Correlation Mechanics for Transferred Analog Block Yield Distributions across Secondary Wafer Foundries
Transferring analog blocks across secondary foundries requires mapping PCM covariance matrices to maintain targeted circuit yield distributions.
Multivariate Covariance Matrix
Meaning
Mathematical representations of variability in multi-parameter systems capture the relationships among several random variables. The construction of a multivariate covariance matrix provides a structured array of numbers showing the covariance between each pair of parameters. This matrix serves as the basis for understanding how process variations affect multiple circuit performances simultaneously.
Statistical Structure
Diagonal elements of the matrix represent the variance of each individual parameter. Off-diagonal elements describe the relationship between variables. When engineers compute a multivariate covariance matrix, they normalize these values.
Yield Calculation
Yield estimation software uses this matrix to generate synthetic parameter sets during Monte Carlo simulations. Accurate representations of the relationships between threshold voltage and gate oxide thickness prevent the generation of unrealistic device states. These simulations define the boundary of the functional yield of the silicon.
Test Optimization
Production test limits are refined by analyzing the covariance data of mature silicon lots. If two parameters exhibit near-perfect correlation, one test can be eliminated from the screening program without increasing the escape rate of defective parts. This optimization reduces the total test time per die on the automated test equipment while maintaining a strict quality standard for delivered parts across all production runs.