Output Improvement in High Volume Memory Fabs by Reducing Recipe Qualifications

Abstract

This paper presents a model to optimize the number of recipes qualified on each tool or chamber in a fleet by evaluating the trade-off between the cost in tool time required to run nonproduct wafers (NPW’s) for qualifying additional recipes versus the benefit of having as many recipes as possible qualified to maximize flexibility. The model specifically addresses an issue observed in high volume fabs such as large memory fabs, where most of the tools are qualified to run several recipes. In these situations, in which the number of tools is much greater than the number of recipes and substantial resources (such as process and metrology tool time, non-product wafers, and labor) are required to qualify each recipe, the cost in tool time for qualifying many recipes on each tool can outweigh the benefits in flexibility. Based on typical product mixes and qualification requirements in such a fab, we demonstrate that using this model could reduce the tool time required to run qualification wafers and increase the time available to run production wafers by 1% or more with minimal or no impact on cycle time.