Chinese semiconductor thread II

[tokenanalyst]

Tianjue Technology released the first 40nm bright-field nano-pattern wafer defect inspection equipment in China, opening a new era of domestic high-end semiconductor inspection equipment​

his is another new phase of progress made by Tianjue Technology less than a year after it officially delivered the TB1000 , a wide-band bright-field defect inspection equipment for 12 -inch wafer 65~90nm technology nodes in August last year.




The Importance of High-Precision Brightfield Defect Inspection Equipment

From a whole wafer to a single chip, in addition to the well-known photolithography machine, a series of necessary production equipment is also required, including diffusion furnaces, etchers, ion implanters, thin film deposition equipment, chemical mechanical polishers and cleaning machines.

Defect detection equipment is an important tool to ensure chip quality, reduce production costs, and promote process iteration, and is indispensable in the chip production process. In particular, as the process continues to evolve, the cost of manufacturing chips is getting higher and higher, and the importance of detection equipment is increasing day by day. Among them, bright field defect detection equipment with higher detection accuracy and more comprehensive defect type coverage is highly favored by the industry.

As a representative of domestic semiconductor equipment manufacturers, Silicon Semiconductor, founded in November 2021, has brought together top talents from well-known domestic and foreign semiconductor equipment companies, wafer foundries, listed companies and research institutions, focusing on the research and development, production and sales of high-end wafer defect detection equipment and components, and striving to fill the gap in the domestic defect detection equipment market . Relying on an excellent technical team and advanced technical strength, it has gradually broken the monopoly of foreign companies such as KLA on the defect detection market, injecting new vitality into the development of the domestic semiconductor industry.

TB1500 is the latest research and development achievement of Silicon Semiconductor. All key components are independently controllable, and advanced signal processing algorithms are used to effectively improve the signal-to-noise ratio and detection sensitivity . In order to meet the process requirements of the 40nm technology node, TB1500 has improved the brightness and sensitivity of the light source, increased the field of view and speed of the objective lens, and can capture smaller defect sizes.

Please, Log in or Register to view URLs content!

Please, Log in or Register to view URLs content!

​

 

[tokenanalyst]

Fuchuang Precision (Fortune Precision) intends to acquire 100% equity of Yisheng Precision for no more than RMB 800 million​

On July 15, 2024, Shenyang Fuchuang Precision Equipment Co., Ltd. (Fuchuang Precision, stock code: 688409) announced that it intends to acquire 100% of the equity of Beijing Yisheng Precision Semiconductor Co., Ltd. held by eight counterparties, including Zheng Guangwen, the actual controller of the company, Shenyang Advanced Manufacturing Technology Industry Co., Ltd., the company's largest shareholder, Beijing Yixin Enterprise Management Consulting Partnership (Limited Partnership), Liaoning Sino-German Industrial Equity Investment Fund Partnership (Limited Partnership), Ruan Yanfeng, Tianjin Xinsheng Enterprise Management Consulting Partnership (Limited Partnership), Zhongtai Fuli Technology Development Co., Ltd. and Liaoning Hesheng Zhongfu Equity Investment Fund Partnership (Limited Partnership), in cash. The transaction is undergoing an audit and evaluation, and the transaction amount has not yet been determined, but it is expected to be no more than 800 million yuan. After the transaction is completed, Yisheng Precision will become a wholly-owned subsidiary of Fuchuang Precision.



Yisheng Precision focuses on the mainstream domestic 12-inch wafer factory customers and can provide maintenance, circulation cleaning and coating regeneration services for non-metallic parts and consumables based on silicon, silicon carbide, quartz, metal parts and consumables based on metal materials such as aluminum, and core components of wafer factories. Some of Yisheng Precision's products have passed the advanced process technology certification of mainstream domestic 12-inch wafer factory customers and have achieved mass production and shipment.

Please, Log in or Register to view URLs content!

 

[Wahid145]

Saw a Post saying the Kirin9100 will have a Transistor Density of 140MTr/mm2. We don't know for sure yet obviously but it's a speculation to benchmark SMIC 5nm for now. For Reference, TSMC and Samsung 5nm has a Transistor Density of 138MTr/mm2 and 127MTr/mm2 respectively.
I can't seem to find a proper source of Kirin9010/Kirin9000s Transistor Density. Some source says its 98MTr/mm2, some goes for 114MTr/mm2 or 111MTr/mm2 or even 108MTr/mm2. Anyone here has a verified source of Kirin9010/9000s Transistor Density?

 

[Hyper]

Saw a Post saying the Kirin9100 will have a Transistor Density of 140MTr/mm2. We don't know for sure yet obviously but it's a speculation to benchmark SMIC 5nm for now. For Reference, TSMC and Samsung 5nm has a Transistor Density of 138MTr/mm2 and 127MTr/mm2 respectively.
I can't seem to find a proper source of Kirin9010/Kirin9000s Transistor Density. Some source says its 98MTr/mm2, some goes for 114MTr/mm2 or 111MTr/mm2 or even 108MTr/mm2. Anyone here has a verified source of Kirin9010/9000s Transistor Density?

Well someone can pay for the TechInsights report.

 

[Hyper]

Saw a Post saying the Kirin9100 will have a Transistor Density of 140MTr/mm2. We don't know for sure yet obviously but it's a speculation to benchmark SMIC 5nm for now. For Reference, TSMC and Samsung 5nm has a Transistor Density of 138MTr/mm2 and 127MTr/mm2 respectively.
I can't seem to find a proper source of Kirin9010/Kirin9000s Transistor Density. Some source says its 98MTr/mm2, some goes for 114MTr/mm2 or 111MTr/mm2 or even 108MTr/mm2. Anyone here has a verified source of Kirin9010/9000s Transistor Density?

To be honest transistor density does not matter. What matters is product performance and benchmarks and user experience and reviews. Nobody busy a chip because it's on such and so node. Performance and reliability is all that matters.

 

[Phead128]

To be honest transistor density does not matter. What matters is product performance and benchmarks and user experience and reviews. Nobody busy a chip because it's on such and so node. Performance and reliability is all that matters.

There is a direct correlation between transistor density and product performance/benchmark. Absent a physical sample of Mate 70 pro/Kirin 9100 until it's release in September, then discussion of transistor density is very much appropriate as that is the best proxy for product performance compared to extant TSMC/Samsung chips.

 

[JPaladin32]

Saw a Post saying the Kirin9100 will have a Transistor Density of 140MTr/mm2. We don't know for sure yet obviously but it's a speculation to benchmark SMIC 5nm for now. For Reference, TSMC and Samsung 5nm has a Transistor Density of 138MTr/mm2 and 127MTr/mm2 respectively.
I can't seem to find a proper source of Kirin9010/Kirin9000s Transistor Density. Some source says its 98MTr/mm2, some goes for 114MTr/mm2 or 111MTr/mm2 or even 108MTr/mm2. Anyone here has a verified source of Kirin9010/9000s Transistor Density?

Transistor density is not a single number. High Performance (HP) library and High Density (HD) library are different. HD is the most dense you can achieve, but such a high density may result in not enough current or too much heat (the reality is much more complicated than what I said here), so you often have a mix of HD and HP on a single die. HP is often used for the performance cores while HD is for effciency cores, for example.

Here are some numbers I collected in my notes over the years in various sources. I may or may not remember where I got them, but at least a couple of them are from Wikipedia and its linked references, and definitely some are from Bilibili. If I have time I may come up with a spreadsheet with sources, but I can't promise anything. People need to take these numbers with caution.

MTr/mm2 HP HD
TSMC 5nm: 92 137
TSMC 4nm: 97 146
TSMC N3E: ? 189
Samsung SF4: ? 137
Intel 4: - 124 (Intel for whatever reason has only one density library for Intel 4)
SMIC 7nm: 82 116

The density numbers you mentioned for Kirin 9000s are likely all correct, depending on which part of the die the density is measured from.

If the rumored 140 MTr/mm2 of the next Kirin chip ends up being true, then it is likely from the HD library and positions itself at about 4nm, even surpassing Intel, or maybe even Samsung (but Samsung GAA has been found in some smart watches, so probably not once its GAA goes into mass production). Also, if this is true, then the scary thing is that all the top foundaries are now entirely in East Asia.

To me, 140 MTr/mm2 feels too good to be true, but who knows?

 

[JPaladin32]

To be honest transistor density does not matter. What matters is product performance and benchmarks and user experience and reviews. Nobody busy a chip because it's on such and so node. Performance and reliability is all that matters.

Actually, density itself may not matter directly, but unlike the old days, these days CPU performance and energy efficiency almost come entirely from having more transistors instead of having better transistors, which is especially true for energy efficiency.

One biggest factor is the size of CPU caches. The energy consumption of data transfer from caches is often one order of magnitude lower than from main memory. Apple is the obvious example. Its remarkably impressive CPU energy efficiency is largely due to having a unified 16MB L2 cache for the two performance cores, which is absolutely insane. The L2 cache for other CPUs like Snapdragon 8 Gen 2 is only 1MB, and the L2 is private for each core, not shared. To give you an idea of how insanely big the cache is, the cache die area of the performance cores in Apple A16 and A17 is even larger than the cores themselves. Qualcomm's latest Snapdragon 8 Gen 3 chip also got 20% bigger to fit in bigger cores and more caches. Without the higher density to squeeze in more transistors or without the die area increase, I personally don't think we could see more than 10% improvement in CPU performance and efficiency in the last 4 years (not 10% each year, but 10% across 4 entire years).

There are other factors like having dedicated circuit like vector units, CPU crypto instructions for higher performance and better efficiency, but again, they come from having more transistor budget. From the data I've seen we are sadly in an age where each transistor no longer improves by much each year. GAA and CFET may change the story, but I'm not sure, but at the moment, having better transistor density to fit in more stuff is the indirect and only path towards better CPUs.

 

[tokenanalyst]

This is a pretty important company with pretty important clients.​

Swedish company acquires vacuum helium leak detector manufacturer Anhui Nuoyi Technology​

On July 10, Atlas Copco announced that it had signed an acquisition agreement with Anhui Noyi Technology Co., Ltd., a manufacturer of vacuum helium leak detectors. Anhui Noyi Technology will become part of the Scientific Vacuum Division of Atlas Copco's Vacuum Technology business area. Anhui Nuoyi Technology is headquartered in Hefei, Anhui Province, has 78 employees and had revenue of approximately US$16.8 million in 2023. Geert Follen, president of Atlas Copco's Vacuum Technique business area, said the acquisition will strengthen Atlas Copco's core in-house helium leak detector technology. The acquisition is expected to be completed in the second half of 2024, and the purchase price has not been disclosed.

Anhui Nuoyi Technology Co., Ltd. is a high-tech enterprise with a registered capital of 32.2 million yuan and a single-building factory. It specializes in the research and development, production and sales of helium mass spectrometer leak detectors , helium leak detection vacuum boxes and recovery systems, airtightness leak detectors, and airtightness leak detection non-standard systems. It has reached strategic cooperation with many scientific research institutes in the field of aerospace in China, and obtained the agency authorization of the first-level vacuum measurement station of the National Defense Science and Technology Industry.
The leak detectors have been exported to Russia, South Korea, India, Thailand, the Philippines and many other countries, and have been highly praised. The company's products are widely used in aerospace, refrigeration and related supporting, automobile manufacturing, power batteries, nuclear industry, vacuum systems, biochemistry, environmental protection, food fermentation, petroleum processing, organic chemistry, Chinese and Western medicine, scientific research and other fields. Since its establishment, it has strictly implemented relevant national standards and industry standards for products. It is committed to the research of high quality, high performance and high efficiency of precision instruments, and applies modern business concepts and management methods. With strong technical support and advanced production and testing equipment, it carefully manufactures NOY brand precision instruments with good stability, high precision, high efficiency and complete functions. With the corporate spirit of "being positive and optimistic, courageous to take responsibility, not making promises lightly, and striving for excellence", it has developed step by step into a company with a wide range of business and considerable influence in the industry among similar companies in China.

Please, Log in or Register to view URLs content!

Please, Log in or Register to view URLs content!

​

 

[tphuang]

Saw a Post saying the Kirin9100 will have a Transistor Density of 140MTr/mm2. We don't know for sure yet obviously but it's a speculation to benchmark SMIC 5nm for now. For Reference, TSMC and Samsung 5nm has a Transistor Density of 138MTr/mm2 and 127MTr/mm2 respectively.
I can't seem to find a proper source of Kirin9010/Kirin9000s Transistor Density. Some source says its 98MTr/mm2, some goes for 114MTr/mm2 or 111MTr/mm2 or even 108MTr/mm2. Anyone here has a verified source of Kirin9010/9000s Transistor Density?

I would ignore that post.

9100 is using same process as 9010s