SEMICONDUCTOR DEFECT REVIEW: FROM INSPECTION MAP TO ROOT-CAUSE EVIDENCE
Editable planning/verification record — Yenra
Updated: October 2, 2026
Article: https://yenra.com/defect-detection/

Purpose: Connect an inspection population to a documented review sample and distinguish observed defect classes from supported root-cause conclusions.
Intended reader: Semiconductor process and yield engineers organizing defect investigations.

HOW TO USE
Save a working copy. Enter the actual specification, units, assumptions and evidence references in each field. Preserve original observations and record changes. Have the responsible reviewer resolve missing or conflicting requirements before release. Examples below are fictional teaching inputs, not process settings or acceptance standards.

RECORD
1. Lot, wafer, process step, recipe and inspection-area identity:
   ____________________________________________________________

2. Candidate population, coordinates and duplicate-handling rule:
   ____________________________________________________________

3. Review selection method, strata and number selected per group:
   ____________________________________________________________

4. Relocation success, unresolved sites and image-quality limitations:
   ____________________________________________________________

5. Class definitions, reference images and reviewer agreement:
   ____________________________________________________________

6. Composition or cross-section evidence and destructive-analysis authorization:
   ____________________________________________________________

7. Process-history comparison, confounders and root-cause hypothesis:
   ____________________________________________________________

8. Controlled confirmation, independent repeat, disposition and closure owner:
   ____________________________________________________________


WORKED EXAMPLE AND LIMITS
Fictional investigation: inspection flags 1,000 candidates. A targeted review of 100 selects 70 edge-cluster sites and 30 other sites. Review finds 60 particles, 25 pattern defects and 15 unresolved sites. The particle share is 60/100 = 60% of this selected sample. Because sampling favored the edge cluster, it is not an unbiased estimate for all 1,000 candidates or a wafer-yield measurement. Keep the unresolved sites and selection method visible.

REVIEW
Prepared by / date: ____________________
Reviewed by / date: ____________________
Open issues, responsible owner, due date: ____________________
Approved next action and evidence: ____________________

SOURCE CONTEXT
These primary references support the linked article. Use exact applicable product, contract, and document versions for operational work.
Applied Materials' defect-control overview
https://www.appliedmaterials.com/us/en/semiconductor/products/analyze/defect-control.html
SEMVision review documentation
https://www.appliedmaterials.com/in/en/product-library/g7-semvision-defect-analysis.html

Return to the guide: https://yenra.com/defect-detection/