Research On SPC For Nested Process Parameters In The Semiconductor Manufacturing

With the fast development of technology, the increasing complexity of process and equipment, the continuous improvement of the electronic components, as well as the customer's rising demands and their higher expectations on product quality in the field of semiconductor manufacturing, enterprises are forced to strengthen their quality management and make further efforts to guarantee their product quality. The only way to guarantee the quality of the product is to implement advanced quality management method, to carry out the statistical process control and to make sure the stability of product development and production process.In this thesis the statistical process control for nesting process parameters is studied based on the nestification of process parameters of batch process and wafer foundry during semiconductor manufacturing, and the unavailability of the conventional statistical process control chart in this case with the method of theoretical analysis and MATLAB simulation. The existing first order nested control chart was thus improved and a two order control chart is also developed. The research results of this thesis will benefit the improvement of the product quality of batch process and wafer foundry during semiconductor manufacturing, reduction of the quality defect, and they have a profound influence in boosting company's market competitiveness..The main contents of this thesis include:1) Analysis of the causes of the nestedness of process parameters in batch process and wafer foundry during the process of semiconductor manufacturing, presenting the method of the nested test, namely, the analysis of variance(ANOVA), which puts forward scientfic evidence of the existing of nestedness of process parameters. The procedure of nested test is proposed, and the specific operation process is described.2) The first order nested model is introduced, and a common error in some publications about the first order nested control chart is pointed out, where the mean variance and mean value variance are always mixed. In view of that, the corresponding simulations is performed and the difference between these two variances is thus clarified, and obviously the first order nested control chart figured out without distinguishing them is wrong.3) Comparing the existing three methods of calculating standard deviation ?xbar in first order nested control chart, the optimal calculation method is selected after that. By this means the first order nested mean control chart is improved. According to the composition of the conventional control chart, a first order nested control chart is proposed, which is composed of two control charts,and the methods of calculating control limits to all the chart are also presented.4) The two order nested model is introduced and the simulation is carried out for the two order model.5) For the Two order nested problem, the two order nested control chart is developed, and it is composed of four control charts. According to the basic principle of SPC technology, the calculation method of the control limit of the control chart is put forward through theoretical derivation and simulation. The control chart can not only be used for statistical process control, but also has the diagnostic function, which can provide guidance for the process adjustment when it is running out of control.6) In order to better apply the findings of this paper to practice, a new quality control software tool is accordingly developed. The software tool can judge by itself whether the data is nested, and if there is a one order nested or two order nested, and automatically select the corresponding control chart. As a result the learning costs of operators will be reduced. The software tool also has the function of generating SPC report, saving much time for writtng it.