Chinese semiconductor industry

Status
Not open for further replies.

ZeEa5KPul

Colonel
Registered Member
It's an inverse problem.

You can't extract information from the results directly like you can with X-ray or electron, which directly from images or spectra that can be directly measured and interpreted. It is literally just like looking at it with your eyes, just enhanced a bit by your instrument.

For optical CD, the optical scattering patterns by itself is meaningless, since you physically cannot extract sub wavelength imaging information directly. Instead there will be hints, but just hints, in the scattering patterns. You have to theoretically calculate the scattering patterns with N free parameters, and see if they fit your model.

Please, Log in or Register to view URLs content!

Please, Log in or Register to view URLs content!

My brain has trouble with trusting inverse problems as there's always the doubt of "but what if the model is wrong/incomplete".

Imaging and spectral data is different. You are directly observing the property of interest. It is model free. The CD is directly observed.

It is like the difference between directly observing a crime happen and piecing a crime together with a forensic team. There's some cases that are airtight, some cases confuse for decades, and sometimes bias puts the wrong guy in jail. But direct observation always works.
In what kinds of problems is optical metrology absolutely necessary? Is it irreplaceable in some domains?
The Jim Reaper.
 

FairAndUnbiased

Brigadier
Registered Member
In what kinds of problems is optical metrology absolutely necessary? Is it irreplaceable in some domains?

The Jim Reaper.
It's way faster than anything electron based, because it works in air and can measure the whole wafer at once. Electron based methods are vacuum only and are small area imaging. But they are the gold standard as it is model free.

X-ray is used for spectroscopy like in XRF or XPS for wafer chemical mapping. It's not used for CD until recently. There is a recent method (within 5 years) to apply small angle X-ray scattering (SAXS) usually used for crystallography to 3D NAND CD.

But there's guys smarter than me working on this stuff.
 

ZeEa5KPul

Colonel
Registered Member
It's way faster than anything electron based, because it works in air and can measure the whole wafer at once. Electron based methods are vacuum only and are small area imaging. But they are the gold standard as it is model free.
It sounds to me like a good problem to sprinkle some AI on. The AI can learn the model from ground truth data provided by electron microscopy and other reliable methods.
 

tonyget

Senior Member
Registered Member
Please, Log in or Register to view URLs content!
Well I guess the news of YMTC being able to resume production has now reached Western media.

I think their anonymous sources timeline of latter in 2024 is a little pessimistic. I would guess start of 2024, but let's wait and see.

It's actually easier to tell YMTC's production status than other firms,because YMTC products are available on consumer market. As long as that availability doesn't change,they are likely okay

Also, so much coping at the end here. Samsung & SK have barely started 232 layer production. Most of their production is 128-layer. YMTC expanding 128 layer is perfectly fine. Based on the current rate of AMEC improvement, YMTC will be able to do 232 layer by 2025

Wait,YMTC's 232 layer products are selling right now,and there is no sign of running out of stock. Doesn't that mean YMTC is able to sustain 232 layer production at the moment?
 
Last edited:

tphuang

Lieutenant General
Staff member
Super Moderator
VIP Professional
Registered Member
It's actually easier to tell YMTC's production status than other firms,because YMTC products are available on consumer market. As long as that availability doesn't change,they are likely okay



Wait,YMTC's 232 layer products are selling right now,and there is no sign of running out of stock. Doesn't that mean YMTC is able to sustain 232 layer production at the moment?
I assume they can keep 232 layer production going, but any new capacity they bring online will be 128 layers. Or until whenever AMEC develops something that can do 232 layers.

YMTC's revenue was said to be 5% of global revenue according to that link. I think @hvpc's table show them to be less than that (especially in Q4). Either way, once fab 2 is ramped up, they can probably reach 10% of global revenue. Fully capable of replacing Micron in the Chinese market. It was really dumb of Biden admin to make memory chips into a national security issue.

We can see with AMEC's progress, they are far more focused on developing tools for 7 to 14nm process than for memory chips.
 

latenlazy

Brigadier
It's an inverse problem.

You can't extract information from the results directly like you can with X-ray or electron, which directly from images or spectra that can be directly measured and interpreted. It is literally just like looking at it with your eyes, just enhanced a bit by your instrument.

For optical CD, the optical scattering patterns by itself is meaningless, since you physically cannot extract sub wavelength imaging information directly. Instead there will be hints, but just hints, in the scattering patterns. You have to theoretically calculate the scattering patterns with N free parameters, and see if they fit your model.

Please, Log in or Register to view URLs content!

Please, Log in or Register to view URLs content!

My brain has trouble with trusting inverse problems as there's always the doubt of "but what if the model is wrong/incomplete".

Imaging and spectral data is different. You are directly observing the property of interest. It is model free. The CD is directly observed.

It is like the difference between directly observing a crime happen and piecing a crime together with a forensic team. There's some cases that are airtight, some cases confuse for decades, and sometimes bias puts the wrong guy in jail. But direct observation always works.
Think of it like detecting stealth planes with indirect signal patterns.
 

olalavn

Senior Member
Registered Member
In order to ensure the security of the key information infrastructure supply chain, prevent network security risks caused by hidden product problems, and maintain national security, according to the "National Security Law of the People's Republic of China" and "The People's Republic of China
According to the Cyber Security Law of the Republic of China, the Cyber Security Review Office conducts a cyber security review of products sold by Micron in China in accordance with the Cyber Security Review Measures.




006DKHfbgy1hcjepdef72j31fs0u0n1z.jpg
 

gelgoog

Lieutenant General
Registered Member
In order to ensure the security of the key information infrastructure supply chain, prevent network security risks caused by hidden product problems, and maintain national security, according to the "National Security Law of the People's Republic of China" and "The People's Republic of China
According to the Cyber Security Law of the Republic of China, the Cyber Security Review Office conducts a cyber security review of products sold by Micron in China in accordance with the Cyber Security Review Measures.
About damn time. Long overdue. They should do the same thing to Qualcomm.
 
Status
Not open for further replies.
Top