jasonclass@[EMAIL PROTECTED]
wrote:
<Summary: Some questions about systematic vs. random yield loss>
I think the distinction between systematic yield loss and random yield
loss comes largely after determining the _cause_ of the yield loss.
For example, particle defects, while (essentially) randomly distributed
across a wafer, would be considered a systematic yield loss mechanism,
because particle defects affect pretty much every wafer in a batch and
every batch in a fab.
Another example of a systematic yield loss mechanism would be processing
through equipment which is only marginally capable of doing what it's
being asked to do. When you send the material into the equipment, you
know that only (say) 70% of if will be good when it comes out. You just
don't know _which_ 70%.
In contrast, yield loss due to a misprocessing / execution error or
equipment failure would be random yield loss. You can't really predict
when somebody's going to make a mistake or when a piece of equipment is
going to fail. (Well, if you're careful and smart enough, you can
probably predict when equipment is going to fail, but then why would you
keep using it?)
You might also be able to categorize yield as systematic vs. random
based on the geographic dispersion of the yield loss. If the outer 1cm
of the upper left quadrant of a wafer is always bad, that's probably a
systematic problem, even if you don't know what's causing it. If yield
loss is uniform across the wafer (any particular die has a 1%
probability of being bad), then the mechanism is probably random.
Just my thoughts...
HTH
Bob Pownall
|
