Chapter 162 Unimaginable Difficulties
“Dr. Chen, is this really feasible?”
The industry has many theoretically feasible solutions, but difficulties will be found in the actual implementation process. Overcoming the difficulties, the distance between the theoretical feasibility and implementation may be hundreds of thousands of miles.
What's more, what they do is completely from zero to one. Crossing the river by feeling the stones, you can know how deep the river is and what kind of preparations are needed. There is no reference to explore a new river. Rivers, the difficulties involved are difficult to estimate.
Chen Yuanguang saw that the person who had just sent the message was Liang Mengsong. He looked directly at him with firm eyes: "It must be possible."
Chen Yuanguang didn't explain anything more. He liked to let the results speak for himself. He had already explained the theoretical feasibility.
Liang Mengsong is considered by the outside world to be the person who knows the most about chip manufacturing in China. He is believed to have a keen sense for chips that is far beyond imagination. For a long time after SMIC was defeated in the internal struggle, he Within SMIC, he is an idler with a position but no real power.
After Huawei planned to use a new process and use 28nm double exposure technology to make 7nm process chips, Liang Mengsong was able to escape from the treacherous struggle within SMIC and began to become the actual person in charge of this project. square.
For him, doing things is far easier than fighting.
Contrary to outside speculation, Huawei's 7nm chips are obtained through a process that sacrifices yield rate. They have not lost much yield rate, which is far less exaggerated than the 50% to 70% yield rate that outsiders think.
For Huawei, using 7nm on their flagship model mate60 can not only make money, but also make a lot of money.
Under his leadership, Huawei and SMIC will next use 28nm process lithography machines to advance chips to 5nm. They have already seen the dawn of victory.
Because of his excellent resume, Liang Mengsong was assigned to the 712 Project to be responsible for chip manufacturing.
He himself was also very interested after receiving the task. He has always been a fan of technology, not to mention that materials such as topological semimetals that only existed in legends in the past have now become a reality. , he accepted this task with almost no hesitation.
The fact that the progress was not going smoothly made him gradually feel uneasy. Can they really spend two years to complete this brand new process? You must know that this is a brand new process, and there is a big difference between the photosensitivity of the so-called topological semimetal and the photoresist.
What's more important is that Chen Yuanguang cannot put all his energy into this project. Even if Liang Mengsong wants to see him most of the time, he will be in meetings. It is very difficult to make an appointment with him in private.
Past experience is failing, researchers in all aspects of the project are gradually becoming impetuous, everyone's unconditional trust in Chen Yuanguang's aura is dissipating, Liang Mengsong feels that the situation is getting out of control, and his inner uneasiness is piling up.
The firmness in Chen Yuanguang's tone made him feel less uneasy. What he feared most at this time was that even Chen Yuanguang himself didn't believe in himself.
"Okay, then we will do process simulation according to this plan." Liang Mengsong said.
After the meeting, Liang Mengsong flew back to Shenhai to promote follow-up specific work.
The entire Project 712 project team is scattered across the country. After all, chip design is in Pengcheng, process manufacturing is in Pengcheng, Shenhai and Jiangcheng, and some large and small suppliers are scattered across the country.
“Do you think two years is really enough?”
There are too many project members, but the engineers responsible for chip manufacturing are the core personnel anyway.
SMIC selected a group of the most experienced engineers to join the project team.
This is not only because of the Chinese official, the Chinese official will only say that they need to cooperate vigorously.
Specifically, Chinese officials don’t know which of your internal engineers are experienced and which are reliable, and they won’t tell them.
The reason why we work so hard and provide the greatest support is more due to SMIC’s own actual needs.
Because no matter in the short or long term, it is impossible for ASML to provide them with advanced process lithography machines.
Cooperating with Huawei can advance the process to 5nm at most, but because the technology is jointly developed by Huawei and them, this process can only be sold to Huawei and cannot be provided to other customers.
And because the process is jointly developed, the benefits enjoyed by SMIC are very limited, and they cannot achieve excess returns through breakthroughs in the 7nm process.
It can be clearly seen from the capital market that SMIC's stock price did not rise due to the release of Huawei's mate60 equipped with a 7nm chip.
Project 712 can help SMIC advance the equivalent process to 1nm or even higher. In SMIC’s view, this is simply the same as when TSMC and ASML jointly developed immersion lithography machines and then defeated them. A once-in-a-lifetime opportunity like any other competitor.
Even if it is only a theoretical possibility, they do not want to give up this opportunity.
After signing the confidentiality agreement, the selected engineers were very excited when they saw the technical parameters of topological semi-metal.
For me personally, being able to participate in such a project, whether I change jobs or continue working here, is a successful experience worth writing about.
In addition, room-temperature superconductors had just come out at that time, and Chen Yuanguang was shrouded in an aura of omnipotence. Everyone had no doubt that this matter could be solved in two years, and they all imagined punching TSMC and kicking Intel in two years.
After half a year has passed and little progress has been made, everyone has lost their morale.
"I don't think it's enough. The problem is that Guangshen thinks it's enough. Before the two-year period is up, I believe in Guangshen."
"You are too optimistic."
“It’s not that I’m optimistic, but what else can I do now but believe? Believe first and then question.
Anyway, they are always giving solutions. ”
“This is just a light source, there are a lot of processes waiting to be improved in the future.
In the current chemical grinding, there is no process for topological semi-metal. ”
Chemical mechanical polishing is a key process in chip manufacturing. It uses machinery and grinding fluid to achieve a flat surface of the silicon wafer, avoiding circuit short circuits caused by different resistance values caused by undulations on the silicon wafer surface.< br>
The uneven surface of the silicon wafer will not only cause circuit short circuit problems, but also cause the inability to focus accurately during the photolithography process, resulting in the failure of line width control
So it's almost an integral part.
“Perhaps topological semimetal does not require chemical grinding?”
“No chemical grinding is required? How is that possible? It is a semimetal, not a room temperature superconductor, and the entire internal resistance is not zero.< br>
It still has resistance. "The existing chemical polishing process requires consumables such as polishing fluid, polishing head and polishing pad. They are all consumables for specific materials. The current process is mainly for silicon grinding and silicon oxidation. Object grinding.
Therefore, there are no suitable consumables and processes for topological semi-metals.
“You can find a solution even if you need chemical grinding, and the process is not difficult.”
“It’s not difficult, the problem is time. What I want to express is how to do it in two years It’s not enough.”
“Why is it stuck for two years? It’s because TSMC’s process will most likely reach 1nm in two years, which is the limit of the current system. What Guangshen wants is to overtake in a corner. , let’s wait for them before they can achieve 1nm.
The problem is, so what if the entire process flow can be solved in the end, let alone two years, even five years is not enough for SMIC.
Shenhai Microelectronics' 28nm nationally produced lithography machine has been clamoring for delivery since 2020, and now it's already 2024 and they haven't successfully delivered it. Isn't it okay? "
"That's right, there's no need for us to be anxious, it's useless to be anxious. ”
SMIC’s internal engineers involved in the 712 project are becoming less and less optimistic that a new process can be completed in two years.
From the perspective of the capital market, the same is true. The 712 project has been After American found out, the stock prices of listed companies such as Intel, TSMC and NVIDIA quickly stopped falling after a slight drop that day
American simply did not believe that they could do it, even with Chen Yuanguang.
"Impossible, absolutely impossible.
It is absolutely impossible for China to use new materials to bypass the limitations of the photolithography machine and advance the chip to the 1nm process."
Although Wall Street does not believe it , but Washington is still facing this possibility. Quantum computers are enough to block their computing power. If they achieve another breakthrough in chip technology, they will still reach 1nm before Americen, then the world will The consumer electronics market is facing a reshuffle.
So they urgently held an internal secret hearing on Capitol Hill and invited Intel CEO Pat Gelsinger and Texas Instruments CEO Rich Templeton to accept the hearing.
Pat said categorically: "Optimizing the existing chip manufacturing process is a huge project. Even if Americen wants to do this, it must rely on global partners including Neon, the Netherlands, Korea, England, etc. This is only possible with the cooperation of dozens of supply chain companies in more than a dozen countries.”
"Mr. Gelsinger, I'm sorry to interrupt you. You just mentioned that optimizing the existing chip manufacturing process requires the cooperation of dozens of supply chain companies. According to history, TSMC has been in the process of dry lithography machines to immersion lithography machines. In the process of process optimization, we only cooperated with ASML," a congressman asked.
Pat thought they were just here to put on a show to reassure Washington, but he didn’t expect that these gentlemen actually made preparations:
“You’re right, from dry style It is relatively easy to enter the immersion type. Adding water between the light source and the silicon wafer is a very weak optimization, and the Chinese want to replace the silicon wafer with topological semi-metal.
This can be described as optimization. Not enough. This is to completely bypass the existing chip manufacturing process system. This is much more difficult than dry to immersion. The difficulty between them is not of the same order.
The difficulty of the former is like going from dry to immersion. The transition from personal PDAs to smartphones, and the latter from smartphones to the metaverse.”
Pat did not correct the other party's mistake. From the dry lithography machine to the immersion lithography machine, not only ASML needs to cooperate, but Zeiss also needs to cooperate and redesign their objective lenses.
"That means they will definitely not succeed? Even with Wright Chen's participation." asked another congressman.
Pat smiled after hearing this, and then straightened his tie: "Sorry, I heard that Wright Chen is nicknamed the God of Light in China, but unfortunately this is the real world and does not exist. God.
I believe this is a feasible technical solution for topological semi-metallic wafers, but I don’t believe that China can do it all by itself
Man, not God.”
Pat and Rich glanced at each other and saw the helplessness in each other's eyes. In their opinion, the gentlemen in Washington had been frightened by Wright Chen and lost their rational judgment. They actually believed that China could subvert the country on its own. Remove the semiconductor industry that the world has built over decades.
Rich said: "I agree with Pat's point of view.
I know Dr. Wright very well. We met once before at an academic conference at MIT. He has extraordinary academic insights.
He will definitely be an outstanding scientist, or he already is.
But industry and academia are two different things, and industry needs more. links require more cooperation, and a single process breakthrough requires the cooperation of other processes
From silicon wafers to topological semi-metals, this involves the replacement of the cornerstone of the entire chip production process. It is a very difficult thing. Although the academic community has been making attempts, it is hoped that new materials can be used to break through the quantum tunnel after silicon reaches its limit. effect, but it’s still too difficult.
I also don’t think Wright can do it all by himself.
China lacks the most cutting-edge technology accumulation in chip technology. ”
“Then do you think topological semimetals are real? "
"It should be that topological semimetals have a long history of technological evolution. It was proposed a long time ago, and it is normal to achieve breakthroughs. Dr. Wright himself has too many things worth mentioning in materials science. It worked.
Although topological semimetal is an exaggeration, it is far less difficult than room temperature superconductivity. Since room temperature superconductivity has been unanimously identified by the global academic community, I think topological semimetal has not been announced to the outside world. It was created by Dr. Wright. I Believe it to be true. ”
Speaking of this, Pat sighed inwardly, if Chen Yuanguang were from America, they would be the first to achieve technological breakthroughs, and with the cooperation of Silicon Valley, they would be able to replace silicon wafers with topological semi-metals. , advancing the process to below 1nm.
This is a great achievement that will allow Silicon Valley to become the world's chip manufacturing center, instead of relying on TSMC for its most advanced chip production. >
"Then if topological semi-metal is true, how long will it take at least for China to achieve process substitution? "
"Ten years. " Rich said.
"Fifteen years," Pat said.
“So can we think that by that time, we have broken through the 1nm limit?”
“That’s right.” The two said in unison.
Not long after the secret hearing ended, Huawei held their Hongmeng Eco Spring Communication Meeting. It was said to be a communication meeting, but in fact it was a product launch.
When the outside world thought that this new product launch conference would focus on the Zhijie S7 and compare it with Xiaomi Automobile, using a large number of far ahead to demonstrate Huawei's influence in car manufacturing, these contents None.
(End of this chapter)