研究目的
To propose a threshold-Poisson model for describing particles with excess zero counts in semiconductor manufacturing and develop a method for parameter estimation, comparing its performance with popular defect models.
研究成果
The threshold-Poisson model effectively describes particle count data with excess zeros in semiconductor manufacturing, outperforming other popular models in terms of determination coefficient, AIC, and BIC. The proposed threshold-c control chart and the method for determining the minimum sample size provide practical tools for monitoring particle counts in clean rooms.
研究不足
The study focuses on particle count data with excess zeros in semiconductor manufacturing, which may limit its applicability to other types of data or industries. The performance of the threshold-Poisson model is compared with a select group of models, and there may be other models not considered in this study.
1：Experimental Design and Method Selection
The study proposes a threshold-Poisson model for particle count data with excess zeros, comparing it with Poisson, ZIP, GZIP, Neyman, and Gamma-Poisson models using 15 measured samples.
2：Sample Selection and Data Sources
Particle count data were collected from 15 Lasair II-100 particle counters placed in different positions of a semiconductor clean room.
3：List of Experimental Equipment and Materials
Lasair II-100 particle counters were used for measuring particle counts in the air.
4：Experimental Procedures and Operational Workflow
Particle counts were measured and classified into discrete sizes using a particle counter. The threshold-Poisson model was applied to the data, and its performance was compared with other models.
5：Data Analysis Methods
The study used determination coefficient (R2), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC) for model comparison. Control limits for the threshold-c control chart were derived, and simulations were conducted to determine the minimum sample size.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
