Database Reference
In-Depth Information
2.2.7 Experimental Data
From the proprietary software system and company database a liated with
the regarded manufacturing process, a subset of data generated for processing
two wafer lots with five measurement positions for each wafer was extracted.
A split of three, i.e., a partitioning of each wafer batch into three subgroups
for individual processing of each partition, was carried out during produc-
tion. Six wafers from the second lot were still staged in the fab for another
experiment, so no data were available for these wafers. The electrical test
data contain redundancies with regard to the particular split, as for each
device specimen, different channel length and width are available. Also, as
already explained, variation of the threshold voltage will influence further
parameters belonging to that particular device. By means of conversion to
an Excel spread-sheet and the application of a standard conversion tool, the
database is converted to the QuickCog system requirements. The QuickCog
system comprises all the methods discussed in this chapter, in particular
the interactive data visualization methods and tools. A first database of 220
vectors with 205 dimensions will be regarded in the following experiments.
It will be denoted by SPLIT in the following. The size of this database is
given by the typical wafer batch size of 25 times the five measurement sites
per wafer. However, the measurement values of six of the wafers from one
set were not available, which reduces the data from the expected 250 to 220
samples. Larger databases could only be generated if larger wafer batches
were made subject to identical split processing. With regard to the associ-
ated effort and cost, the aim of this work was to assess the applicability of the
regarded methods also for rather sparse data of this application. No general
limitation of the approach is implied by the choice of this practically relevant
problem, as the regarded methods themselves scale well for large database
sizes [2.23], [2.24].
Complementing the parameter data, class a liations were generated in
two files. A three-class file was generated, regarding split information only for
the complete database. The data labeled by this class file will be denoted by
SPLIT3 in the following. Additionally, a six-class file was generated according
to lot and split a liation of each wafer/measurement location. The labeled
data will be denoted by SPLIT6 in the following. Additionally, according to
the underlying lots the data have been separated into two databases denoted
by SPLITTrain3 and SPLITTest3 with three classes each, corresponding to
the underlying split of 3 of each lot. Finally, for the novelty classification
purposes, a training set was extracted from the first lot containing data only
from one split. This will be denoted by SPLITTrainOCC in the following.
Search WWH ::




Custom Search