Burn-in failures and local region yield: an integrated field-reliability model

Neighborhood selection for I/sub DDQ/ outlier screening at wafer sort

“…[8][9][10][11] NNR extends this relationship to continuous-valued measurements. Previous articles defined the NNR Î DDQ (x, y) as the median of at least eight dies closest to the (x, y) location.…”

Development of a Dual Burn-in Policy for Semiconductor Products Based on the Number of Defective Neighborhood Chips

“…Recently, several authors [18][19][20][21] developed yield-reliability models taking the time variable into consideration. Kim et al 19,20 and Kim et al 21 introduced the time variable into the yield-reliability model for describing the breakdown of gate oxide, a specific structure within a semiconductor device.…”

“…However, these studies did not consider such an explanatory variable as d and dealt with only a specific type of failure (i.e., gate oxide breakdown). On the other hand, Barnett and Singh 18 developed a general yieldreliability model that includes d as a parameter and considered the time variable without limiting the model to a specific type of failure. An advantage of their theoretical model is that it consists of parameters (e.g., average number of killer defects, degree of defect clustering, ratio of average number of latent defects to that of killer defects, etc.)…”

“…4, the actual burn-in yield of the product considered in this paper shows a piecewise linear pattern rather than a smooth one as predicted by the theoretical model in Barnett and Singh. 18 Therefore, in the present study, an empirical yield-reliability model based on actual data is developed to overcome the above-mentioned problems with the existing studies.…”