Is this new breakthrough in EUV lithography real?

In the past few days, confusing rumors about lithography have come again.

Things revolve around EUV, and rumors package it as "having noses and eyes", which probably means that it is better than nothing. The deeper the fermentation, the more lingering the rumors become.

This has also triggered heated debate among EEworld friends. This solution is like a sniper rifle with an accuracy of 1,000 meters. We make a 999-meter gun barrel, and a photolithography factory can produce lights of any length. Although the imagination is rich, it is still difficult to distinguish between true and false. (If you have any other ideas to communicate with engineers, you can move to EEWorld’s original post for discussion: https://en.eeworld.com/bbs/thread-1256513-1-1.html)

Could this solution be the key to bypassing ASML?

Not only that, there are many equipments that are stuck in China, but the lithography machine is the one with the slowest growth in localization rate.

For chips, EUV lithography machines must be used after 5nm. This is because when the metal spacing shrinks below 30nm (corresponding to process nodes exceeding 5nm), the resolution of the lithography machine is not enough.

From the formula "Lithography machine resolution = k1*λ/NA", we can know that the larger the NA, the higher the lithography machine resolution and the more advanced the process. However, the NA aperture is not that easy to increase, so the lithography machine chose to change the light source and replace the 193nm DUV light source with a 13.5nm EUV light source, which can greatly improve the resolution of the lithography machine. Not only that, EUV can also reduce process steps and improve yield.

What is the difficulty with EUV? According to ASML official information, an EUV lithography system includes multiple components such as laser, objective lens, optical path system, test bench, exposure bench, measurement equipment, and shock absorption device. Each component has extremely high requirements, is very precise, and requires Perfect fit, extremely difficult to manufacture:

• Contains 100,000 parts and weighs approximately 180 tons;

• Requires 40 freight containers, transported via more than 20 trucks and 3 cargo planes;

• The reflectors used in EUV systems need to be very flat. If enlarged to the size of Germany, the largest bulge would be less than 1 mm high;

• EUV light is generated by firing a high-energy laser onto a drop of molten tin - a process that occurs 50,000 times per second;

• The EUV system can control the beam so precisely that it is equivalent to shining a flashlight from the Earth and hitting a 50-cent coin placed on the moon;

• The EUV system contains a large vacuum chamber weighing 7,600 kilograms.

In addition to EUV, there are other technologies. Research on lithography around the world has been very active. It has a history of nearly 60 years, and there are far more lithography technologies than imagined.

Although EUV is not the only way to advanced processes, it is just not that good or too expensive. EUV is still the main direction now.

For the chip industry, the ideal photolithography machine technology should be low cost, high throughput, small feature size, and independent of materials and substrates.

Now, except for optical lithography, most lithography technologies cannot meet the needs of mass production. Direct-write lithography technology is generally used for masks or customized chips.

Optical lithography is a technology that has been screened through layers of history and finally won. It continues to extend from ultraviolet lithography (UV), deep ultraviolet lithography (DUV) and extreme ultraviolet lithography (EUV). Its wavelength It also extends from 436nm, 365nm, and 248nm to 193nm and 13.5nm.

IEEE has always planned what lithography technology to use for each process technology node, and it has been going on for decades. It is the International Device and System Roadmap (IRDS).

According to the roadmap, extreme ultraviolet lithography (EUV), guided self-assembly (DSA) and nanoimprint lithography (NIL) are technology candidates for the next generation of nanoprocess nodes. However, nanoimprint lithography (NIL) is not currently capable of large-scale production, and is mostly used for small-scale chip customization or mask making.

Process technology development routes and potential technologies, image source丨IEEE

First of all, the results of this paper must be epoch-making. We believe that the power of domestic products is becoming stronger. But correspondingly, the "added fuel and jealousy" gossip from the outside world is not credible and can be said to be just rumors.

This solution is not impossible, but it has the feeling of being broken. It is this sense of ambiguity that makes people feel that the rumors are true.

American psychologists Allport & Postman proposed a formula for rumor transmission in the 1940s, which is "rumor spread opportunity = personal attention level × ambiguity of event evidence." In other words, the uncertainty of events and satisfying people's curiosity are key. The truth is far less important than the need for support for one's own position and opinions, so people are only willing to believe what they want to believe.

For market information, we need to look at it rationally:

On the one hand, we must consider the practical application of technology. There are many new routes written in papers in the world, such as quantum computing and brain-inspired computing. These technologies are very good and can be realized by spending a lot of money, but they lack commercial practicality, so even Intel only says that we are researching them. But the official launch will have to wait until there is a market. Investment institutions will not blindly invest money, nor will they blindly invest large amounts of money without any reason. Everything ultimately depends on business logic, and the road must be walked step by step.

On the other hand, technical implementation issues must be taken into consideration. Lithography machines are not only gold-swallowing beasts, but also electricity-swallowing beasts and water-swallowing beasts. How much electricity such a large machine uses, how much water it uses, and how to treat wastewater are all issues that need to be considered; just having an EUV light source does not represent technology. Breakthrough, mask, photoresist, and photolithography system construction are all issues that need to be considered; after all the twists and turns, what is the final production yield and production efficiency? If it is not as expected, is it practical to take risks to build such a machine?

On the EEworld forum, engineers believe that it feels like a direction, but practicality must be considered. If you can do 4nm but the output is slow, it doesn't make sense. Some engineers said that as the saying goes, don't watch the ads, watch the efficacy. It's hard to tell now, we can only wait for a while.

In We Media, we are making breakthroughs in lithography technology every day and making the world tremble every day. We believe that the future will be better, but there is no need to brag about hard-won paper results without considering practical problems. Time never speaks, but it will give you the answer you want.