To gain the necessary insight into process
variations occurring within the sub-micron
IC device structures, scientists and engineers
require high magnification digital images,
such as SEMs; similarly, to analyze a
specific equipment failure mode or manufacturing
process excursion, they require other
types of spatial data, such as defect
inspection results files. The procedure
to obtain, examine and analyze the images
and spatial data from hundreds of wafers
each day is time consuming, expensive
and prone to errors in human judgment.
At present, however, manually sorting
through all these data has been the only
way to determine the source of the problem.
Engineers try to identify the root cause
of the defects by recognizing a known
pattern or defect signature. Clearly,
what is needed is an automated and intelligent
method that quickly extracts meaningful
information from the data or image and
relates its spatial signature to a known
process excursion. However, the integration
of high magnification image analysis and
spatial signature analysis into yield
enhancement systems has been limited by
the complexity of automatically extracting
critical process attributes and signatures
from the spatial data.
|