It is the faults that are caused by defects, not the defects
themselves that cause yield loss.To make the best use of your yield
enhancement efforts, it is important to know the yield loss
associated with each defect type, not just the defect count.
The output of EYES gives not only the
yield of the chip, but also a breakdown of the yield into the yield
loss associated with different defect types. This information can
be used to assess the real cost of defects and help target defect
reduction efforts where they will be most cost effective.
Yield loss by fault type for the stacs_noimp
chip.
Where the fault type does not correspond to a single defect type
it can be split into one or more different fault types, associated
with a particular defect type. This allows the yield loss of a
particular defect type, with perhaps a know source, to be costed. A
yield prediction will more accurately indication of the scale of
the problem and can be used to estimate ROI from defect reduction
measures targeted at those defects.
For example, there may be two defect types that cause shorts
(classified as dome and pancake defects) and two defect types that
cause contact/via fails (resist bubbles and straight forward
"defects").
Yield loss by a finer classification of fault
type.