AN IMAGE PROCESSING APPROACH FOR WAFER SCRATCH IDENTIFICATOIN

Abstract

During an electrical testing stage, each die on a wafer must be tested in order to determine whether its functionality is the same as designed. If there is a scratch on the wafer, it is possible that some bad dies in the scratch area can not be detected by the tester. To avoid the bad dies being proceeded to final assembly, a testing factory has to assign some workers to check the wafers and mark the bad dies in the scratch or closer to scratch area by hand. In this paper, we propose an automatic wafer-scale scratch identiffier using a median filter and clustering approach to detect the scratch and mark all the bad dies. The results verify that our algorithm indeed detects the scratches, and it can be used in chip probing stage to automate the testing of wafers.