| Summary: | 碩士 === 國立臺灣科技大學 === 機械工程技術研究所 === 86 === The purpose of this article is to discuss the feasibility of applying the Chemical-Mechanical Polishing (CMP) to the Ultra-Large Scale Integrated (ULSI) flatting process. This paper investigates some questions, i.e. the "Machining", "Process Parameters Optimization", "Prediction", and the "Process Integration" the might be important for a better understanding of this application.
In the beginning, we did the first research of the CMP process Machining Parameters by using the Analysis of Variance (ANOVA), finding out the obviously important role of the Process Machining Parameters in the CMP experiments. Then, with the Grey Relation Analysis, we had the advanced study and experiments to exactly know the features of CMP Process and Parameters. In the way, we can make sure the main effects among Process Machining Parameters and then make order. Followingly, we could get the huge and complicated relations among Chemical-Mechanical Removal Parameters (for example: Down Force Pad Speed, Carrier speed, Oscillation etc.) Removal Rate, Non-Uniformity and Surface-Roughness. Based on the results of Analysis of variance and Grey Relational Analysis, the experimental outcomes, we used the Optimization method to approach the purpose of Process Parameters Optimization of the Chemical-Mechanical Polishing. Furthermore; we motivated the Chemical-Mechanhical Polishing. Furthermore; we motivated the Adaptive Neuro-Fuzzy Inference System (ANFIS) in order to lessen the experiment times making correct predictions and Predicating the best Surface Roughness. At last combining the Adaptive Neuro-Fuzzy Inference System (ANFIS) and Optimization Theory we hoped to get the purpose of prediction the best Process Parameters.
After the experiments, We proved that the results of the best Process Parameters match the Optimization Theyory and Adaptive Neuro-Fuzzy Inference System (ANFIS). Therefore, we could roughly predict the best accurate Parameters. Through this study, we can then precisely have the CMP process parameters.
|
|---|
