Reliability analysis of gate oxide using bi-Poisson model to evaluate the crystal defect effect

In actual LSI processes, it is occasionally found that oxide weakspot densities vary discontinuously or are condensed locally within a wafer. In such a case, neither Poisson nor Gamma distributions are applicable to estimate weakspot densities which are necessary in quantitative reliability evaluation. To solve this problem, we propose a bi-Poisson model. The model is based on the assumption that the wafers can be divided into two regions of normal and higher weakspot density, and that weakspots distribute randomly in each region. The model is applied to the analysis of the weakspots in thermal oxide on an intentionally prepared defect-containing Si wafer. The effect of crystal defects on oxide reliability was evaluated quantitatively using this method