Re: [Sci.nanotech] Re: End of lithography at 32 nm?

From: Jim Logajan (JamesL_at_lugoj.com)
Date: 08/30/04 

Date: 30 Aug 2004 00:40:28 GMT

erincss@aol.com (erincss) wrote:
> well how far can micrometer chips built in three dimensions go, if
> done using modern techniques?

Not very far.

Lithography and related technologies are modestly fast ways to duplicate 2-
D patterns from a set of templates. Doubling the number of features (or
length) along either of the first two dimensions (call them X and Y)
doesn't double the duplication time - the duplication time is somewhat
independent of the pattern complexity (or dimensional length). But building
into a 3rd dimension (call it Z) causes your building time to increase
roughly linearly with the length of Z. That's bad enough but even worse is
the fact that your yield is going to drop exponentially with increasing Z.
Suppose for example that the normal 2-D lithography results in 1% of your
chips having defects. Then by the time you've built 100 layers 63.4% of
your 100-layer chips will have defects.

Relevant Pages

  • Re: [Sci.nanotech] Re: End of lithography at 32 nm?
    ... > length) along either of the first two dimensions ... > doesn't double the duplication time - the duplication time is somewhat ... dissipation is already a huge problem in modern systems. ...
    (sci.nanotech)
  • Re: Spectrum loading problem
    ... the rest of the computer scaled to the same dimensions ... - imagine how much more powerful it could be, ... the chips and other electronics inside also don't take one ...
    (comp.sys.sinclair)