Production Systems

ADC systems, integration with inspection tools, and continuous learning

Automatic Defect Classification (ADC)

ADC systems automatically classify defects in real-time as part of the production inspection flow:

Inline ADC: Classification happens on the inspection/review tool itself, immediately after image capture. Low latency but limited compute.
Offline ADC: Images are transferred to a server for classification by more sophisticated models. Higher accuracy but adds delay.
Hybrid: Fast pre-classification inline, with uncertain cases sent to a more powerful offline system.

Production ADC requirements:

Speed: Classify 1,000+ defects per wafer in seconds
Accuracy: >95% agreement with expert human classification
Purity: Critical defect categories (e.g., "killer defect") must have very high precision — false negatives are costly
Adaptability: Models must handle new defect types as processes change

Key Concept: Continuous Learning

Semiconductor processes constantly evolve — new recipes, new materials, new pattern densities. ADC models must be continuously updated with new training data. This requires a pipeline: flag uncertain classifications → expert review → relabel → retrain → redeploy. Active learning prioritizes the most informative samples for human review.

Classification & Root Cause

Understanding Yield

Knowledge Check

1 / 1

What is the key challenge for ADC systems in production?

Maintaining accuracy as processes evolve, requiring continuous model updatesGetting enough compute powerConvincing engineers to use the systemConnecting to the internet