Abstract: A method for analyzing the effectiveness of polishing frequency and the number of polishing times on the polishing pads having different patterns and profiles while performing the chemical-mechanical polishing process on the wafers is described. This method is to convert the images of various patterns and topography of the chips and grinding pads into binary images, and then calculates the binary images by numerical matrix method, which only needs to calculate the modified model of the position changed and the frequency of grinding during the rotation and deformation of different patterns and topography during relative movement, and then uses overlay model of effective grinding frequency to predict the distribution of effective grinding frequency at a fixed period of grinding time under a set grinding path.

Abstract: The TRIZ decision process of the clustering method proposed by this invention uses the characteristics and invention rules from the contradiction matrix table resulting from massive quantities of patent inferences to find a similar or approximate character group and invention rule group of the physical meanings, and also applies statistics to calculate the number of display times of the groups to be the basic foundation. Apart from the number of display times, Bayes probability, fuzzy object oriented method and Bayes probability combined with fuzzy object oriented method can be used as the system. The reading value is utilized as a foundation for prioritizing the sequence of consideration for the groups, in which the system reading value constructed by different models gives designers lots of options to perform the reading, so as to acquire the undesired result features of the prioritized consideration.

Abstract: A method of building a set of experimental prediction model that requires fewer experimental frequency, shorter prediction time and higher prediction accuracy by using the advantages of combining the experimental data of Taguchi method and neural network learning is disclosed. The error between the experimentally measured result of photolithography and the simulated result of the theoretical model of near field photolithography is set as an objective function of an inverse method for back calculating fiber probe aperture size, which is adopted in the following Taguchi experiment. The analytical result of Taguchi neural network model of the present invention proves that the Taguchi neural network model can provide more accurate prediction result than the conventional Taguchi network model, and at the same time, improve the demerit of requiring massive training examples of the conventional neural network.

Abstract: A non-destructive method to inverse-calculate a fiber probe aperture size, and a prediction method of the simulation and fabrication profile of near field photolithography are provided. The error between an experimental result of the photolithography and a simulation result of the theoretical model of near field photolithography is set as an objective function to inverse-calculate a fiber probe aperture size that can match with the photolithography experiment and the theoretical model of near field photolithography. Finally, by comparing the fabrication profile of the photolithography experiment and that of the simulation result of the inverse-calculated fiber probe aperture size, it is verified that the inverse-calculated fiber probe aperture size is reasonable and acceptable.

Abstract: A method for analyzing the effectiveness of polishing frequency and the number of polishing times on the polishing pads having different patterns and profiles while performing the chemical-mechanical polishing process on the wafers is described. This method is to convert the images of various patterns and topography of the chips and grinding pads into binary images, and then calculates the binary images by numerical matrix method, which only needs to calculate the modified model of the position changed and the frequency of grinding during the rotation and deformation of different patterns and topography during relative movement, and then uses overlay model of effective grinding frequency to predict the distribution of effective grinding frequency at a fixed period of grinding time under a set grinding path.

Abstract: The TRIZ decision process of the clustering method proposed by this invention uses the characteristics and invention rules from the contradiction matrix table resulting from massive quantities of patent inferences to find a similar or approximate character group and invention rule group of the physical meanings, and also applies statistics to calculate the number of display times of the groups to be the basic foundation. Apart from the number of display times, Bayes probability, fuzzy object oriented method and Bayes probability combined with fuzzy object oriented method can be used as the system. The reading value is utilized as a foundation for prioritizing the sequence of consideration for the groups, in which the system reading value constructed by different models gives designers lots of options to perform the reading, so as to acquire the undesired result features of the prioritized consideration.

Abstract: A method of building a set of experimental prediction model that requires fewer experimental frequency, shorter prediction time and higher prediction accuracy by using the advantages of combining the experimental data of Taguchi method and neural network learning is disclosed. The error between the experimentally measured result of photolithography and the simulated result of the theoretical model of near field photolithography is set as an objective function of an inverse method for back calculating fiber probe aperture size, which is adopted in the following Taguchi experiment. The analytical result of Taguchi neural network model of the present invention proves that the Taguchi neural network model can provide more accurate prediction result than the conventional Taguchi network model, and at the same time, improve the demerit of requiring massive training examples of the conventional neural network.

Abstract: A non-destructive method to inverse-calculate a fiber probe aperture size, and a prediction method of the simulation and fabrication profile of near field photolithography are provided. The error between an experimental result of the photolithography and a simulation result of the theoretical model of near field photolithography is set as an objective function to inverse-calculate a fiber probe aperture size that can match with the photolithography experiment and the theoretical model of near field photolithography. Finally, by comparing the fabrication profile of the photolithography experiment and that of the simulation result of the inverse-calculated fiber probe aperture size, it is verified that the inverse-calculated fiber probe aperture size is reasonable and acceptable.

Abstract: A method for analyzing polishing frequency and number of polishing times for chemical planarization polishing wafer with different polishing pad profiles is disclosed. First, drawings of a wafer and a polishing pad are provided and then are converted into pixel arrays. Pixel arrays are processed to be black/white images. The black/white images are converted into binary matrices. The effective polishing frequencies of all points in the binary matrix are calculated. Following the calculated polishing frequencies, the coordinates of all binary matrices are redefined according to a displacement condition, and then new coordinates of all points and corresponding effective numbers of polishing times for a time increment are calculated so as to form an effective polishing times matrix for the time increment. Further, all effective numbers of polishing times within a total polishing time interval are added together.

Abstract: A method for analyzing polishing frequency and number of polishing times for chemical planarization polishing wafer with different polishing pad profiles is disclosed. First, drawings of a wafer and a polishing pad are provided and then are converted into pixel arrays. Pixel arrays are processed to be black/white images. The black/white images are converted into binary matrices. The effective polishing frequencies of all points in the binary matrix are calculated. Following the calculated polishing frequencies, the coordinates of all binary matrices are redefined according to a displacement condition, and then new coordinates of all points and corresponding effective numbers of polishing times for a time increment are calculated so as to form an effective polishing times matrix for the time increment. Further, all effective numbers of polishing times within a total polishing time interval are added together.