Leveraging tritone AttPSM masks for improved reticle inspections: ADC system design and implementation

This paper presents an advanced Automated Defect Classification (ADC) system designed for Die-toDatabase inspection of Tritone Attenuated Phase Shift Masks (AttPSM), which are critical for highresolution lithography in advanced semiconductor nodes. Tritone masks incorporate a chrome (Cr) layer over a halftone MoSi surface, reducing diffraction and extending phase-shift benefits to line patterns. Conventional ADC systems, optimized for binary masks, often misclassify tritone structures due to inadequate tone recognition and simplified lithography modeling. To address this, the proposed system introduces: • Accurate identification of three-tone GDS layers, including polarity-aware handling of the chrome layer. • A two-level set mask representation for improved lithography simulation fidelity. • Enhanced image-to-GDS registration and refined topology markup for precise defect localization. • Updated classification rules tailored to tritone-specific defect types. Validation using a production dataset demonstrates a significant improvement in defect filtering accuracy— 11% improved—with zero under-classification. These results highlight the system's potential to improve inspection reliability, yield, and cost efficiency in semiconductor manufacturing. The paper details the implementation of this approach and its integration into ADC workflows. By leveraging advanced tritone lithography modeling and refined classification techniques, the proposed system delivers a robust solution for accurate defect analysis in Tritone attPSM reticle inspections.