Re: Revolution in chip manufacturing!!

There's another outfit that is serious about going after ASML.

That's all I can say.

Resin-based 3D printing does amazingly well with bog standard 4k LCDs and UV LEDs, achieving 20 micron voxel size. Simon uses it for his automatically generated cradles for test jigs.

I’m a fan.

Chip litho has been better than that since the dawn of the planar process. You can easily do 5 microns with contact printing, even if you’re a bit cross-eyed.

A lot of the advances in lithography technology are based on using the very high contrast of photo resist to multiply resolution.

You expose using a dose that just exceeds the resist threshold to make a grid of narrow lines spaced by a bit more than half a wavelength. (Details vary depending on the numerical aperture and refractive index.)

Then you develop the resist, forming the narrow lines. Repeat once for 2x resolution, or three times for 4x.

Once you have this grating structure, you have to cut the lines in appropriate places to form the actual circuit outline. This is harder.

Also, of course you have to lay out your circuit so that this highly restricted geometry can build it. Not too easy, but that’s how chips have been made for many years now.

All of which is very fiddly.

State of the art litho is already using multiple exposures of 13.5 nm light. That’s a photon energy of 92 eV, which dwarfs the band gap of any material whatsoever. (Diamond is the champ at 5.5 eV.)

Thus there’s no way to make a LED with a wavelength that short.

A super high density array of super small 200-nm LEDs could be pretty useful if it wasn’t too expensive, but you ain’t doing maskless litho at advanced nodes with that.

There’s a _lot_ of money spent on litho R&D, So perhaps there’s a way to do it using an array of LEDs and multiple exposures, at least at nodes where 200 nm lithography still works.

Not going to replace advanced light sources and wafer scanners anytime soon, and anyway the extreme-precision motion and positioning will still be needed.

Might be great for making semi-custom chips based on arrays of gates or transistors.

Cheers

Phil Hobbs

I think that some of the first ICs were silk screened.

It should be possible to inkjet some power mosfets or something.

This sounds very interesting. FELs are well understood, but were not designed or used for IC production. Research only.

ASML did try all manner of deep UV light source approaches, but FELs didn't make the cut back then; don't know why, but accelerator technology has advanced since then.

Joe Gwinn