Intelligent control system of semiconductor manufacturing device and technology

Abstract

The invention provides an intelligent control system of semiconductor manufacturing device and technology. The system comprises a data preprocessing unit, an outlier detection unit, a supporting vector machine training unit, an intelligent monitoring unit and a continuous learning unit. The data processing unit comprises a preprocessing module and a clustering processing module. The preprocessingmodule is used for carrying out preprocessing on data acquired by the semiconductor manufacturing device and technology. The clustering processing module carries out clustering processing on the datawhich is preprocessed by the preprocessing module. The outlier detection unit carries out outlier detection on the data which is subjected to the clustering processing so as to obtain an outlier set.

Description

technical field [0001] The invention relates to the technical field of semiconductor manufacturing equipment and process control, in particular to an intelligent control system for semiconductor manufacturing equipment and process. Background technique [0002] In the process of semiconductor manufacturing and production, it is difficult to directly establish a relationship between the abnormal quality of semiconductor devices and the behavior of semiconductor manufacturing equipment and processes. In the existing technology, it is usually only possible to check and judge by technicians after the quality of semiconductor devices is abnormal. After formulating the next processing measures, each inspection and judgment of the abnormality of the semiconductor device is an iterative process of speculation and verification, and the technician only consults the relevant semiconductor manufacturing equipment and process measurement data when the quality of the semiconductor device i...

AI Technical Summary

Problems solved by technology

[0003] (1) The impact of human factors is large: In the above mode, the use method and effect of the test data of the conductor manufacturing equipment and process are determined by the experience and judgment of the technicians themselves, so the process results after the occurrence of abnormal events are very different. Large, and different engineers also have different differences with each other, especially the difference in process results between engineers with less experience and more experience is very large;

[0004] (2) Lack of coherent learning mechanism: the handling of each abnormal event is also a learning opportunity, which can provide help for future processing of related events, but due to the lack of system support, the historical data of each abnormal event processing, including abnormal event phenomenon analysis, processing Opinions and implementation effects are scattered in different spaces, which cannot form a continuous learning mechanism;

[0005] (3) It takes a long time to manually process data: technicians may spend several hours on the decision of an appropriate control parameter, which seriously reduces production capacity;

Images