Abstract :
Yield enhancement engineering usually focuses on three areas of interest: investigation of low yielding lots (or lots with abnormal fail signatures), elimination of in-line defect process excursions and improvement of baseline product yield. These tasks require that engineers digest the data necessary to lead them to what needs to be done to find root cause for a given yield issue. The process of gathering and digesting the data necessary to arrive at the root cause of a yield problem can take a significant period of time (i.e., days or, in some cases, weeks.) There are many reasons why it takes so long to gather the necessary data needed for yield analysis, among them are: (1) data often needs to be extracted through many different software interfaces each of which may require a different language knowledge, (2) the data extract itself may require access to a particular machine and require a custom (i.e., new) database call to be written, (3) the multiplicity of database extracts require a significant portion of time to complete (i.e., a relational database is not always used), (4) once extracts are completed data must be formatted for analysis, (5) data overlay from several sources is often not available and manual methods must be employed. It is very clear to anyone who has faced the aforementioned data “islands” that integration of data sources into one database which is easily and quickly accessible through one user interface will significantly reduce time to root cause for many yield issues
Keywords :
inspection; integrated circuit yield; production engineering computing; production testing; relational databases; baseline product yield; data format; data integration; data overlay; data sources; fail signatures; in-line defect process excursions; yield enhancement; Data analysis; Data engineering; Data mining; Dispersion; Image analysis; Image retrieval; Optical detectors; Relational databases; Testing; User interfaces;
