Ushio has received the first acceptance of its laser-assisted discharge produced (LDP) EUV light source. After 15 years of development, the company may earn a place in the EUV arena after all – in the mask shops, not the fab, though.
In 2008, Japanese specialist lighting company Ushio acquired full ownership of XTreme Technologies. The German company, originally a joint venture between Lambda Physik and Jenoptik Laser, supplied the light sources for the first two EUV scanners ever built: ASML’s so-called alpha demo tools, which were installed at Imec in Leuven and CNSE in Albany to probe the new lithographic technique. Joining forces with Philips Lighting, Ushio aimed to scale up the demo technology to (much) high power levels.
To produce EUV light, the partners bet on a technique called laser-assisted discharge plasma (LDP), which works by having two discs rotate in a bath of liquid tin, covering their surface with the metal. The film is evaporated with a laser and subsequently ‘ignited’ by an electric discharge. The resulting plasma emits EUV light.
LDP had to compete with the laser-produced plasma (LPP) method championed by the American firm Cymer. With LPP, falling tin droplets are shot with lasers used to create the EUV-emitting plasma – no electric discharge involved.
EUV front-runner ASML, which builds the scanners the source is intended for, worked with both the Ushio/Philips duo and Cymer for a while. However, ASML CTO Martin van den Brink eventually became convinced LPP was the better choice long-term and intensified the relationship with the American company. This wasn’t a good omen for Ushio.
But Cymer dropped the ball. It kept missing deadlines, forcing Van den Brink to order a few LDP sources for preproduction tools – the first EUV scanners to be installed at customer sites. This revived hopes at Ushio that it still had a shot at the prestigious business. By then, in 2010, Philips had already given up and had sold its EUV activities to the Japanese company.
ASML CTO Martin stuck to his guns, however, despite the delays. “Cymer’s execution is really disappointing,” he admitted to Bits&Chips in 2011, but he stayed confident LPP was the right choice long-term. ASML initially just sent ‘assistance’ to San Diego, yet moved to acquire Cymer in 2012 (link in Dutch). The Veldhoven equipment maker cherishes its supplier-oriented business model, but for “this exceptional piece of technology it doesn’t work,” a spokesman explained.
It took another five grueling years or so to get the EUV show on the road, but the Cymer acquisition ended Ushio’s dream, it seemed. Except not quite: it retreated from Aachen to its home base in Japan, yet did not stop development. Instead, it switched applications. Rather than building equipment to feed EUV scanners their precious light, the Japanese firm decided to focus on another application: sources for mask inspection.
EUV mask blanks – EUV masks before they have a pattern – and the masks themselves have to be near-perfect at the atomic level. Since even the slightest defect or speck of dust can deform the pattern projected on the wafer, the ability to inspect blanks before patterning them or masks before using them in production is rather important.
For years, the lack of source power was put forward as the biggest roadblock for the introduction of EUV in commercial IC manufacturing, but as ASML and Cymer got their affairs in order, the infrastructure surrounding EUV technology became an increasing cause of worry. With all the delays in EUV development, many companies had adopted a wait-and-see attitude. Not before they were sure EUV would make it, they would start investing.
At industry conferences, presenters would typically use traffic light colors (green for ready, and so on) to convey progress. While progress was made on, for example, mask blank inspection and the pellicle (mask protecting device), the very last item to have a red light next to it was mask inspection related.
To be more precise, the problem child was actinic mask inspection, meaning inspection using the same wavelength of light as is employed during the actual production process. For years, there was discussion about whether actinic inspection is absolutely necessary, as a combination of various other methods – optical, x-ray and e-beam – could potentially deliver the same results. The common view today is that it isn’t absolutely necessary, but nonetheless very much desirable to increase yield and decrease costs. This is why chipmakers have started employing EUV in their fabs, even though actinic inspection wasn’t available.
Until recently, that is: Ushio repurposed its partially developed source as an inspection tool, which recently got thumbs up from a customer, a manufacturer of inspection equipment. “Testing equipment was one of the things holding back EUV. With that piece in place, production efficiency and yields can go up,” Ushio CEO Koji Naito told the Japan Times. The EUV business will start contributing to his company’s profit next fiscal year, he added.
After fifteen years of development, that would be well-deserved.