[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Open Fabs

  From: "[email protected]" <[email protected]>

-------- Original Message --------
From: jim bell <[email protected]>
 ----- Original Message -----From: Zenaan Harkness <[email protected]>

On 7/29/15, [email protected] <[email protected]> wrote:

 >>That's the way things were about 25 or so years ago.  Wafers with photoresist were exposed with machines using UV, which over the years used ever->>decreasing  wavelengths of UV, in equipment:436. 365nm, and 248nm, eventually reaching 193 nanometer UV.  In the 1970's, entire wafer-masks (which >>covered the entire wafer) were used.  This became impractical as feature-sizes were reduced.  Step-and-repeat devices ("wafer-steppers) >>https://en.wikipedia.org/wiki/Stepper ;  then allowed exposure of a much smaller portion of a wafer.  (these eventually used mirrors, rather than lenses, >>because it is hard to process UV in a solid lens.)    >>However, because the wavelength of light eventually became a large portion of the size of a chip feature (a line or a space) it was increasingly difficult to >>'draw' the picture necessary to expose the resist on the wafer.  Due to many ever-more-heroic technologies, it eventually became possible to use 193 >>nanometer wavelength to expose features far smaller than the wavelength itself, which would have been considered phenomenal in the 1970's.  >>These days, EUV ("extreme ultraviolet") has been used for ever more small features.   https://en.wikipedia.org/wiki/Extreme_ultraviolet>>https://en.wikipedia.org/wiki/Extreme_ultraviolet_lithography ;  EUV is strongly absorbed by air, so such exposure is typically done in a vacuum.
>Is this using near-field optics? There also has been experimentation with soft x-rays using wiggler-type linear accelerators.>The differences between photons and electrons enables e-microscopes and EBL (which also operates in the vacuum) to avoid the optical-based >limitations.
I really don't know about modern EUV lithography.  That's why I cited the Wikipedia article, above.   It has a huge amount of information I haven't been keeping up on.  (I've had excuses...)   It is clearly extremely difficult/expensive to develop, which is in large part why they took so long to go from 193 nm to EUV's 13.5 nm.  Just reading that article is painful.  But they have to do it, because they want to get to feature sizes of 10nm and below.  It takes about 10 'square features' to make a DRAM cell.  A DRAM whose storage array is 1 cm^2 might, in principle, contain 100 billion DRAM cells.                 Jim Bell



  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cpunks.org/pipermail/cypherpunks/attachments/20150729/186f7067/attachment.html>