Integration of automated defect classification into integrated circuit manufacturing

Automated defect classification (ADC) has been identified by SEMATECH and its member companies as one of the important enabling technologies required for the next generation advanced wafer fabs. ADC can provide significant improvements in resource allocation and yield enhancement. In the evaluation of a new tool for ADC, several important aspects regarding integration of this technology into manufacturing were investigated. In that this tool represents technology that will, in effect, make decisions regarding defectivity, the use of fuzzy logic was required. Through fuzzy logic, a supervised self-learning approach was implemented to train the ADC tool. Additionally, defect images were "quantified" in terms of fuzzy truth values to facilitate process analysis and new defect introduction into the process. This paper will discuss the increased need for ADC based upon the current defect reduction schemes that are strongly desired for improved company/product competitiveness. This paper also seeks to address some of the challenges in implementing ADC in the wafer production fab.