AUTOMATIC PATTERN CLASSIFYING METHOD BASED ON DEFECT- REFERENCING SYSTEM

摘要

PROBLEM TO BE SOLVED: To identify causes of defects using a standardized qualitative method during in-process inspection of a wafer, by assigning generalized symbols to attributes of a single subset of a plurality of database objects corresponding to feature attributes, ad storing identifiers corresponding to features in a database. SOLUTION: A semiconductor wafer is processed, using a process tool or tools to determine defect attributes (205). Then, a table or tables are created, where defects are described in the form of a concept flowchart (210). Thereafter, each attribute is generalized by expressing it qualitatively, rather then quantitatively. Then, each attribute is correlated to a category, which is a subset of database objects (215). After categorizing all the defect attributes, their identifiers are assigned, which are prepared by arranging in a predetermined sequence, symbols of the categories to which the defect attributes belong (220). The defect identifiers, identifying related cause information, are stored in a storage device (225). Stored process parameters and data are also linked with the defect identifies (235).