In February 2021, Professor Tang Chuanxiang's team from Tsinghua University collaborated with German scientists to publish a paper in Nature, reporting that they had completed certain experimental verification of the theory of producing ultra-deep ultraviolet light sources based on the SSMB principle. But today, two years later, this work was suddenly hyped by some self-media as the establishment of an EUV lithography machine factory in China. However, can it be so fast to independently manufacture a photolithography machine?

Written by Wang Jie

In the past week, a piece of science and technology news has circulated on the Internet. The news is that Chinese scientists have discovered a new principle of generating ultra-deep ultraviolet light sources that can break through the technical problems of lithography machines. Many people even say that our country has started to build a photolithography factory in Xiongan. "There are pictures and the truth." They have a nose and an eye.


Many people ask me if it is true. Let me tell you the answer first: the new light source generation principle is true, but it was proposed as early as 2010.It is still in the principle verification stage, and there is still 15 to 20 years to go before it is truly practical.The paper by the Tsinghua University scientist that was being hyped this time was actually published in early 2021. I don’t know why it was suddenly dug up and hyped two and a half years later. As for the photolithography factory to be built in Xiongan, it is just a rumor and a lie.

I just want to use this topic to talk to you today.Why is it so difficult to build a lithography machine? Is it possible for China to completely independently develop the most advanced lithography machine?

Photolithography machines are key equipment used to produce chips. Every computer we use and the chips in every smartphone are produced using photolithography machines.


To measure the technological advancement of a chip, the unit of xx nanometers (nm) is used.Nano is a unit of length, 1 nanometer is equal to one billionth of a meter.Didn't Huawei release its latest mobile phone, Mate60pro, two weeks ago? As soon as this mobile phone came out, everyone exclaimed, wow, the chip used in this mobile phone is made of a 7nm process, which is incredible. Here is an explanation of what the 7nm process means. Simply put, the electronic components on the chip, that is, the transistors, are engraved, just like we are engraving on a rubber stamp.On the same area, the more transistors that can be carved out, the more advanced the chip will be.In the field of chips, nanometers are used to express the advanced level of the chip. The smaller the number, the more advanced the chip is. 10nm is more advanced than 14nm, and 7nm is more advanced than 10nm. Don’t worry about why there are numbers like 5, 7, 10, and 14. There are complex historical reasons behind them.

Chips are carved on silicon wafers using lasers, so the smaller the transistor is to be carved, the shorter the wavelength of the laser is required.The light source used by the most advanced lithography machines in the world is called extremely deep ultraviolet light, or EUV in English, with a wavelength of 13.5 nanometers.It was developed by an American company, but the American company has now been acquired by the Dutch company ASML. However, there is a concept to be clarified here. It does not mean that the 13.5-nanometer wavelength laser can only engrave 13.5-nanometer chips. It can actually engrave 7-nanometer, 5-nanometer, or even smaller process chips.


The light source used by lithography machines that is worse than EUV is deep ultraviolet light, which is abbreviated in English.DUV, the wavelength is 193 nanometers, which is an order of magnitude larger than EUV. The 7nm process Kirin 9000s chip used in Huawei's latest mobile phone is carved using DUV. Yes, at a wavelength of 193 nanometers, a 7nm chip can be carved using a technology called multiple exposure. But even this kind of 193-nanometer photolithography machine is still not available in our country. The only companies in the world that can produce DUV are Japan's Canon and Nikon, and the Netherlands' ASML. Yes, you heard it right, neither can the United States.


By the way, here is what multiple exposure technology is. Let me use the simplest analogy to try to explain. For example, now you have a machine that draws square grids, but the side length of the square grid it can draw is 100 mm. Is there any way you can use this machine to draw a square grid less than 100 mm? It is possible. The method is that I first draw many connected grids on the paper to form a grid. Then I moved the machine slightly and drew again on the paper. This will draw a new grid. The two grids overlap and the lines will intersect to form a smaller grid. You can try it yourself with a pen on paper.

Each time a photolithography machine engraves a chip, the process is one exposure. The same is true for using DUV to produce 7nm process chips. If you can't do anything at one time, just expose it a few more times. After each exposure, move a small step before exposing again. This allows for carving out smaller transistors. Of course, this is not without side effects, that is, the possibility of errors is greater. In mass production, many failed chips will be wasted. In professional terms, the yield rate of chips is relatively low and the defective rate is relatively high.


Let's get back to the topic,How difficult is it to build a photolithography machine?

I'll decide on my personality first.The lithography machine is by far one of the most precise and complex machines that humans are capable of manufacturing.A photolithography machine consists of three key parts.The first part is the light source, the second part is the optical system, and the third part is the etching workbench.The technical challenges of each section are comparable to landing on the moon.

Let’s talk about the light source first.To generate ultra-deep ultraviolet light with a wavelength of 13.5 nanometers, the current method is to use a high-power laser to bombard a small ball of tin (that is, metallic tin) with a diameter of only 1/30 millionth of a meter. But this sentence is not enough to describe its difficulty. I need to expand.


First, a laser beam is required to accurately hit a small solder ball that is moving at a speed of about 200 miles per hour. When the temperature of the small solder ball reaches 500,000 degrees, a laser beam is then used to bombard it. At this time, extremely deep ultraviolet light with a wavelength of 13.5 nanometers can be produced.To continuously and stably produce this kind of ultraviolet light, small solder balls need to be bombarded at a frequency of about 50,000 times per second.There is only one German company in the world that can produce this kind of laser. This German company called TRUMPF took ten years to successfully develop it. This laser alone has more than 45,700 parts. But you may not have thought that TRUMPF's laser relies on a Lithuanian company to provide key equipment. Without the light source equipment manufactured by this Lithuanian company, TRUMPF would not be able to do it. It is simply like a mantis stalking the cicada and the oriole behind it. The next difficulty is how to collect this ultra-deep ultraviolet light to form a ultra-deep ultraviolet laser? This is the next key part.


Optical system.This optical system developed for EUV can only be manufactured by one German company in the world, and it is the famous Zeiss. You may have heard that the camera lenses produced by Zeiss are among the best in the world, but comparing camera lenses to lenses used in EUV optical systems is like the difference between an airplane with propellers spraying pesticides and a jet fighter. This optical system involves at least the following technical challenges: high-precision aspheric surface processing, multi-layer film mirrors, high-quality melting, ion beam polishing technology, and extreme precision grinding. You don’t need to go into the technical terms just mentioned, you just need to know,The ultimate goal is to create an absolutely smooth and flat lens, how smooth should it be? It’s the smoothness of water droplets in the three-body system.The fluctuation of the lens is an error of about one atom, which is close to the theoretical physical limit.If we use Zeiss's own promotional metaphor, even if this lens is enlarged to the size of the entire Germany, the fluctuation does not exceed 0.1 mm. If a virus falls on this mirror, it will be like a hill rising up to 100 meters high. Therefore, this optical system must work in a vacuum without any interference. But having a light source and lens is not enough. It is just like having a carving knife for engraving. The next step is to carve tens of billions of transistors on a silicon chip the size of a fingernail.

Precision instrument workbench.In order to carve tens of billions of transistors, we need a console with extremely high precision. It is difficult for me to find an accurate metaphor to describe the difficulty of manufacturing it. This console is composed of 55,000 high-precision parts, and these parts rely on at least patented technology provided by Japan, South Korea, Taiwan, the United States, Germany, and the Netherlands. It would not work without any country.

The above is probably the difficulty of manufacturing the most advanced photolithography machine in the world. Its research and development history is roughly like this: In 1997, Intel Corporation and the U.S. Department of Energy jointly invested in a company and began to develop EUV lithography machines. In 6 years, this company has developed most of its core patented technologies. However, neither Intel nor the U.S. Department of Energy intends to build lithography machines themselves, because they feel that building lithography machines does not actually make money. It is better to license the core technology to a foreign company and let them build lithography machines. Later, ASML in the Netherlands obtained the authorization for these core technologies, and with the help of companies such as Samsung and TSMC, it finally produced the first EUV lithography prototype in 2010. It spent another 9 years testing, optimizing, and upgrading, and finally produced the first EUV lithography machine that could be officially put into commercial production in 2019, which took a total of 22 years.


However, although the EUV lithography machine is produced by ASML in the Netherlands, it is nothing more than an assembly plant. Only 15% of the parts are produced independently, and the other 85% of the parts are imported. And because the U.S. Department of Energy owns almost all core patents for lithography machines, ASML's production of lithography machines requires authorization from the U.S. Department of Energy. This is why if the U.S. government says it is not allowed to sell photolithography machines to China, the Dutch ASML company can only listen to it. It can be said thatAn EUV lithography machine is made up of seven or eight countries forming a circle, jamming ASML’s neck..

If China wants to break through the technological blockade and independently produce lithography machines, it needs to achieve complete independent innovation in all three key parts. All we can say now is that in the first light source section, we see a little bit of hope.

In 2010, Zhao Wu, a Chinese professor at Stanford University and a distinguished visiting professor at Tsinghua University, worked with his doctoral students to propose a new principle for generating extremely deep ultraviolet light sources. This principle is called"Steady-state micro-bunching", the English abbreviation SSMB, uses a huge particle accelerator to generate extremely deep ultraviolet light.In 2017, Professor Tang Chuanxiang's team at Tsinghua University worked with colleagues in Germany to complete the theoretical analysis and physical design of the experiment, develop a laser system for the test experiment, and conduct certain principle verification. In February 2021, their paper was successfully published in the journal Nature[1], Professor Tang’s doctoral student Deng Xiujie is the first author, Professor Tang and another professor from the Helmholtz Center for Materials and Energy Research in Berlin, Germany, are the corresponding authors. By the way, here is a general rule in the academic circle. The first author generally refers to the person who has made the greatest contribution to the research topic, while the corresponding author is the person in charge of the topic and the beneficiary of the results.


In March 2022, Professor Tang Chuanxiang and Dr. Deng Xiujie published a paper of the same name in my country's "Acta Physica Sinica"[2]. Maybe they themselves didn't expect that more than a year later, for some unknown reason, probably on September 13, 2023, some self-media posted a video with a title like "It's against the sky!" Tsinghua University’s SSMB-EUV light source was born, with power 40 times that of EUV lithography machines.” Then, like a fire, various self-media platforms began to hype Tsinghua University’s SSMB solution with various titles starting with the words “outrageous”. I was dumbfounded when I saw it.

What I hope everyone can calm down is that we are still a long way from realizing the production of ultra-deep ultraviolet lithography machines.Don't get ahead. First of all, Tsinghua’s official website states that in 2021, Professor Tang Chuanxiang has applied to the National Development and Reform Commission to list the SSMB experimental device as a major national science and technology infrastructure during the 14th Five-Year Plan. However, I didn't find any news about the project. Considering that this is a civilian scientific research project, not a military project, public announcement is required if the project is approved. Therefore, at least so far, this project has not been approved.

Even if we are optimistic, the project can be approved next year, but it will be difficult to build a scientific research device of this level within five years. After it is completed, we will be more optimistic and test it successfully in 3 years, and then spend another 5 years to build a light source that can be used commercially. This will be 13 years ago. However, can the other two key parts of the lithography machine be completed in these 13 years? There is not even a shadow yet.

Moreover, we don’t know if the Americans and the Dutch will have developed a more advanced next-generation lithography machine in 13 years, and we have to continue to pursue it.

Finally, I want to say something that I personally don’t like:

Within 20 years, it is impossible for any country in the world to be able to completely independently build a lithography machine that represents the most advanced level in the world, and the United States is no exception.

Of course, this only represents my personal opinion, and I really hope to be slapped in the face.

The reason why I want to express this point of view is because I really do not want the tragedy of the past "Great Leap Forward" to happen again. Chinese people are very smart, but that does not mean that we Chinese are made of special materials. All races in the world are Hominidae, Homo, and Homo sapiens. There is almost no genetic difference between Chinese and foreigners. We are not stupid than foreigners, but we are not much smarter than foreigners.

Seeking truth from facts is the right way to develop science and technology. For ultra-precise and complex machines like photolithography machines, seeking the greatest range of international cooperation is the best solution.

access:

Jingdong Mall