Research On The Variability Measurement And Performance Evaluation Of Semiconductor Chip Assembly Test Production Line

With the increasingly fierce competition in the semiconductor industry,it is the company's goal to achieve a short production cycle,a high production rate,and a low work-in-process level without increasing production capacity.There are many variability factors in the semiconductor chip packaging test production line.Quantifying these variability factors and establishing their quantitative relationship with production performance are prerequisites for optimal control of production performance.This thesis supported by the National Natural Science Foundation of China's "Study on the variability of semiconductor manufacturing system performance prediction and optimization methods"(Grant No.71671026).Based on the existing theories of plant physics,this thesis started from the variability measurement of the production line,improved the measurement model of process variability and flow variability,and then built a production performance forecast model for multi-product mixed flow conditions based on the variability measurement results.Finally,the performance evaluation of a specific product was achieved.First,the preemptive and non-preemptive variability measurement methods of the factory physics were used to measure the variability of the specific products on a link,and the variability of multiple products on the links was integrated by the weighting the proportions of release.Measurement for mixed-process-variability of multiple product was realized.Based on the support vector regression model,the sample data set was extracted from Arena simulation model.The rule that the variability of multiple links in parallel aggregating to the whole station was described,and the measurement of the process-variability of parallel link structures was developed.Then,to address the problem that the existing flow variability measurement model did not consider the loss of the lot(the basic unit of wafer flow on the production line)caused by quality problems in the semiconductor packaging test production line.the impact of quality loss on the variability transfer process,especially,the impact of the departure time interval was analyzed in this thesis.And an improved flow variability measurement model considering quality loss was proposed,the effectiveness and accuracy of the improved model were verified by simulation data.Finally,a quantitative relationship between the variability measurement results and the three major production performance indicators was established based on the VUT equation and Little's law.Based on this relationship,the performance of the production line was predicted.Especially through the division of logical work station to make prediction of the production performance of a specific product.the “three benchmarks” performance evaluation method was used to qualitatively and quantitatively evaluate the predicted production performance.Based on Arena's experimental verification,the proposed method was proved to be feasible and effective.Based on the above research,a prototype software system for variability measurement and performance evaluation was designed and implemented.Based on the system,the company can execute the "production planning design-performance evaluation-optimization plan-reassessment" process to obtain the optimal production plan beforehand.That can effectively help companies improve production performance.