China May Be Constructing EUV Lithography Machines on a Massive Scale

Over the past few days, news on the possibility of China circumventing US sanctions through an unexpectedly innovative approach called SSMB-EUV has generated substantial public buzz in China. With this cutting-edge technological breakthrough, China could construct an “EUV factory” to replace ASML’s individual EUV lithography machines. But is this feasible?

Basically, why are lithography machines so important? Why is it so difficult to manufacture one, and can China independently develop cutting-edge lithography machines?

Microchips, or semiconductors, are the brains behind almost all of our electronic devices, from computers to phones, cars to appliances. They’re essential to our digital world, and advances in chipmaking can have far-reaching impacts. A key process in chipmaking is lithography.

The production of cutting-edge microchips below 5 nm is heavily reliant on extreme ultraviolet (EUV) lithography machines, which are considered the crown jewels of the semiconductor industry. The only company in the world that can produce EUV lithography machines is ASML in the Netherlands.

To understand why lithography machines matter so much for making advanced microchips, let’s first take a look at how lithography works.

Lithography machines are devices that use light to create tiny patterns on a silicon wafer, which is cut from a bar of 99.99 percent pure silicon and refined to perfection. The patterns are like the instructions that tell the microchips what to do.

To make these patterns, lithography machines use a thin plate with a circuit diagram on it, called a mask. The mask is like a stencil that you can use to spray paint a letter on a piece of paper. The mask blocks some parts of the light and lets others pass through, creating a pattern of light and dark areas.

The light then goes through a system of mirrors and lenses that shrink and focus the pattern onto the silicon, which is coated with a special material that reacts to the light, called photoresist. The photoresist is like the paper that you want to print the letter on. The photoresist changes its properties when it is exposed to the light, making it easier or harder to remove.

By removing the unwanted parts of the photoresist, the pattern on the mask is transferred to the silicon. The silicon then goes through other processes to become a microchip.

The smaller the patterns on the mask, the more powerful and efficient the microchip can be. However, there is a limit to how small the patterns can be, which depends on the type of light used. The type of light is determined by its wavelength.

To print smaller patterns, lithography machines use different types of light sources with different wavelengths. The most common types are deep ultraviolet (DUV) and extreme ultraviolet (EUV) lithography systems. However, there is a limit to how small the patterns can be, which depends on the wavelength of light used. Shorter wavelengths allow smaller patterns.

The most advanced lithography systems use deep ultraviolet (DUV) or extreme ultraviolet (EUV) light.

To print smaller patterns, lithography machines use different types of light sources with different wavelengths. The most common types are deep ultraviolet (DUV) and extreme ultraviolet (EUV) lithography systems.

DUV lithography systems use light with wavelengths in the ultraviolet range, which are shorter than visible light. DUV systems can print patterns as small as 7 nanometers (nm), which is about 10 times smaller than a human red blood cell. However, to achieve this level of resolution, DUV systems use various techniques to improve image quality and reduce errors, such as using water or oil between the lens and the silicon, printing multiple layers of patterns, and using mathematical models to correct distortions.

EUV lithography systems use light with a wavelength of 13.5 nm, which is almost in the x-ray range, which is even shorter than ultraviolet light. EUV systems can print patterns as small as 3 nm, which is about 20 times smaller than a strand of DNA. EUV systems enable the making of more complex and efficient microchips with new designs and architectures.

Now we have a basic understanding of why lithography is so critical for advanced chips. But a lithography machine itself is one of the most precise and complex machines humans can currently produce. An EUV lithography machine has three key components with over 100,000 high-precision parts:

From concept to implementation, it took over 20 years for EUV lithography to become commercially viable. ASML spent 13 years developing the first EUV prototype and another 10 years to reach high-volume manufacturing.

Although produced by ASML, EUV lithography machines are only 15% in-house, with 85% of parts imported. EUV lithography is a collaborative achievement circling seven to eight countries.

From concept to implementation, it took over 20 years for EUV lithography to become commercially viable. ASML spent 13 years developing the first EUV prototype and another 10 years to reach high-volume manufacturing.

For China to break technical barriers and independently produce advanced lithography, it would need to achieve complete self-reliance across all three key components.

However, EUV systems also face many technical challenges, such as generating enough power for the light source, keeping the optics in high vacuum conditions, and producing flawless masks and photoresists. EUV is hitting fundamental limits, with source power capped around 500 W. This strains patterning for the 3 nm node, raising costs.

To overcome the shortage of EUV machines and circumvent the US embargo, the brightest minds in China have been working around the clock to find a solution. Rather than following the traditional path set by ASML, they have devised a remarkably different approach.

EUV machine is a massive presence with roughly the size of a bus. And it’s already as small as the engineers can make it. For ease of shipment and transportation, they have tried everything to miniaturize the behemoth to its current dimension. Just like with other electronic devices, making a gadget smaller is always more challenging than making it bigger. This holds especially true for the sophisticated light source of EUV.

So, why not build a giant EUV light source if it’s never intended to be exported or shipped to foreign customers?

This novel concept is the aforementioned SSMB-EUV technology.

In a remarkable paper titled “Steady-state micro-bunching accelerator light source”, published in the Chinese journal Acta Physica Sinica, scientists from Beijing’s Tsinghua University propose a plan to build an accelerator the size of a football field and use it as a light source for a few dozens of EUV machines working simultaneously. The accelerator in question is a circular ring-shaped device where charged particles, such as electrons or protons, are accelerated to high speeds and kept in a stable orbit using magnetic fields. The particles circulate within the ring repeatedly, allowing them to undergo multiple interactions and experiments. In such particle accelerators, electrons are typically arranged close together in a controlled manner, forming bunches to optimize their behavior and interactions. It’s called electron bunching.

In the context of steady-state micro-bunching (SSMB), the mechanism described in the research paper, electron bunching is achieved through laser manipulation. Laser beams are used to interact with the electrons, causing them to oscillate and form tightly packed groups or micro-bunches. This laser-induced bunching process is precise and allows for the creation of electron bunches six orders of magnitude smaller than what is achievable with a conventional particle accelerator.

By harnessing the powerful coherence of radiation generated through micro-bunching and coupling it with the high repetition rate of an accelerator, an SSMB ring can produce high-average-power, narrow-band coherent radiation spanning wavelengths from terahertz (THz) to soft X-ray. This novel light source holds tremendous potential for accelerator photon science and various applications in industries like EUV lithography, which is used for high-resolution imaging and manufacturing processes.

SSMB-EUV could have tremendous potential for EUV lithography’s high-resolution imaging applications in chip manufacturing. The public interest stems from hopes of getting around US blocks, commercializing Tsinghua’s research, and advancing domestic chip fabrication, further buoyed by Huawei’s recent 5G chip self-sufficiency.

Currently, the most advanced EUV lithography machine made by ASML in the Netherlands employs a CO2 gas laser with a power greater than 20 kW to bombard liquid tin, creating a plasma that generates Extreme Ultraviolet (EUV) light at a wavelength of 13.5 nm. Through continuous optimization, the machine achieves an EUV light power of approximately 350 W at the intermediate focal point.

However, this technology has reached its upper limit in terms of light power, which is around 500 W, and cannot support further development of next-generation lithography technology.

The SSMB (Steady-State Micro-Bunching) EUV light source offers several advantages：

A storage ring design for steady-state microbunching to generate EUV radiation Photo: Tsinghua University

In the end, what the Chinese scientists are actually working on is not an EUV machine, but rather a whole new type of EUV lithographic complex that could produce leading-edge chips in high volume. The SSMB concept was proposed in 2010 by Stanford professor and Tsinghua distinguished visiting professor Alexander Chao and his Ph.D. student Daniel Ratner. Chao has continued promoting SSMB research and international collaboration.

A group from Tsinghua University, Helmholtz-Zentrum Berlin and Physikalisch-Technische Bundesanstalt experimentally tested steady-state microbunching at the Metrology Light Source, a synchrotron in Berlin.

In 2017, Tang Chuanxiang and Chao launched the SSMB experiment. Tang’s team and their German partners led the theoretical analysis, physical design, and laser system development, achieving initial proof of principle. Their seminal paper was successfully published in Nature in February 2021.

Recognizing SSMB-EUV’s importance for lithography R&D, China has since invested to construct a scientific facility in Xiong’an dedicated to SSMB research.

So SSMB-EUV is scientifically validated and engineering efforts are underway. The research team is committed to realizing SSMB-EUV, with ample funding approved. However considerable obstacles persist before industrial EUV lithography with SSMB can be achieved. Tempered expectations are prudent – the LPP-EUV approach took over 20 years from concept to implementation.

Nonetheless, under US pressure, China is compelled to aggressively pursue creative solutions like SSMB-EUV, combining ingenuity and engineering. With sufficient perseverance and resources, breakthroughs may not be so distant. This could be a game changer if China makes it work.