Localized voids may originate at the pre-bond surface, during bonding, or during annealing. A final defect map can confuse correlation with causation; pre- and post-bond data and controls in the same coordinates are needed.
Confirm void-map resolution, false positives, and sampling limits first.
Register particles, surface profiles, wafer warpage, and void locations.
Compare post-bond and post-anneal data to establish defect timing.
Key point: Register coordinates before assuming a single cause
Register post-bond void maps to pre-bond particles, Cu surface profiles, and wafer topography. If defects cluster at fixed coordinates or equipment regions, check stage, cleaning, and bonding uniformity. Group defects appearing only after annealing separately.
BLNKK editorial notes
Key point: Separate detection capability from root-cause capability
Detecting 1–2 µm defects does not establish their cause. Confirm sensitivity, false-positive rate, and sampling, then validate representative defects by cross-section or destructive analysis to connect inline signals to the process.
Experts have not yet been invited. These are BLNKK editorial notes. Contributions require submission and review; they are not published automatically.
