Integrated circuit (IC) manufacturing involves complex processes and may
require months to complete. Thousands of messages will be generated during each process,
and most messages can easily be identified and analyzed. However, ambiguous information remains as a kind of tacit knowledge that is one of the most essential issues of R&D
management. Integrated innovation is an application of scientific/technological creative
solutions to complex processes. This research uses a case study of a semiconductor company in Hsinchu Science Park in northern Taiwan. Traditional methods only yield the
results inferred from explicit knowledge, and omit the results based on tacit knowledge.
The integrated method can be designated as a clear direction for the R&D which uses
the multivariate statistical analysis as virtual sensors. By involving Hotelling T2, and
the principal component analysis (PCA), to generate specific results corresponding to the
core of the high density plasma chemical vapor deposition (HDP CVD) equipment or
process, eliminate inaccurate information using the experience rating in a 12-inch fab.
This provides an approach that can guide the R&D engineers and illuminate the entire
process. In sum, both process stabilization and cost savings are the major advantages of
virtual sensors.
Relation:
International Journal of Innovative Computing, Information and Control, 14(4), p.1371-1388
