Defect spatial signature analysis for semiconductor process

The equipment in manufacturing process of semiconductor tends to cause defects in specific distribution over substrates when it fails, so the defect spatial signatures are efficient information for diagnosis of its operations. However, it is difficult to automatically detect each signature from inspection results because it depends on each failure and the distribution of defects becomes complex if several sets of equipment fail simultaneously. Therefore, we developed a new method using ICA (Independent Component Analysis) and SOM (Self-Organizing Maps) to detect the defect spatial signatures, and we introduce effectiveness of the method.