Shortcomings and gaps in the traditional approach.
Traditional outlier and quality control methodologies often assume that data will conform to ideal, Gaussian distributions. In practice, this is rarely the case. Upstream variability in manufacturing—ranging from process variation to equipment inconsistencies—can significantly impact the behavior of otherwise good die. Identifying subtle yield detractors requires examining a broader set of wafer-level test parameters and their data signatures. These signatures may follow non-Gaussian distributions such as Log-Normal, Uniform, Chi-Square, or Multi-Modal, and must be analyzed in conjunction with spatial considerations like wafer zonal effects.
To read more, click here.
 |
 |
 |
 |
 |
 |
 |  |
 |  |
 |
 |
Leave a Reply