Optimizing defect inspection strategy through the use of design aware database control layers

摘要

Resolution limitations in the mask making process can cause differences between the features that appear in a database and those printed to a reticle. These differences may result from intentional or unintentional features in the database exceeding the resolution limit of the mask making process such as small gaps or lines in the data, line end shortening on small sub-resolution assist features etc creating challenges to both mask writing and mask inspection. Areas with high variance from design to mask, often referred to as high MEEF areas (mask error enhancement factor), become highly problematic and can directly impact mask and device yield, mask manufacturing cycle time and ultimately mask costs. Specific to mask inspection it may be desirable to inspect certain non-critical or non-relevant features at reduced sensitivity so as not to detect real, but less significant process defects. In contrast there may also be times where increased sensitivity is required for critical mask features or areas. Until recently, this process was extremely manual, creating added time and cost to the mask inspection cycle. Shifting to more intelligent and automated inspection flows is the key focus of this paper. A novel approach to importing design data directly into the mask inspection to include both MDP generated MRC errors files and LRC generated MEEF files. The results of recently developed inspection and review capability based upon controlling defect inspection using design aware data base control layers on a pixel basis are discussed. Typical mask shop applications and implementations will be shown.