AI Application in Yield and Failure Analysis to Reduce Overall Time-to-Defect and Failure Root-Cause Isolation

Abstract

This paper presents conceptual application of AI in Failure Analysis to connect to various databases in semiconductor manufacturing and generating interactive data visualization to isolate root cause of failure faster vs traditional methods. Generally available low-cost software application like Microsoft Power BI (Business Intelligence) is utilized to visualize big data to isolate failure modes at wafer, die, and package level. This historic data visualization knowledge is further used by failure analyst to process failure mode isolation much faster based on failed package unit history. Semiconductor manufacturing companies have various big data such as wafer fab processing, die level test, or wafer sort and packaged die testing including customer return. MS Power BI application has ability to connect to these separate big databases and create unified data visualization to isolate failure modes through faster inter-connectivity and "connecting the dots" to provide bigger picture or drill down to finer unit level detail. This level of visualization utilizes already available info/data to help reduce overall time-to-defect. With this failure background, engineers can plan fault isolation and analysis and reduce overall time to find root-cause of failure.