Defect Density from Yield
Infer defect density from measured chip yield and die area using the Poisson yield model. Formula: D0 [cm^-2] = -ln(Y [%]/100) / A [cm²], where Y is the measured die yield and A is the die area. Inputs: 0 < Y < 100 %, A > 0 cm². Output: D0 ≥ 0 cm^-2. Assumes uniformly random defects, Poisson statistics, and that every defect kills a die.
Description
Infer random killer-defect density from measured die yield and die area with the Poisson yield model.
Infer defect density from measured chip yield and die area using the Poisson yield model. Formula: D0 [cm^-2] = -ln(Y [%]/100) / A [cm²], where Y is the measured die yield and A is the die area. Inputs: 0 < Y < 100 %, A > 0 cm². Output: D0 ≥ 0 cm^-2. Assumes uniformly random defects, Poisson statistics, and that every defect kills a die.
When to use Defect Density from Yield
- Infer random killer-defect density from measured die yield and die area with the Poisson yield model.
- Compare fabrication or device scenarios while holding coefficients and unit conventions constant.
- Check a hand calculation before moving to a higher-fidelity process, circuit, or TCAD model.
How the calculation works
Exactly 100% measured yield maps to zero inferred density; very small sample counts still carry large statistical uncertainty not shown by the point estimate.
D0 = −ln(Y)/A Interpreting the result
Exactly 100% measured yield maps to zero inferred density; very small sample counts still carry large statistical uncertainty not shown by the point estimate.
Assumptions and limitations
- The Poisson model assumes spatially independent uniformly distributed defects and that each defect kills a die.
- Use parameters measured for the same material, geometry, temperature, and operating regime whenever the result informs engineering work.