Comments on: SPIE Advanced Lithography Symposium 2018 – day 2

By: Semiconductor Engineering .:. Blog Review: Mar. 7

Semiconductor Engineering .:. Blog Review: Mar. 7 — Wed, 07 Mar 2018 17:08:31 +0000

[…] of this year’s SPIE Advanced Lithography Symposium. Check out the blogs on day three, two, and one for an example of the conference’s dominant […]

By: Sjoerd

Sjoerd — Fri, 02 Mar 2018 01:54:44 +0000

In reply to Bob. And on top of that also plot the total industry turnover and profit. And then you have Moore's Law. It's not only about investment, it's about investment versus rewards really. Of course all large companies emphasize the costs and the investments, but what about the rewards and the increasing company and share holder value?

By: Dan

Dan — Thu, 01 Mar 2018 17:13:28 +0000

Bob,
As a technologist (first SPIE paper given in 1985) turned entrepreneur/investor–with plenty of failures and success in both areas–I’ve learned that investing large amounts of money is rarely the best predictor of success–and almost never a good predictor of the best/winning product technical solution. More money is often an attempt to “outspend” a competitor OR to continue funding a product/project that has missed its original promised commercial aviability date. I suspect that the enormous EUV spending is the latter.

A more relevant graph would be a bar chart showing the amount of money spent (cumulatively by the whole litho eco-system) on each new litho technology that has been successfully implemented (e.g. scanners to steppers to Step-and-Scan, to DUV, to immersion, to DP, etc.) Of course, OK to normalize for size of entire semi industry.

Then create a second bar chart that also normalizes for either years or number of nodes where the technology is used. This would be a rough relative measure of Return on Investment.

Then plot where EUV (and imprint, if you like) might fit on these two charts. Feel free to add a outline above the EUV bar (and imprint) on additional spending required.

I might also include a neighboring bar that normalized for the size of the market (e.g. EUV might only find a home in CPU and Foundry, but not memory, and Imprint might only be NAND.)

In my suggested version, the EUV bars would still make the imprint data “almost indistinguishable to the x axis,” but EUV might also dwarf ALL the other successively implemented lithography technologies.

By: Bob

Bob — Thu, 01 Mar 2018 14:35:03 +0000

Generate a graph with the Y axis as billions of dollars invested and the X axis the year. Now plot data for both imprint lithography vs EUV lithography. It will look like you only added data for EUV, since the imprint data line will be almost indistinguishable to the x axis.

By: Dan

Dan — Wed, 28 Feb 2018 17:04:23 +0000

What about the Imprint papers on Tuesday?
Imprint seems to making excellent progress for Memory. Imprint may still have some issues to resolve–but LER/LWR is not one of them (since the resist is confined/contained in the mask.)
And I’m still looking for someone to take the other side of my bet ($10K?) that we’ll see a commercial/volume device made with Imprint before EUV.
And love your SPIE blog every year…