Approach to the clustering modeling for the strong correlative control measurements for estimation of percent of the suitable integrated circuits in the semiconductor industry

摘要

The problem of increase of the suitable things percent exists in any manufacture. It may be solved by different ways, both technological, and by the general methodology of quality control. In the paper, this problem related to the semiconductor industry with own features is discussed. In particular, there is a problem of increase of the potential achievable percent of the suitable integrated circuit at a line production and simultaneous experimental designing on the same equipment. The technological path consists of several hundred operations, so revealing of their influence degree to result is a complicated problem demanding application of nonparametric statistics. On the other hand, the facility of modeling should be robust and simply, that allow to any operator make a decision on admission of a half-finished item to the further technological operations. The cluster analysis as a simple technique of the processes in semiconductor industry is given on example of the real data from semiconductor enterprise. The independent variable is the percent of the suitable integral circuit, and other variables represent results of the intermediate control on all extent of a technological path. The developed approach is illustrated by analysis of data from semiconductor enterprise, so received results have a practical value for manufacture of semiconductor devices.