die-to-die variation

**Die-to-die variation** is the **parameter spread observed across different dies on the same wafer due to spatial process non-uniformity and module-level gradients** - it drives performance binning, guardbands, and per-lot yield outcomes. **What Is Die-to-Die Variation?** - **Definition**: Across-die statistical variation for metrics such as Vth, Idsat, leakage, and speed. - **Scale**: Macroscopic, spanning die locations across wafer radius and angle. - **Primary Drivers**: Film thickness gradients, CD shifts, implant non-uniformity, and thermal variation. - **Measurement Basis**: Wafer sort parametrics, scribe-line structures, and monitor arrays. **Why Die-to-Die Variation Matters** - **Binning Economics**: Larger spread increases low-bin population and revenue loss. - **Yield Risk**: Tail dies can violate limits even when average process is on target. - **Design Margins**: Timing and leakage guardbands must account for across-die spread. - **Process Control**: D2D metrics are core KPIs for fab uniformity improvement. - **Customer Consistency**: Lower variation improves product predictability lot-to-lot. **How It Is Used in Practice** - **Spatial Decomposition**: Separate radial, azimuthal, and random D2D components. - **Binning Simulation**: Predict distribution of speed-power bins from measured spread. - **Control Actions**: Tune module uniformity and monitor long-term drift by tool and lot. Die-to-die variation is **the macro-uniformity metric that directly connects wafer process control to product performance distribution** - reducing D2D spread is one of the highest-impact yield and revenue levers.