Machine learning tools for analyzing semiconductor yield analysis and diagnosis data

A non-transitory computer readable medium encoded with a computer program comprising instructions to cause a processor to:

• train a classification machine-learning (ML) model to correlate feature values of yield diagnostic data generated for training wafers of integrated circuit dies, to defects in the training wafers, wherein,
• the yield diagnostic data comprises wafer-level test data, wafer region based defect densities, and functional circuit block based defect densities,
• the wafer region based defect densities are a function of numbers of defects within regions of the training wafers and physical areas of the regions, and
• the functional circuit block based defect densities are a function of numbers of defects of functional circuit blocks of the integrated circuit dies and physical areas of the functional circuit blocks.

A method, comprising:

• receiving yield diagnostic data for wafers of integrated circuit dies;
• and
• using a suite of one or more machine-learning (ML) models to infer information related to defects in the wafers based on feature values of the yield diagnostic data, wherein,
• the yield diagnostic data comprises wafer-level test data, wafer region based defect densities, and functional circuit block based defect densities,
• the wafer region based defect densities are a function of numbers of defects within regions of the wafers and physical areas of the regions, and
• the functional circuit block based defect densities are a function of numbers of defects of functional circuit blocks of the integrated circuit dies and physical areas of the functional circuit blocks.