On Tuesday, the Optical Lithography conference begins, and with four of the five conferences going in parallel things are in full swing. Still, every time I walk into one of the cavernous meeting rooms that last year would have been standing room only and this year is more than half empty, I’m reminded of the difference that one year can make.

I began by sneaking into the Alternative Lithography conference (hoping no one would recognize me) to get on update on ASML’s EUV program. I found out that the throughput of the alpha-demo tool (of automotive junkyard fame) had been improved to 4 wafers per hour. But lest you become too excited about that speedy feat, the maximum throughput for a week was 100 wafers. When I talked a user of the tool, however, he said that last week was in fact a very good week: 42 wafers processed. But then, no one defined what was meant by a wafer. Because the flare on the tool is so bad, you have to separate the exposure fields by many millimeters to keep each field from influencing its neighbors. Thus, a “wafer” has on the order of 10 exposure fields (sometimes less) – several factors of two less than a product wafer. So how far off is the throughput of the alpha-demo tool compared to the beta tool spec?

I spent most of the day in the Optical Lithography conference. The highlights of the day were the keynote by Bruce Smith (he always gives a good talk) and the Chris Bencher paper on “gridded design rules”: how to scale SRAMs, and possibly logic, for another three or four generations using sidewall-spacer double patterning. After seeing that presentation I have become a fan of the sidewall-spacer approach – I think it will take us a long way.

My biggest complaint of the day was about session organization. After an entire session of talks by IBM on source-mask optimization, I was tired of the topic and IBM’s take on it. The Applied Materials session wasn’t as monotone (and wasn’t as long) but still was too much of one company. The conference organizers (you know who you are!) need to prevent such one-company sessions.

Tuesday night is usually a time for hospitality suite overload. Not this year. This pickings were sparse and they were soon picked over by hungry (and cheap) lithographers. I ended up spending most of the night at Gordon Biersch.