Chinese semiconductor industry

continue last post

深蓝：上微800系列的干式双工台版本光刻机，在产线调通了吗，可以大批量制造芯片了否？
havok：90nm差不多了，不是燕东4月要试产了么？90nm光刻机在燕东调试快有2年了，20年的时候就在那里了

王鑫：浸没式内部结构像不像ASML的浸没式？
havok：干式像 浸没就肯定像，90nm机和28nm机的内部结构没多大差别

深蓝：Has the 800 series dry-type dual-station version lithography machine of SMEE been adjusted in the production line, and can it manufacture chips in large quantities?
havok：90nm is almost ready, isn't Yandong going to trial production in April? The 90nm lithography machine has been debugging in Yandong for almost 2 years now, it has been there since 2020

王鑫：Is SMEE's immersive-type machine's internal structure similar to ASML's immersive-type?
havok：Yes, if the dry-type is similar than the immersive-type will be similar as well，the internal structure of 90nm machine and 28nm machine is not much different

havok just confirmed that the dual-station in the picture has been used on the new 90nm machine

havok：而且90nm光刻机就是用上面图片中的工件台

tonyget said:

havok：65nm双工件台及调试台
havok：65nm dual-working station and debug bench

View attachment 105755

Click to expand...

tonyget said:

continue last post

深蓝：上微800系列的干式双工台版本光刻机，在产线调通了吗，可以大批量制造芯片了否？
havok：90nm差不多了，不是燕东4月要试产了么？90nm光刻机在燕东调试快有2年了，20年的时候就在那里了

王鑫：浸没式内部结构像不像ASML的浸没式？
havok：干式像 浸没就肯定像，90nm机和28nm机的内部结构没多大差别

深蓝：Has the 800 series dry-type dual-station version lithography machine of SMEE been adjusted in the production line, and can it manufacture chips in large quantities?
havok：90nm is almost ready, isn't Yandong going to trial production in April? The 90nm lithography machine has been debugging in Yandong for almost 2 years now, it has been there since 2020

王鑫：Is SMEE's immersive-type machine's internal structure similar to ASML's immersive-type?
havok：Yes, if the dry-type is similar than the immersive-type will be similar as well，the internal structure of 90nm machine and 28nm machine is not much different

Click to expand...

This confirmed that Yandong bought 90nm lithography machine.

tonyget said:

havok：90nm is almost ready, isn't Yandong going to trial production in April? The 90nm lithography machine has been debugging in Yandong for almost 2 years now, it has been there since 2020

Click to expand...

Trial production in April means it can go in volume production by year end. In this case it will be 3 years from delivery to volume production.

It is not so long time considering that the machine included the main sub-modules are totally new and that the machine has been probably developed on-site. The next step, the immersion version, probably will have a shorter ramp-up, because, as havoc said, the internal structure of the 2 machines, included the U-Precision's dual stage, is very similar, and also, very important, the gained on-the-field experience and software improvements of the dry machine will be ported to the immersion machine.

If the immersion machine, now under internal testing/debugging in SMEE, will be shipped to customer within the next months, we can assume end of 2024 / beginning of 2025 for volume production....just in time for next US elections.

As a side note, if ASML immersion machines will be banned this year, development of the 28nm fabs, now under construction in China, could be delayed by about 18 / 24 months.

I'm not sure is a good deal for US to give up a clear market leadership in China, that would have lasted for at least the next 10 years, in exchange of a possibly 2 yearish slowdown in China's IC development.

ASML beats earnings forecasts, sees 2023 growth amid China worries​

Regarding ASML's exports to China CEO Peter Wennink said "We just have to wait for the governments and the politicians to keep talking and come to a reasonable solution"

"We can still ship DUV (older) ... tools" to mainland China, Wennink said.

european_guy said:

Trial production in April means it can go in volume production by year end. In this case it will be 3 years from delivery to volume production.

It is not so long time considering that the machine included the main sub-modules are totally new and that the machine has been probably developed on-site. The next step, the immersion version, probably will have a shorter ramp-up, because, as havoc said, the internal structure of the 2 machines, included the U-Precision's dual stage, is very similar, and also, very important, the gained on-the-field experience and software improvements of the dry machine will be ported to the immersion machine.

If the immersion machine, now under internal testing/debugging in SMEE, will be shipped to customer within the next months, we can assume end of 2024 / beginning of 2025 for volume production....just in time for next US elections.

As a side note, if ASML immersion machines will be banned this year, development of the 28nm fabs, now under construction in China, could be delayed by about 18 / 24 months.

I'm not sure is a good deal for US to give up a clear market leadership in China, that would have lasted for at least the next 10 years, in exchange of a possibly 2 yearish slowdown in China's IC development.

Click to expand...

Havok did mentioned that DUVi will be certified in production line by the middle of this year. I think a few units already delivered to customers.

On Yandong, I don't remember if this was previous posted, but they are finally expecting to start mass production of the 12 inch line this year and reach max production by 2025.
expected to start production in April and eventually reach 20k by July 2024 with first phase. 2nd phase start trial production in April 2024 and reach full production of 20k by July 2025.

燕东微表示，基于成套国产装备的特色工艺12英寸集成电路生产线项目的总投资为75亿元，公司将利用现有的净化厂房和已建成的厂务系统和设施进行局部适应性改造，并购置三百余台套设备，建设以国产装备为主的12英寸晶圆生产线。该产线完成建造后，月产能将达4万片，工艺节点为65nm，产品定位为高密度功率器件、显示驱动IC、电源管理IC、硅光芯片等。

Click to expand...

On Yandong's domestic tool involvement. There states a full set of domestic equipment for a specialty production line with investment of 7.5 billion RMB. The second part indicates that for the full plant, it uses mostly domestic tools and will reach for production of 40k wpm with focus on 65nm in power chips, display drive IC, power management IC and silicon photonics chips and such.

Also, I would caution people to blame all of these 2 years on SMEE DUV being trialed. The entire factory and production line is new for Yandong. There is no doubt that a lot of the time will be spent on fixing issues for other part of their production. This is their first 12-inch wafer plant and there are other domestic tools that are getting qualified.

I don't think it's reasonable to expect SMIC trialing DUVi to take as long because SMIC will already have experience producing 28-40nm wafers on 12-inch lines that are de-americanized.

My question overall is still the 55nm line at ICRD. If Yandong is already doing their own 65nm line and finished debugging it. Why does ICRD still need to do it. It seems like they are qualifying a more advanced variant for of SSA800 dry for 55nm.

Interesting. Loongson is coming out with more SoC. This 2K3000 is an 8 core (LA364) CPU along with its own integrated GPU LG200. According to this, 2K3000's LA364 went through redesign with different IPCs, redesigned processor core and lower bandwidth? Basically, the efficiency improved and the new core is at Cortex A75 level and the usage goes beyond industrial control. Since 2K3000 will be built with 14nm process rather than 28nm (like 2K2000), one can see why the LA364 went through improvement.

It's expected to be cheaper, with higher dimension, smaller surface area and lower power consumption than 2K2000. It's expected to be taped out in the first half of this year.

Above is a table of the different CPU and GPU core designed by LoongSon for different purposes. LG100/120 is their first generation low end GPU. LG200 seems to be second generation more powerful GPU to be placed on their SoCs and such.

Now, it says the main rationale for having 8 cores is to win in certain domestic bids that requires 8 core CPU (which 4-core CPUs like 3A5000/6000 don't qualify). But from what I can see, the low power consumption aspect allows this to be a SoC used on lower end smartphones. Or at least it is a step toward that. I would imagine this gets integrated with openHarmony and Huawei would have places to use it.

This one talked a little bit more about it along with other future Loongson plans
2K3000 is expected to have 3A5000 capabilities but at lower cost and power consumption. Making it I think more appropriate for IoT devices, more demanding industrial computers, lower end smartphone/tablet.

Keep in mind Kirin 810 that was launched in 2019 and made using 7nm process had 2 core of A76 performance along with 6 cores of A55 little core + an integrated GPU. If 14nm redesigned LA364 is at A75 performance, then 8 core LA364 with integrated GPU could very well have better performance than Kirin 810. It seems to me that if Huawei is forced to go 12/14nm route for 5g SoC, then something along the line of 2K3000 should be considered (or something similar to it with higher powered and lower powered core). But we will need to wait and see its power consumption performance.

Exclusive-Dutch officials headed to Washington to talk controls on chipmaking gear - sources​

(Reuters) - Dutch and U.S. officials will meet in Washington on Friday to discuss potential new controls on exporting semiconductor manufacturing gear to China, with a deal possible by the end of the month, according to two sources familiar with the matter.

mst said:

Exclusive-Dutch officials headed to Washington to talk controls on chipmaking gear - sources​

(Reuters) - Dutch and U.S. officials will meet in Washington on Friday to discuss potential new controls on exporting semiconductor manufacturing gear to China, with a deal possible by the end of the month, according to two sources familiar with the matter.

Please, Log in or Register to view URLs content!

Click to expand...

Great, another article that says absolutely nothing materially and tells us that any agreement won't be announced for some time. What's the point of keep posting these articles?