Canon Nanoimprint lithography semiconductor manufacturing system (5nm)

Re:Canon Nanoimprint lithography semiconductor manufacturing... by Karliss on Monday October 16, 2023 @02:14AM (#37896407)

How do you even create a mask of that size? With photolitography you can at least reduce feature size compared to mask size with the help of lens. But here you need a process capable of making 5nm features (or whatever 5nm actually means) to produce a tool for making 5nm features. Seems a bit like chicken and egg problem. You could do it using photolitogrpahy, but wouldn't that defeat the point?

Is there a manufacturing process which can produce features that small, and is cheap enough for producing masks, but not cost effective for the quantities of producing final product?

[ Parent | on HN ]

Re:Canon Nanoimprint lithography semiconductor manufacturing... by imtringued on Monday October 16, 2023 @02:22AM (#37896439)

Are your joking? They obviously use electron beam lithography to make masks.

[ Parent | on HN ]

Re:Canon Nanoimprint lithography semiconductor manufacturing... by mitthrowaway2 on Monday October 16, 2023 @02:51AM (#37896563)

I think the idea is that you only have to make far fewer masks than wafers, so you can make it slowly. I'm sure it's quite difficult to make, and Canon must have been working to get it right for quite some time.

Good to see they haven't given up in their competition with ASML.

[ Parent | on HN ]

Re:Canon Nanoimprint lithography semiconductor manufacturing... by orbital-decay on Monday October 16, 2023 @02:52AM (#37896573)

> How do you even create a mask of that size?

https://en.wikipedia.org/wiki/Electron-beam_lithography

It can easily make small features, the only issue is low throughput, insufficient for mass production. But it's widely used to make masks.

(however, "5nm" in this case means nothing)

[ Parent | on HN ]

Re:Canon Nanoimprint lithography semiconductor manufacturing... by adrian_b on Monday October 16, 2023 @03:01AM (#37896624)

There are no lenses for extreme ultraviolet.

The mirror system used to project EUV might be able to do a reduction of the mask image by a small factor, but even so, the size of the features on the mask is too small to allow the production of the masks by photolithography.

As another poster has said, since many years ago the masks are produced by electron-beam lithography, even for traditional UV photolithography.

Electron-beam lithography and ion-beam lithography allow the production of chips with features as small as they can ever be possible (due to atom size constraints), but they make the images too slowly for mass production (because the beam scans sequentially the surface of the wafer), so they can be used only for making masks or for making devices used in research projects, where the cost per device does not matter.

[ Parent | on HN ]

Re:Canon Nanoimprint lithography semiconductor manufacturing... by tullatulla on Tuesday October 17, 2023 @04:11PM (#37920874)

Well, you would need a template with 8x smaller features than a EUV mask (at least in one direction, 4x in the other). Ebeam lithography can in principle do this, but it's at the resolution limit and the shot noise will mean there will be a lot of defects.

A sufficiently defect free EUV mask can be worth >>1E6$, for a nano imprint template this would be much much more and it's lifetime probably a fraction. That's why nanoimprint litho is no alternative to EUV.

[ Parent | on HN ]