So, what’s the minimum pitch found on a 45 nm half-pitch process?

No, it’s not a trick question. And yet, surprisingly, the answer is not necessarily obvious. I’ve talked to a few people and it seems that the phrase “XX nm half-pitch” is becoming a name, not a numerical description. We’ve known for years that node names, once themselves descriptive of the minimum half-pitch of the chip, are now the sole purview of marketing departments and can mean just about anything. That’s why people began to specify minimum half-pitch. “Here are the results from our 32 nm node process, which has a 45 nm half-pitch …” But if a company quotes the half-pitch to describe a process, what happens if that company decides to relax the minimum pitch due to lithography challenges? The easy solution is to relax the pitch without changing the name you’ve given to that process. We are starting to see “45 nm half-pitch” processes with a metal 1 (minimum) pitch of 100 or 105 nm. We in this business are nothing if not innovative.

This is a short blog today. Too many hospitality suites last night.

Last year SPIE published a book I wrote called the Field Guide to Optical Lithography. No, it doesn’t contain any illustrations of the plumage of the male lithographer during courting. SPIE picked the title – it is one of a series of field guides on optical topics. Of course, they are selling the book here this week, but I was a little surprised to see their method of advertising – a pair of 8 foot tall posters each with a giant photo of my graying, smiling head. I find them a little disconcerting to look at, so I can only imagine their impact on the general lithography book-buying public. Mark Smith threatened to take a marker and write “actual size” on the posters next to my picture. Isn’t that what friends are for?

It’s always fascinating to see which sessions are hot and which are not. Last year, a session entitled “polarization and hyper-NA effects” would have been packed. This year it was empty. Not only that, but all of the papers were essentially the same (how many ways can we rewrite the Jones matrix?). The industry tends to work like that – as a mob, recognizing and working on the same problems at about the same time. We spent the year working on vector pupils, now we are done. The ‘hot’ topic this year? Double patterning, of course. Standing room only during that session.

My favorite papers so far:

On Monday in the metrology session, Tim Brunner put scattering bars too close to a feature in order to do what lithographers never want to do – make the NILS low. The result – a pattern with extreme sensitivity that can be used as a focus and exposure ‘canary’ monitor.

In the resist session on Tuesday, John Biafore explained some of the stochastic phenomena behind line edge roughness – including totally cool movies that illustrated the main effects in a way an equation never will.

Bernd Geh made sense of the Jones matrix on Tuesday in the optical lithography conference – something I thought couldn’t be done.

The first day of the Advanced Lithography Symposium began, as it always does, with the plenary session. David Williamson, possible the world’s foremost catadioptric lens designer, won the 4th Zernike Prize for Microlithography. Much deserved. And after Burn Lin, Grant Willson, and Tim Brunner, it’s good to see that someone who didn’t work for IBM is deserving of this honor.

The plenary talks began with Hans Stork, CTO of Texas Instruments, describing the challenges of scaling CMOS down to the 32 nm node. Surprisingly, though, after giving a very good description of the challenges, he failed to mention the TI strategy for addressing those challenges. Of course, that strategy became well known a month ago when TI announced that it was laying off its R&D staff and relying on foundries for manufacturing below 45 nm. There were over 1000 people in the room for that plenary talk, and one very large elephant. I thought it was important to point to the elephant.

George Gomba then gave a very nice sales pitch for the IBM lithography cabal. I have to admit that I like the new buz-phrase “computational lithography.” I found it interesting that the logo for their lithography efforts was a single tear.

Mark Melliar-Smith followed up with a sales pitch for nanoimprint lithography as a viable alternative for 32 nm half-pitch CMOS manufacturing. On one of his concluding slides I counted up about 5 – 6 orders of magnitude improvement required to make imprint work at this node. That’s a lot, but I think it is still less than the relative orders of magnitude improvement required for EUV.

The sessions began after a coffee break, and immediately there were multiple talks I wanted to be at at the same time. Welcome to SPIE. And it seems like the meeting just keep going. At 9pm I caught the tail end of a panel discussion on double pattering, “Twice the gain for twice the pain?”. It reminded me of the old story of the frog in a pot.

Throw a frog in a pot of boiling water and he will feel the pain and jump out. But put him in cool water and he is happy. Now if you slowly start heating up the water, the frog doesn’t notice the rising temperature until it is too late, and he is cooked. So my question for everyone working on 65 nm, 45 nm, and 32 nm half-pitch lithography solutions: how hot is it now?

It’s that time of year again, when a large proportion of the world’s lithographers make the annual trek to San Jose to attend their biggest conference of the year. Attendance is expected to surpass 4,000 – and that’s a lot of litho engineers. I hope the area bars are well stocked with beer.

2007 promises to be a very interesting year for lithographers. Two EUV “alpha demo” tools (whatever that means) were installed last December, so I suspect we’ll learn a great deal, good or bad, about how much work is required to make EUV practical. We’ll probably also see the next hyper-NA installment in 2007 – tools with numerical apertures in the range of 1.3 to 1.35. Is this the highest we can go? And if so, is this finally the end for optical lithography? But wait, what about double patterning? I think 2007 will be the inevitable let-down year for double patterning, a depression of reality after all the hype and expectations from last year. Then, following the normal cycle of emotions, in 2008 we’ll either figure out how to make double patterning work, or give up on it. The future of advanced lithography is cloudier than I ever remember seeing it before.

And so people come to this conference looking for answers. But people come to Advanced Lithography for many reasons. TI researchers will come looking for jobs. Freescale lithographers will be looking for recommendations on where to live in New York. Intel will come to listen, but they won’t talk. And of course, the suppliers (oh, there are soooo many suppliers) will come looking for business. What am I looking for this week? As always, two things. First, community. I’ve come here to reconnect with the people I see each year at this event, many of whom I am proud to consider my friends. And second, inspiration. The collective creativity of the men and women presenting papers at the conference is absolutely astounding. Each year I come away with far more ideas of things to do than I can possible attempt in one year.

And that’s what this conference is all about – a community of scientists and engineers taking inspiration from each other. As each of us runs as fast as we possibly can towards the brick wall at the end of Moore’s Law, its good to know we’re not alone.

On July 4 of last year my friend (and fellow lithographer) Grant Willson was hit by a drunk driver – a head-on collision that crushed his hip, among other injuries. Many people have asked me for an update on Grant’s recovery. It started off very slow. He didn’t heal well from the major reconstructive surgery on his hip. Grant spent a lot of time in a wheelchair, fighting pain and an infection. Finally around Thanksgiving he had hip replacement surgery. The impact was dramatic – he was like a new man almost immediately. When I saw him on New Year’s eve, he had barely a perceptible limp and was in the best of moods. Recovery will soon be complete.

What about the drunk driver who hit him? It was his third offense. He received a seven year prison sentence, and will be up for parole in five years.

Last Thursday, I found myself in Bellingham, Washington. Of course, no one goes to Bellingham – I just happened to be driving from Seattle to Vancouver, British Columbia (to attend a philosophy of science conference – I know, I have strange hobbies). Besides being a little bit hungry, I had no reason to stop and wouldn’t have given the exit from the highway another thought until I was struck by a nagging familiarity with the name of the town. Bellingham. Then it occurred to me – this was the home of SPIE, the optics professional society that hosts the two largest conferences in the semiconductor lithography world (BACUS, the former home of world-class geek entertainment, and Microlithography, now dubbed Advanced Lithography so that no one will think we are doing any of that mundane kind of microlithography we used to give papers on).

Bellingham, Washington. The place we Fed-ex our extremely overdue manuscripts in the hopes that they might still be included in the proceedings. The return address on the advanced program that we get in the mail every year telling us about the few papers we will go see and the many, many papers we will miss at the massive Microlithography (sorry, Advanced Lithography) symposium. The place where we assume real people sit and talk to us on the phone when call with a question, but of course we are never sure.

SPIE Headquarters. I don’t know anybody whose has ever been there. So I decided to go. I veered off the highway at the last minute, searched and finally found the very well hidden building that was nestled in the hills and trees of this small coastal Washington town. I asked for the people I know there – June Thompson, who used to manage the BACUS conference, and Brian Thomas, who spear-headed the effort to upsize Micro into Advanced. Luckily, they were both in town. June gave me a tour of the building (it’s very nice!), then indulged one of my (mostly) harmless obsessions and took me to a brewpub for lunch (I can say with confidence – there’s good beer to be found in Bellingham). I exchanged gossip with Brian (though, quite frankly, lithography gossip is pretty boring) and lent him a sympathetic ear as he complained about the problems running SPIE’s most successful conference. All in all, a nice time.

So if you find yourself driving up interstate 5 through Washington some day with a little time on your hands, stop by and say high to the folks at SPIE. They’re nice people, and they get lonely.

David Pan, a friend of mine and professor at UT Austin, told me recently “Moore’s Law isn’t dead – it’s just having a mid-life crisis.” Although I wrote an article a few years ago provocatively titled “The End of the Semiconductor Industry as We Know It”, I think David is right – our industry is in mid-life crisis. I can relate.

Gordon Moore penned his famous law in 1965, observing a doubling of the number of the transistors on a chip since the birth of the integrated circuit in 1960. Now it just so happens that I was born in 1960. I certainly don’t claim any cosmic connection to continuous semiconductor improvement due to this coincidence of birth dates, but it does mean that me and the technology driver of the Information Age share at least one thing – we are both getting old.

Now I certainly don’t feel ‘old’, or that my useful days are behind me, but I’m not young either. I can’t pull all-nighters anymore, they way I used to when I could start and finish a conference paper 12 hours before I gave it. I’m unwilling to put my life on hold when a customer calls and says he needs something yesterday. I can’t work in the fab – that’s a young person’s job. And yes, the cries of “mid-life crisis” could be heard from all of my friends when I bought that Lotus sports car last year. I’m definitely older, but I like to think that I’m wiser too, and that this wisdom is more than enough to make up for a little slowness in step. But is the same thing true of the IC industry? It better be, or things will get pretty ugly fast. Working harder and faster because we have to keep with Moore’s Law is not good enough any more. The IC industry took off because the early pioneers took the science of semiconductors and turned it into technology. For that technology to keep going, we’ll have to bring in a whole bunch of new science. Most of that science will come from the universities, unlike in the past when most innovations came from the IC companies. Increased support for univeristy research is needed now, and hopeful it is not already too late.

Moore’s Law is getting old – let’s hope it gets wise as well.

I just got back from visiting Grant Willson in the hospital here in Austin. One week ago, while driving to pick his wife up at the airport, he was hit head-on by a drunk driver.

Almost everybody working in the field of microlithography knows of Grant Willson. While at IBM he invented chemically amplified resists and has subsequently received every award I can think of for that work, including SPIE’s Frits Zernike Award for Microlithography, our field’s highest honor. For more than a decade he has been teaching at the University of Texas at Austin and his irreplaceable leadership in the education of scores of students has produced results that may even eclipse his scientific contributions. For those of us lucky enough to know Grant personally, he is affable, giving, and a man of the highest possible integrity.

The accident banged him up badly, crushing parts of his hip. When I saw him last Thursday, less than two days after the accident, he was in good spirits, but in obvious pain. Today he was much better, and he should be heading home soon (in typical Willson fashion, ahead of schedule). The hope is that after 2 – 3 months of physical therapy (learning how to walk again) he’ll be right as rain.

Best wishes for a speedy recovery, Grant.

If you have poked around my site much, you may have noticed a section on the history of semiconductor lithography. It still needs a lot of work, so anyone who has any good information to share, I would appreciate it.

In particular, I’m trying to find out the year and circumstance around the purchase of the David W. Mann Company (maker of photorepeaters for mask making) by GCA (who eventually turned the technology into wafer steppers). Any and all help would be appreciated.

The SPIE Microlithography Symposium is without question the premier annual conference in the field of semiconductor microlithography. But all is not well in litho conference land. Many of the papers are simply not worth listening to. Of course, with any event this big you have to expect a range of quality in technical papers – to get the good one must accept the bad and the ugly. As the conference has grown over the years, the very good papers have stayed very good. But the bad papers have gotten worse, and the average quality of papers at the conference has steadily declined as the conference has grown. The reason for this is clear to me: an increased influence of sales and marketing goals over technical goals. With the conference’s growth in size has come a growth in influence, and a desire by many to control that influence.

What can be done to fix this problem? I’ve written a short whitepaper, A Modest Proposal, with concrete recommendations that I believe can improve paper quality. If you disagree, please let me know. If you agree, please let the conference organizers know.