Within the past few years, scatterometry has been embraced for many in-line measurement and disposition applications
in the semiconductor industry. Yet, there remains some hesitation to fully rely on scatterometry for advanced process
development, and instead to depend on CDSEMs and traditional failure analysis imaging methods (cross-section, TEM,
FIB) to provide CD as well as profile information. This paper investigates whether scatterometry can be used as a
suitable tool to supplement and sometimes replace XSEM metrology, and is an extension of the work from M.
Sendelbach et. al. entitled "Improving STI etch process development by replacing XSEM metrology with scatterometry"
from the 2005 SPIE Microlithography conference. A very large number of cross-sections were completed on the
scatterometry grating targets as well as in-line disposition target and compared to optical measurements of the gratings
with the purpose of decisively answering this question. The targets used for this work were etched 65nm node NFET
gate structures. Measurements from cross-sections and scatterometry were processed to understand the top CD, middle
CD, bottom CD and sidewall angle correlations.
The investigation led to some interesting results, such as the existence of significant variation within a grating. In fact,
there was enough variation to indicate that one or a few cross-sections may not represent the actual process or the
"average" state of an array of lines, making XSEM metrology a poor quantitative method when used for process
development. Scatterometry measurements of middle and bottom CDs show excellent correlation to cross-section
results of both the grating and disposition targets.