A Novel Resource Optimization Approach for Yield Learning

摘要

In this paper, we describe a new integrated framework for yield learning, based on linking traditional inspection sampling, and current Automatic Defect review and (ADC) Classification procedures. The elements of a yield learning cycle, and the drivers, are identified. We then review results concerning integrated inspection-classification/review procedures that reduce yield loss detection; these incorporate new optimized control charts that incorporate killer and non-killer defect types, with classification errors, as well as integrated queuing-hypothesis testing approaches combining resource management and excursion detection. We briefly touch upon tactical approaches for achieving source isolation and prioritizing source isolation and root cause analysis.