Advanced Methods for the Analysis of Semiconductor Manufacturing Process Data - Advanced Techniques in Knowledge Discovery and Data Mining - page 71

were not pertinent to the split, the selection is soundly based. The bases are
differences between the two lots, between single wafers in each lot, and even
variations with regards to the position on the wafer. Again this is properly
accounted for in the projections, further validating our approach.
In addition to these o ine analysis and knowledge-extraction methods,
dedicated classification techniques for online observation and potential con-
trol of the underlying process have been investigated. The feasibility of OCC
and the proposed NOVCLASS method for selective data storage could be
confirmed.
In this early stage of the work, the proposed methods were confronted
with actual high-dimensional process data from a practical but, in terms of
available samples N , small-scale problem. Most of the presented methods are
more sensitive to the increase in the number of dimensions M than in the
sample count N . Thus, it can be rightfully assumed that the methods will
scale well with larger databases.
Future work will emphasize the improvement of the visualization tool
and the integration of the algorithms and tools into the existing industrial
environment for meaningful large-scale method application, assessment, and
improvement based on more comprehensive data and data containing hereto-
fore unknown information on the process.
Acknowledgments
The contributions of Michael Eberhardt and Robert Wenzel to the QuickCog
System and Acoustic Navigator are gratefully acknowledged. Michael Eber-
hardt made part of this work feasible by contributing a data-converting tool
and adapting Acoustic Navigator to the task presented. Thanks go to Bernd
Vollmer and Christian Esser for friendly support and encouragement and to
Klaus Franke for providing the photographs in Section 2.2.
References
2.1 Aarts, E., and Korst, J., Simulated Annealing and Boltzmann Machines ,Ad-
dison Wesley, 1988.
2.2 Semiconductor Industry Association, International Technology Roadmap
for Semiconductors , Semiconductor Industry Association, San Jose, CA,
http://notes.sematech.org/ntrs/Rdmpmem.nsf, 1999.
2.3 Braha, D., and Shmilovici, A., Data mining for improving a cleaning process
in the semiconductor industry, IEEE Transactions on Semiconductor Manu-
facturing , 15(1):91-101, Feb. 2002.
2.4 Broomhead, D. S., and Lowe, D., Multivariable functional interpolation and
adaptive networks, in Complex Systems 2 , pp. 321-55, 1988.