Engineering-depth documentation for process engineers and quality teams running evaluations. Contact us for access.
Detailed treatment of per-recipe adaptive threshold calibration. Covers the calibration algorithm, convergence behavior (lots 1–50, 51–200, 200+), drift correction mechanisms, and benchmark comparison against static threshold systems. Includes false-positive data from 7nm logic, 5nm logic, and 28nm DRAM process nodes. Provides validation protocol for process engineers transitioning from human-assisted inspection.
Statistical analysis of defect escape rates across three shifts at a 300mm logic facility over a 6-month period. Quantifies first-to-third-shift detection rate degradation by defect class (particles, scratches, edge chips). Establishes methodology for demonstrating machine inspection equivalence to first-shift baseline in process qualification documentation.
Technical overview of the convolutional neural network architecture used in Lenspathio's base detection model. Describes training data composition, defect library curation, feature extraction layer design, and per-class precision/recall performance. Includes resolution analysis comparing 7nm versus 28nm input requirements.
All three white papers are available to evaluation teams. Contact us with your role and facility type to receive access.