Corliss was talking about IBM achieving 34 wafers an hour from an ASML EUV scanner using a Cymer 40W light source.
“My hope is that the 40W will be doubled in the next few months,” said Corliss. “The goal is to have it at 80W.”
Why not use Gigaphoton’s light source, which is claimed to be 92W?
“Because it’s not yet integrated in a system,” replied Corliss. “System integration is the big deal.”
However, anyone’s light source could be incorporated into anyone else’s scanner, so there’s nothing to stop ASML using the Gigaphoton source except for the fact that ASML bought Cymer last year.
It has been reported that Cymer has achieved 100W in the lab, but the light source is still some way from being integrated into a working tool.
The general industry expectation is that 125W is needed for 60wph and 250W for 120wph but, if IBM is getting 34wph off a 40W light source, then there’s a chance that the industry can find an EUV which is cost-competitive with multi-patterning (the alternative to EUV) before 250W is reached.
“Six hundred wafers a day is enough to support technology node development,” said Corliss.
Asked why IBM is supporting EUV development – a programme it has funded for 14 years – when it doesn’t make scanners and may not be making chips, Corliss responded: “We have an interest in developing technologies for our own systems development.”