Making EUV: from lab to fab

The dawn of EUV

The EUV story begins in the mid-1980s in Japan when, building on multilayer mirror research done in Russia in the 70s, Hiroo Kinoshita projected the first EUV images. Labs in the US and the Netherlands soon also began to explore this potential new development in lithography. Originally called ‘soft x-ray’ lithography, the name ‘extreme ultraviolet’ was inspired by the term used by astronomers for the same light wavelengths and photon energies.

In lithography, using shorter light wavelengths enables chipmakers to shrink the size and increase the density of the features (or transistors) on a chip, making the chip faster and more powerful. When ASML was founded in 1984, the industry was using mercury-vapor lamps that produced light of 436 nanometers (nm), known as g-line, and later, ultraviolet (UV) light of 365 nm, called i-line. Early EUV researchers pursued several wavelengths ranging from 4 to 40, but eventually settled on 13.5 as the sweet spot for generating EUV light with tin plasma.

EUV wasn’t the only technology that researchers explored to enable future generations of ‘shrink’. Electron beam lithography and ion beam lithography seemed to be other viable options, but ASML placed an ‘educated bet’ on EUV lithography, because it was the technology that seemed best suited to continue transistor scaling while remaining affordable in mass production.

Not everyone was immediately sold on the idea of EUV technology, however. At a 2020 SPIE conference retrospective, Hiroo Kinoshita, who had then been a researcher at NTT, described the challenges of convincing his fellow scientists that EUV lithography had a chance. “[I presented my results] at the yearly meeting of the Japanese Society of Applied Optics in 1986,” he said. “Unfortunately, the audience was highly skeptical of my talk. However, my belief did not change.”

Andrew Hawryluk, long-time semiconductor industry veteran and then-researcher at the Lawrence Livermore National Laboratory, recalled similar frustrations at the conference. He explained how in December of 1987, a professor came to visit him and his team. After learning about their groundbreaking research in EUV, the professor asked, “But can you actually do anything useful with this stuff?”

“His words haunted me,” said Andrew, who went home for Christmas break and returned two weeks later with a 30-page whitepaper on EUV lithography. He and his team presented the paper at a conference later that year, but even then, finding support was difficult. “You can't imagine the negative reception I got at that presentation,” Andrew recalled. “Everybody in the audience was about to skewer me. I went home with my tail between my legs vowing never to talk about EUV lithography again.”

 

But a week later, Andrew got a call from William (Bill) Brinkman from Bell Labs, who subsequently succeeded in getting the US Department of Energy to fund national programs for EUV lithography research at Lawrence Livermore and Sandia Laboratories, in what came to be called the ‘Virtual National Laboratories’.

Meanwhile, in the Netherlands, Fred Bijkerk, a professor of EUV optics, projected the first EUV images in the Netherlands in 1990. It wasn’t until the late 1990s, however, that research into the industrialization of EUV lithography began in Europe. In 1997, ASML brought in Jos Benschop (senior vice president of Technology) to kick off our EUV program.

With Jos’s help, in 1998, ASML formed a European industrial R&D consortium dubbed ‘EUCLIDES’ (Extreme UV Concept Lithography Development System) with our long-standing partner the German optics manufacturer ZEISS and synchrotron light source supplier Oxford Instruments. ASML also worked with other suppliers that included Philips Research and CFT, TNO-TPD, FOM-Rijnhuizen, PTB (the German national metrology institute) and Fraunhofer-IWS. ASML and EUCLIDES joined forces with the American EUV LLC in 1999, while Japan pursued EUV technology development through their ASET program (Association of Super-Advanced Electronics Technologies) and later the Japanese Extreme Ultraviolet Lithography Development Association (EUVA) program.

In 2000, Jos was able to present the EUCLIDES program’s first results at SPIE. Having seen what was possible, in 2001 ASML assigned a small group of people and modest resources to build a prototype EUV system.

The team achieved their goal in 2006, when the first EUV prototypes were shipped to imec in Belgium and SUNY (College of Nanoscale Science & Engineering) in Albany, New York. There, the prototypes were used to better understand EUV and how this new technology could fit into the semiconductor manufacturing process. Things were starting to get real.

“We started work on this in 1997,” he says. “Suddenly, it’s 2010, and you’re flying overseas to South Korea. You take a long taxi ride, change clothes, go to the fab [factory], turn the corner, and there it is: the machine. That is a great moment. What started as a dream became reality in a customer’s factory.”

It was a great moment, but it still took nine years before our customers began releasing products containing EUV technology. “The machine basics were there, but we struggled to achieve the productivity that our customers needed in order to make EUV an affordable choice,” Jos explains. “It’s important to remember that the R&D phase at our customers also takes time,” says Jos. “After EUV’s ‘launch’, it still took one or two nodes for chipmakers to mature the technology and learn how to use it fully before commercial products could be built with it.”

Technology challenges along the way

“It took longer and was much more work than I imagined,” says Jos, looking back on the process. “You could say that we were either brilliant or just stubborn,” he laughs.

To make the jump to extreme ultraviolet light of 13.5 nm required innovations in practically all areas of the lithography machine, including the light source, the optics that shrink and focus the light onto the silicon wafer, and the reticle containing the blueprint of the chip to be printed. To make matters more complicated, EUV light is absorbed by almost everything, so it has to be produced and used inside a high vacuum environment. “The whole switch to vacuum was quite an adventure,” Jos recalls. “It made the materials a challenge due to outgassing – what materials you could use and how you had to handle them and clean them.”

“Our research really gained traction when ZEISS showed it could make specialized EUV optics,” explains Jos. Optics had always been the biggest anticipated challenge for EUV, but thanks to ZEISS, it quickly dropped off the list of critical issues.

What’s next? EUV 0.55 (High-NA) and beyond

In our quest to enable ever-smaller chip features, we continue to innovate and are now increasing our EUV machines’ numerical aperture (NA) from 0.33 to 0.55, which means that the optics in the new systems will allow light with larger angles of incidence to hit the wafer, giving the system a higher resolution and decreasing the size of the features that can be printed. “Because this new lens is fundamentally different, it meant that we had to come up with a completely new way of manufacturing it and measuring its features,” explains Jos. “Our teams at ZEISS and in Wilton are breaking one world record after another.” The first shipments of EUV 0.55 ‘EXE’ machines to customers for R&D purposes are planned for the end of 2023, and we expect them to be used in high-volume manufacturing in 2025.

But in 2022, our biggest challenge is ramping up production of EUV systems from 40 to around 60 systems per year. Hiring more people and building more cleanroom space are only part of the problem – the crux of the difficulty lies in ramping up capacity in our supplier network to meet the demand for EUV machines, according to Sheila Leenders, vice president of our EUV Factory. “Our supply base is facing the challenge of a lifetime,” she says. “We’re working hard to help them increase capacity and improve reliability where it’s needed at our suppliers around the world.”

Overcoming challenges through collaboration

Over the years, ASML pushed through many setbacks, unexpected challenges and delays. But Jos says that whenever they saw a new challenge looming, they always had ideas for how to tackle it. “Through our network of trusted partners, we always found ways to solve it,” Jos says.

“The way we approach any technology is by working through collaborative networks,” he continues. “This allows us to make use of the knowledge available and to progress with an ‘acceptable’ risk profile.”

Throughout all phases of EUV technology development – research, industrialization and now high-volume manufacturing – working together across borders has been key. Today, we procure some critical EUV modules from the US, such as the droplet generator for the light source as well as sensors, reticle handlers and stages, and other modules from Germany, such as our ZEISS optics and TRUMPF’s lasers for the EUV source, and we collaborate with companies and institutions all over the world in continued R&D for the next generation of EUV machines. Acting as the system architect at the center of things, ASML succeeded in building and mobilizing a global ecosystem to industrialize what many considered to be impossible technology. And we’re now confidently taking it into high-volume manufacturing.