fun (if cynical) web site

They used to make all their content available after a blackout period, which was a lot more useful. They want $1k per year for a subscription.

The free stuff is also interesting.

Cheers

Phil Hobbs

Moore observed that transistor density grows exponentially. He didn't note that the difficulty grows exponentially too.

It didn't, for a really long time. Classical Mead-Conway scaling worked for many generations, so that the industry was basically lithography-driven, from crayons down to sub-100 nm. Transistor performance peaked at about 65 nm--they've been getting slower, leakier, and much more numerous since then. Analogue performance peaks at around

For a legal case a year or two back, I got to look at quite a lot of engineering drawings for 14-nm-class wafer scan tools. The amount of automatic tweaking and calibration in those things is unbelievable.

Cheers

Phil Hobbs

Was that DUV immersion? Maybe it used my laser controller!

One EUV scanner now costs $150 million. One big fab can cost $10 billion. The complexity is boggling. Each scanner must come with a team of support people to keep them running. Where are they going to find them?

And lots of spares!

It bogs my mind that steppers have become scanners. Both the reticle and the wafer are in motion during the exposure, with nanometer repeatability.

The shaped beam electron beam microfabricator that I was working on around

We were using a laser interferometer to keep track of where the stage was w ith nanometer accurcy, and we could move the electron beam around equally p recisely (and a good deal faster) - albeit over much smaller distances. We could pick up little squares of gold put on the wafer surface as position references - the electron beam worked just as well as an electron microscop e probe for imaging as it did as a writing tool. Stopping the wafer while w e wrote on it would have been a waste of time.

I'm not all that surprised John Larkin's mind gets boggled by 35-year-old t echnology - he's not exactly the best informed person around here.

The case was ASML v. Nikon in the International Trade Commission. I was mostly looking at Nikon's stuff.

That's a good deal easier than maintaining sub-nanometre geometric accuracy over a large field in a staring system, especially at a wavelength where you can't use lenses.

Using 1-D scanning you can take out most of the aberrations using dynamic adjustment of focus and tilt.

Cheers

Phil Hobbs