different demographics call for different strategies and also, China is China, not giant international bootleg USA. Just like in politics and economics, China achieves its victories by being uniquely China, not by copying America.

Moores law been dead for a long while now...

If push comes to shove China just needs to lodge a missile at TSMC and entire globe will be stuck at 5nm for the next decade
If the aim is to reunite back with Taiwan then your suggestion of bombing Taiwan even if it's "just their crown tech jewel" will not bring about that eventuality. Let's not be hasty, and too emotional when discussing pertinent issues for China’s future.
 
Last edited:
Status
Not open for further replies.
Top