Statistical Quality Control Techniques in Semiconductor Manufacturing

Abstract: Research addresses Statistical Quality Control (SQC) techniques designed specifically for semiconductor manufacturing processes. Packaging, test, and assembly steps normally receive little SQC analysis although proper control is equally important. Steps assigned high defect rates require special attention to offset the effect on product yield. SQC tools directly influence semiconductor production yield, for both internal process yield and wafer-scale yield. The SQC solutions incorporate process capabilities, control charts, and sampling strategies. Recommended charts and control limits for all critical processing steps are provided by data from a major electronic product manufacturer. Special CUSUM and EWMA charts assess subtle process drifts. Process capability indices for non-normal data address the unique nature of defect counts. Bayesian techniques allow prior information in the analysis of any semiconductor step and posterior updating improves decisions on unknown parameters such as process mean and variance. A full integration with yield and defect analysis quantifies the contribution of control on yield improvement. Further development of statistical techniques is supported as overwhelming volumes of process data made real-time monitoring possible.

Keywords: Data; Tax compliance monitoring; Public revenue systems; Decision support; Policy evaluation; Policy design; Administration arrangements; Policy instruments; Tax gaps; Revenue risk.