The atmosphere at the International Electron Devices Meeting (IEDM) in San Francisco was quite a contrast to the Facebook bash at its Menlo Park headquarters. Engineers in black suits listened intently to long, detailed presentations that typically involved advanced math.
There was a gravitas about the event that I liked and have felt before at top notch engineering gatherings. This diverse, global community maintains a civil dialog and respectfulness rarely seen in America these days.
They share a deep understanding of physics, chemistry, materials science and what it all means for the future of the semiconductors that give us iPhones and data centers that can run social networks for a billion people. They tolerate with kindness questions from former English majors turned reporters.
So I was glad to single out Kurt Ronse who leads the lithography group at the IMEC research center in Belgium. To my thinking, he is at the core of the core because he leads a team pioneering extreme ultraviolet lithography, the technology for making what could be the last few generations of CMOS chips.
EUV is famous for being delayed for years. Some say it may never work.
Ronse’s boss, ILuc van den Hove, said in an IEDM keynote that there are no fundamental problems with EUV, it is just an engineering challenge. Ronse said he winced to hear that characterization because he thought it made the work ahead sound easier than it is.
EUV systems need to turn out more than 100 wafers an hour to be useful. Today they turn out less than 20. Intel, Samsung and TSMC threw billions of dollars at the sole EUV systems maker, ASML, to acquire Cymer and create a new and better light source. Now the heat is on, but whether it produces the needed light is still an open question.
Even if it does, Ronse suggested, the industry may need to invest in EUV masks, too. There is no one big ASML in the mask world. A handful of smaller companies are struggling and complaining, Ronse said. Each one makes only a few EUV masks a year and sees no way to invest what’s needed to solve the EUV mask defect problems they face.
Maybe on his way to the airport, Ronse should have made a stop in Menlo Park. There’s a billionaire down there who probably doesn’t know how much he is betting this whole EUV thing will succeed.
Facebook likes wimpy cores, CPU subscriptions
Even with Intel's chips on the table, EUV still no sure bet