On the use of wafer positional and spatial pattern analysis to identify process marginality and to de-convolute counterintuitive experimental results

Abstract

The use of wafer randomization and positional analysis in manufacturing is ubiquitous and well established. Wafer electrical and yield data can be traced back to specific operations in the manufacturing process with the help of wafer sequencing records. Tight process windows or complex process technologies may however require that the statistical parameter versus sequence signal be combined with other variables including parameter spatial distribution on wafer, batch loading sequence, and an intimate knowledge of the process tool itself. In this paper case studies are presented to demonstrate how the above were combined to optimize the robustness of an especially critical gate oxidation process module. Finally possible enhancements to analysis techniques are discussed.