Marvell: Accelerating silicon yield improvement through mass-volume scan diagnosis

Advanced semiconductor nodes such as 5nm and 3nm often show low initial yield, increasing cost pressure and making early identification of systematic manufacturing issues more critical. Effective yield improvement depends on fast, accurate detection of structural defect patterns and timely feedback to the foundry.

Tessent YieldInsight provides a strong framework for diagnosing these systematic yield limiters. By using mass‑volume ATPG scan test data, it enables early recognition of defect trends during initial silicon screening, shortening the path to high‑volume production and improving yield and profit margin.

The flow is driven by two key steps: layout‑aware scan diagnosis and YieldInsight signature analysis. Each step incorporates advanced techniques such as Root Cause Deconvolution (RCD), which uses physical layout information to generate realistic and actionable defect insights. Together, these capabilities accelerate yield learning & support continuous manufacturing improvement.