Research On Process Parameters Decision & Optimization For Function Ceramics In The Precision CMP

With the rapid progress of ultra-precision machining, researches and applications of function ceramics are continuous in-depth in the fields like microelectronics, information, etc. However, there is a big gap at home and abroad in the efficiency and precision of ultra-precision machining. Being one of the main methods of ultra-precision machining, chemical mechanical polishing(CMP) is in the situation of low efficiency, bad quality stabilization and high machining cost as the impact of the human factors on polishing process. Scientific decision and optimization ways have become the key point in the study of the CMP in order to reduce the influences of human factors. Thus, for the sake of establishing an effective process parameters decision support platform for precision polishing of function ceramics, it is urgent necessary to study the decision and optimization technologies of process parameters for function ceramics in the CMP. Based on the combination of orthogonal test, Taguchi method, BP neural network, the hybrid combination of case-based reasoning and BP neural network method, database technology, the following efforts are focus on the process parameters optimization, the surface quality prediction and the process parameters decision for function ceramics in the CMP. The main works in this thesis can be concluded as the following:1) Taking the surface roughness as evaluation criterion, the relationship among polishing velocity, polishing pressure, density of polishing fluid and surface roughness has been studied firstly by using the orthogonal test and Taguchi method. And then, process parameters have been optimized to achieve their optimal combination in the precision CMP.2) Taking polishing velocity, polishing pressure, density of polishing fluid as input and surface roughness as output, the mapping model for the simulation of the precision CMP of function ceramics has been founded based on BP neural network. And the model is trained with the datas collected in the step 1) in order to finish the establishment of the SiC prediction model, and realize the effective prediction of the surface roughness in the precision CMP.3) Based on the steps 1) and 2), the process parameters decision strategy of function ceramics with the precision CMP has been determined by the hybrid combination of case-based reasoning and BP neural network. And a scientific basis is supplied for the process parameters decision in the CMP.4) Referring to the technical manuals, experiences of engineering workers and experimental datas, the process parameters database for function ceramics in the CMP has been built by Microsoft software Access. 5) On the basis of the above efforts, the process parameters decision support system for the precision CMP of function ceramics has been developed by applying PowerBuilder software, which can realize the process parameters optimization, the surface roughness prediction and process parameters decision. Examples show that the prediction results are coincident with the experiments well.In conclusion, the researches in this paper provide a scientific solution to change the lag status of the process parameters empirically chosen in the CMP, and lay a theoretical foundation for the enhancement of the overall level of the home CMP technology.