Review sample shaping through the simultaneous use of multiple classification technologies in IMPACT ADC [IC yield analysis]

Among the various options available for yield improvement tools is automatic defect classification (ADC). ADC reduces the amount of manual inspection required as well as increasing the amount of classified data available. It can serve to entirely remove operator-based manual classification. The classified data can then be used to identify the cause of the yield-limiting problem in the line. To allow this, ADC can sample a limited yet representative number of defects from the total count, but does so under less constraints than those used by an operator who will, depending on experience and knowledge, adjust their chosen sample for each case. As such, the sampling methods employed can have critical effect on the overall results. This paper looks at how the sampling criteria should be established in order to provide the best representation of existing problems in the inspected wafers, and whether this is actually possible using ADC.