The final day at PMJ had the papers most interesting to me (that is, more related to the use of masks rather than their fabrication). Here’s a few things I learned at the conference (if you’re not a serious litho techno-nerd, skip this bit):

• Mask makers have started to ask for resists that are less sensitive (shot noise has become a problem), after three decades of complaining that resists are not sensitive enough.
• The non-ideal behavior of a scanner has a noticeable impact on proximity effects, and thus on the behavior of OPC tools.
• EUV is hard to do.
• No one really understands line edge roughness.
• Double patterning sounds good to mask makers – they get to sell twice as many masks, but mask errors probably won’t be the number one problem facing the lithographer.
• EUV is really hard to do.
• At the 45 nm node, lithography will be the #2 problem facing chip makers (strain variability will be #1) – how disappointing!

Overall, the average paper quality at PMJ is about the same as at other SPIE-sponsored lithography conferences, but with fewer really bad papers and fewer really good ones. There was one presenter who seemed to randomly change slides, both forward and backward, without any relationship to what he was saying. A couple of papers had none-too-subtle marketing messages sprinkled between graphs of data and self-serving conclusions. And of course, there was the guy whose company wouldn’t let him present any data, so he removed the numbers from the axes of every graph (why did he even bother to present?). But I’ve come to expect this from a conference that caters mostly to industry, and the number and egregiousness of the violations of paper propriety were relatively small. Most papers described solid but small incremental advances – the kind of thing that has been pushing Moore’s law forward for 40 years.