composite yield

Composite Yield is a yield model that partitions die yield into systematic (fixed) and random (defect density-driven) components — Y_{composite} = Y_{systematic} \times Y_{random}, allowing separate optimization strategies for each component.

Composite Yield Model

• Systematic Yield: Y_{sys} — yield loss from design-process interactions, edge effects, and pattern-dependent failures that affect the SAME die every time.
• Random Yield: Y_{random} = e^{-D_0 A} (Poisson) or similar — yield loss from random defects (particles, contaminants) distributed across the wafer.
• Negative Binomial: Y_{random} = (1 + D_0 A / \alpha)^{-\alpha} — accounts for defect clustering (alpha = cluster parameter).
• Separation: Separate systematic and random yields by analyzing die failure patterns — systematic failures are spatially correlated.

Why It Matters

• Targeted Improvement: Systematic yield requires design or process changes; random yield requires defectivity reduction — different solutions.
• Mature vs. New: New processes are dominated by systematic yield loss; mature processes by random defects.
• Prediction: Composite models predict yield more accurately than single-component models.

Composite Yield is dividing blame between design and defects — separating systematic from random yield loss for targeted improvement strategies.