As EUV lithography becomes increasingly prevalent in the fabrication of advanced IC chips, the ability to provide defect-free EUV masks through reliable pattern defect inspection tools is crucial for successful high volume EUV wafer fabrication. While the existing mask pattern inspection tool is based on 193nm DUV wavelength and has been used for EUV mask production for many years, the development of actinic pattern mask inspection (APMI) tools has enhanced detection capabilities for printable defects on EUV wafers. However, both DUV-based and APMI tools are costly inspection scenarios. This study aims to evaluate the Lasertec X9ULTRA, a new 193nm EUV mask pattern inspection tool, as a more cost-effective solution for high volume N3 EUV mask making. By optimizing shading calibration conditions and improving tool optics components for illumination and speckle noise reduction, the signal-to-noise ratio (SNR) of N3 mask DOIs were enhanced. As a result, X9ULTRA has the capability to detect all printable defects of interest (DOIs) accurately through die-to-die pattern inspection, meeting the sensitivity requirement of N3 mask productions. The validation of real N3 mask productions confirmed the tool's false defect count performance also. Furthermore, the X9ULTRA tool can output full mask pattern images to TSMC in-house GPU-based image computers and data servers, allowing for the development of a TSMC in-house die-to-database defect inspection algorithm. This study identified the SNR of DOIs, the defect detection capability, and the false count performance of the in-house die-todatabase algorithm using N3 masks.
|