| Summary: | 碩士 === 中華大學 === 電機工程學系碩士班 === 102 === The semiconductor industry is very important and this industry always have new technologies rapidly and evolving. Within decade ago, just as predicted by Moore's Law, as [3], the chip size continue to be required to reduce its size and also quickly increase integrated circuits (IC) number of transistors on the chip, the IC manufacturing process technology has become more complex.
In response to the high demand of electronic components in this booming electronics industry, in addition to the semiconductor wafer mass production, the process becomes sophisticated and demanding miniature. Meanwhile, process defect detection becomes the key elements which will dominate the whole process generation development. This paper will focus on chemical mechanical polishing (CMP), . the key process of IC fabrication, how to detect the abrasive defects on pad surface effectively and these defects on the pads will caused not only semiconductor wafers unnecessary yield loss and wafers scrapped but also semiconductor production plant capacity drop and unnecessary losing the resources. Therefore, the surface of the polishing pad related defect detection technology has become an important research topic, based on the search of relating literature, we can understand, so far no such CMP pad defect detection procedures has been built, on the contrary, due to the cost concern, the visual way to check the pad quality is still the mainstream of the production line quality control(QC). Currently, the QC of CMP pad detection couldn’t be avoided with "people". The challenges were encountered and influenced with fatigue, the environment, and the emotional stress.
This study attempts to identify a new way which is from the pad’ surface image to identify the surface defect of polishing pad. This method could propose an efficient, low-cost and high recognition rate with excellent automatic identification for pads’ surface imperfections, moreover, the fast and effective identification and classification of defects could increase its true positive rate (TPR) of up to 87%. And the computation time is about 10 seconds of the experimental test results, this study has been proved of polishing pad surface defect identification procedures in the industry, and provides a systematic and effective identification technology.
|
|---|
