Chinese semiconductor thread II
- Thread starter vincent
- Start date
The guy was encouraged to say that by a random poster. And that random was a known doomer on the Chinese social media and had a history of posting weird things. For example, he had used fried chicken prices to reach conclusions about the US economy and the prices on the post were wrong too. He had zero days of experience or education in anything related to semiconductors. You would expect media workers to check the credibility of the resources they use and read on the topic they are writing on but these are apparently too much to expect from think-tankies.
Sounds like Brookings Institute material.
Sounds like pretty massive news. And the post has so many details it sounds credible. Here is a machine translation for the whole post:
Good news about domestic photoresist, key technology and application of 14-28nm process deep ultraviolet photoresist: compared with similar foreign products, the cost is reduced by 45%, forcing them to reduce prices. The overall cost of the integrated circuit industry has dropped by 30%, replacing 10% of foreign imports, and the chip yield has increased from 80% to 97%.
The first product to be replaced is the American photoresist
==========
This project is a key raw material in the integrated circuit manufacturing process-28nm process deep ultraviolet photoresist key technology research and application research. The products cover different fields of integrated circuit manufacturing and compound semiconductors, and are the key "neck" materials to be overcome in chip manufacturing.
The research and development and application of project products break the high technical barriers and market barriers of Japanese and American companies, realize the substitution of imported photoresists by domestic production, and strongly support the development of my country's key integrated circuit material industry.
With the support of Jiangsu Province's scientific and technological achievements transformation projects, this project has realized the key technologies and applications of 28nm process deep ultraviolet photoresist. The main innovations are as follows:
1. Process technology for preparing deep ultraviolet photoresist below 28nm and its related raw materials.
2. Replacement of Dow Chemical Company's deep ultraviolet photoresist EPIC and UV series products.
3. Two group standards for integrated circuits, T/ICMTIA4-2020 "ArF Photoresist Monomer Part 1 Liquid Methacrylates" and T/ICMTIA5-2020 "ArF Photoresist Monomer Part 1 Solid Methacrylates", have been formulated.
This project innovatively developed a full industry chain technology route integrating photoresist monomer-photoresist photosensitizer-photoresist resin-photoresist formula-photoresist preparation and production. By fully mastering the core technology of each link, four types of products were obtained: ArF photoresist hole type HTA112X series, groove type HTA116X series, KrF photoresist HTK10X series and KrF negative photoresist.
Technical indicators, monomer metal ion control 10ppb level, deep ultraviolet ArF photoresist meets the various indicators of N90/N55/N28 nodes.
ArF photoresist products include wet ArF photoresist hole HTA112X and trench HTA116X series, both of which are based on the independent mass production of high-purity low-gold impurity monomers, designed and optimized ArF photoresist resin structure, and developed ArF photoresists with excellent performance, breaking foreign restrictions on the import of semiconductor photoresists and their materials in my country, and realizing the localization of high-resolution ArF photoresists for logic devices at 65nm-14nm technology nodes.
The dry ArF technology capability reaches the process requirements of 120nm line width half-pitch at the N90 technology node and 90nm line width and half-pitch of through holes at the N55 technology node; the immersion ArF reaches the resolution capability of 50nm half-pitch of through holes at the N28 technology node.
The HTK10X series of KrF photoresist products are developed for the "28nm and below integrated circuit chip high-resolution photoresist development project". Through independent resin structure design and formula optimization, it successfully breaks the technical barriers of KrF photoresist for advanced processes and meets the process requirements of 160nm half-pitch line width at the N65 technology node and 120nm half-pitch line width at the N40 technology node and below.
The HTKN601-X series of KrF negative photoresists uses a completely independent raw material supply chain and is a photoresist used for the lift-off process of surface acoustic wave filters. Surface acoustic wave filters are the core components for signal transmission and reception in mobile communication systems. High-generation (3G, 4G and 5G) filters have long relied on imports and have been jointly identified by the National Development and Reform Commission and the Ministry of Industry and Information Technology as the second "device that affects national security". This product can be used in high-dose ion implantation lift-off processes and surface acoustic wave filter production processes. It is one of the key materials and currently achieves import substitution for such products.
The project products have high technological content, high added value and broad market prospects. Some products fill domestic gaps, solve the problem of stable batch production technology of raw materials and supporting material series, and fundamentally solve the domestic substitution of "stuck neck" materials from the entire supply chain. And it has independent intellectual property rights and has authorized a number of invention patents, truly achieving self-sufficiency and domestic photoresist that is not restricted by foreign countries. The implementation of the project has achieved good economic and social benefits. Compared with similar foreign products, the cost of the project products has been reduced by 45%, the gross profit margin has reached 55%, and the products have replaced 10% of foreign imports, reducing the overall cost of the domestic integrated circuit industry by 30%, and the chip yield has increased from 80% to 97%. The implementation of the project is of great significance to breaking through the gap, mastering core technologies, breaking the high-price monopoly of imported products, realizing the localization of the product, and enhancing the status and core competitiveness of my country's integrated circuit industry.
Achievement category: applied technology
Achievement level: domestic leading
Evaluation form: acceptance
Research start and end time: 2015-04～2019-06
Achievement storage time: 2023
The 7nm chips for Huawei, whose masks are they using? I thought US has mandated all suppliers cant supply anything below14nm to China. Same with ArFi photoresist.
Want some opinion from the experts, from my research and reading from this thread the most difficult transition is going from planar to Finfet, once you successfully conquered such technical feat then you have a pathway to lowered the node. 14nm, 7nm and 5nm all use the same materials (mask, photoresist and others) and tools (SSA800A and NXT 2050) if applied using DUVi, so from the look of things we can predict a complete domestic 7nm line within the year, this is one of many evidence with the new SMIC Beijing Fab as another.
From the article:
ArF is the most difficult to manufacture, which is also an indispensable raw material in the 14nm/7nm chip manufacturing process. The technology of the chip is also graded. The technology level of tablet computers and automobile chips is not high. The real difficulty lies in the 7nm chip.
Mobile phone chips of this process not only require advanced EVU lithography machines to produce, but also require high-end photoresist as auxiliary materials and a large number of chip raw materials to successfully produce 7nm chips. The lithography machine is monopolized by companies in the United States and the Netherlands. Now the EVU is in a state of disconnection to China. The EVU lithography machine purchased by SMIC for 1.2 billion has not yet arrived; chip raw materials, although some domestic Raw materials are produced independently, but raw materials such as silicon wafers, photomasks, electronic gases, polishing materials, sputtering targets, and wet electronic chemicals are completely dependent on imports.
In addition, although the most advanced EVU lithography machine has been cut off, the supply of the DVU lithography machine with relatively low process technology has not been cut off, and the DVU lithography machine can also be used for 7nm process chips. Achieving the fabrication of a 7nm chip. This means that in 2022, the existing technology may be able to mass-produce 7nm in advance.
Jun 22, 2022 — ArF photoresist is the most difficult to manufacture, which is also an indispensable raw material in the 14nm/7nm chip manufacturing process.
Not that surprising. Intel is hemorrhaging cash with these EUV fabs. Why do you people think they tried to hold on to their DUV tools for as long as possible? Tools are extremely expensive and it is hard for Intel to finance the new fabs with increasing competition from both AMD and ARM (in client and server chips). These fabs will require huge government subsidies and creative financing.
Intel's GPU line (Arc) is a huge money sink. Now it turns out their top end higher clocked chips, their cash cow, are failing prematurely.
Intel is in deep sh*t.
People talk as if the Kirin 9000S is the first SoC that HiSilicon fabbed at SMIC. But they had already launched the Kirin 710A at 14 nm FinFET way back in 2020. If you believe Gina Raimondo and these other US muppets the goal of the US government sanctions was to stop all Chinese FinFET chip production. The US totally failed at that.
The Huawei Mate 60 is just rubbing salt on the wound. Since this proved that SMIC had attained parity with Intel despite being sanctioned.
SMIC can produce its own masks. They have their own mask shop.
Trying to ban "FinFET" by restricting tools or materials is a fool's errand which could only have come out of Eric Schmidt's useless AI commission. The legislation was conceived and written by people with basically no knowledge of how semiconductors are made.
"FinFET" is a transistor architecture and is not something which can be banned with tools or materials restrictions like that.
The FinFET 16nm chips made by TSMC used similar immersion lithography tools to those used to make 28nm planar chips. To further increase density beyond that requires more precise machinery than those tools had but the basic tool setup is the same. There are a couple of Chinese companies who supposedly work on making that kind of precise machinery. It would not be surprising if someone in China could modify existing lower spec tools to make them work on higher density processes. But that is besides the point. Because the original FinFET process was the Intel Tri-Gate at 22nm. Which is a process with tool technical requirements similar to 28nm.
You can technically make a FinFET transistor in any process you wish. It is an architecture. The only question is if it is worth using those instead of planar transistors or not. The industry moved to FinFET architecture to reduce leakage at smaller transistor geometries. But it is not like FET or FinFET is new.
Masks, and photoresist are roughly similar for both ArF and ArFi since those operate at the same light wavelength. So the US would have to ban sales not just to the producers of "FinFET chips" but anything 55nm and higher. Plus China can already produce those materials. As you could find out by reading this thread. The only thing is production of those materials in China is still not enough to cover Chinese industry demand. But Western sanctions would easily solve that by forcing Chinese industry to make investments into scaling up their production. Right now much of the industry is still dragging its feet on adopting Chinese materials.
Last edited:
Then ASML is screwed, aside Intel who will buy their High NA EUVL? TSMC and Samsung the other 2 EUVL operator had stated they will still use their old Euvl machine instead of buying the latest one.
Tsmc will continue to buy 0.33NA euv. They have delayed high-na for now.
so it looks to me that Chinese govt itself has no idea what the domestic chipmaker market share is. Or at least they are horribly behind in times. No wonder we keep seeing articles saying they are at < 10% for domestic share.