The 20-year journey to the chips of tomorrow

EUV lithography turns a corner in 2016

Along with our 2016 financial results released today, we noted how Extreme Ultraviolet (EUV) lithography turned a corner this year. The year 2016 has turned out to be an important step further in our journey towards taking EUV to full production in the field. Following years of development and testing, our EUV systems have now reached levels of productivity that have given our customers the confidence to place a significant number of orders.

The semiconductor industry is now gearing up to use EUV lithography for high volume chip production. Introducing this completely new lithography technology will be one of the biggest technology transitions that the semiconductor industry has ever undertaken.

ASML’s journey to EUV lithography started in 2006 with the shipment of the Alpha Demo Tools ©ASML

It has been a long time coming, as the EUV story goes back nearly 20 years, to when long-term roadmaps were being drawn up by the industry in the late 1990s. To continue Moore’s Law, towards ever smaller, cheaper, more powerful and energy-efficient semiconductors, EUV lithography emerged as the most viable technology to pursue for the next-generation of chips.

ASML has led the journey to deliver EUV lithography, but like many breakthroughs, it’s been a story of co-operation and collaboration. Working together with suppliers like Carl Zeiss SMT and major customers, many new technologies and processes have had to be mastered, and challenges solved and overcome. It required close relationships and team work based on trust and mutual respect.

Initial research into the concept of using EUV for lithography had already started back in the 1990s and early 2000s, including at NTT in Japan, at the Bell Labs and Lawrence Livermore National Laboratory in the U.S. and at the University of Twente in the Netherlands.

An illustration of the Alpha Demo Tool (2006) ©ASML

Building on this early research, ASML developed two prototypes (known as Alpha Demo Tools) in 2006. They were sent to two research institutes — imec in Belgium and the SUNY Polytechnic Institute in the U.S. — so that researchers could test the fundamentally new system which involves a plasma source, reflective optics and a vacuum.
 
The resulting data was positive and promising enough to convince ASML to invest in the next phase of development in 2007. In spite of the prevailing economic crisis and even though the technology wasn’t fully established, ASML committed to building a commercial EUV tool.

The first TWINSCAN NXE:3100 machine was shipped to a major customer in Asia in 2010. It delivered on the key promise of EUV: the ability to image smaller features, one of the drivers that underpins Moore’s Law. But it soon became clear that the step from the R&D fabs of chip makers — where future generations of chips and process technology are researched — to the high-volume manufacturing fabs was a big one.

“You just have to tackle one issue at a time and then move onto the next one.”

Chip factories need to run 24 hours a day, seven days a week. So to be a viable new technology for chip manufacture, EUV systems not only need to process a large number of wafers in a short amount of time, but also do so continuously. What followed was a long, hard slog of solving all the engineering challenges this presented — it took longer and was harder than anticipated at the time.

The new technology required a completely new way of thinking. Some things were easier — for example, the vacuum removed the temperature fluctuations that can cause problems in other lithography systems. But many were more difficult, such as unforeseen chemical interactions in the vacuum environment. As a program system engineer, Henk Meijer had been involved in the EUV program from the early days: “We were trying to do things no-one had done before. It always felt like a bit of a gamble and we could have been overwhelmed by all the different puzzles we had to solve. But you just have to tackle one issue at a time and then move onto the next one.”

Sometimes the teams even felt like they were conquering the impossible. “We started researching a pellicle (protective membrane for the mask) for EUV”, continues Meijer. “This film has to be 1,000 times thinner than household plastic wrap and withstand being heated to 500 degrees Celcius without cracking. Initially we thought it wouldn’t be possible, but then we stumbled on the solution through a research project set up to solve a different problem.”

By 2012 there was no doubt that the EUV scanners could print chip features for the next technology nodes, but the systems were still not ready for full production. Moving from the research phase to the manufacturing phase turned out to be the most challenging step, and more R&D was needed. To accelerate the work, ASML reached an agreement with its three biggest customers, chip makers Intel, TSMC and Samsung, to contribute 1.38 billion euros over five years. ASML hired hundreds of engineers as a result.

Christian Wagner, program system engineer for the EUV light source ©ASML

In the same year, ASML also agreed to acquire partner company Cymer, which was developing the light source, to help accelerate progress. “We were working with their team in San Diego, and there was lots of drive to solve things, but we realized that we needed to add more resources to industrialize the source,” explains Christian Wagner, who joined ASML in 2001 and currently is the program system engineer for the light source. “No-one thought this issue would be so complex. To put it into perspective: there were only about 250 people working on the EUV light source in those days — now, there are over 1,000 people working together as one team in San Diego and Veldhoven.”

“One thing that amazed me over all the years was the team work. People are there for each other.”

At the end of 2013, the TWINSCAN NXE:3300B was shipped to customers, followed by the TWINSCAN NXE:3350B in late 2015. The mood in the industry was changing and confidence grew due to the improved results being reported — the systems were more reliable and could expose more wafers in just one day. ASML was focused on delivering 24/7 manufacturing, striving to reach the threshold of cost-effective production at 1,500 wafers per day.

An illustration of TWINSCAN NXE:3350B ©ASML

Then in 2016, EUV lithography turned a corner. Productivity and availability met levels that gave ASML’s customers enough confidence to place orders, showing their commitment to introducing EUV into volume manufacturing in 2018/2019. Until then, ASML will be driving the performance even higher, to the level required for volume manufacturing, while continuing to build up manufacturing capacity for EUV systems at ASML and in the supply chain, as well as the capabilities of its service organization.

It has been a long journey of highs and lows, and it sometimes threw up near impossible challenges. A constant factor has been the dedication of the teams and a belief that we could successfully overcome the technology challenges.

“One thing that amazed me over all the years was the team work,” says Wagner. “Maybe it’s because the challenge is so big, but people really help each other, they’re there for each other. And it’s very creative — so if you meet a problem, you get the right people together and you will get there.”
 
Important note for investors: Please read the information on forward-looking statements published here.