In semiconductor manufacturing, particle removal is critical for high yield. This study investigates cleaning efficiency on challenging surfaces like deep trenches, wafer bevel, and bonded dies, using various methods. Intentional contamination was applied using silica, silicon nitride, and ceria particles. Cleaning methods tested include high-velocity biphasic sprays, cryogenic processes, and polymer peel techniques. Results show that while polymer peeling offers excellent particle removal efficiency (PRE) on blanket surfaces and wafer bevels, it struggles with high aspect ratio features due to polymer film dewetting. The study provides a new methodology for generating cleaning tool fingerprints and highlights the importance of key contamination parameters, such as particle nature, media characteristics, and queue time. The findings underscore the need for evolving cleaning processes to address the complexities of modern semiconductor surfaces.
Read full abstract