Modeling and optimization of wafer-level spatial uniformity with the use of rational subgrouping

The goal of this paper is to present the rational subgrouping methodology for modeling and optimization of the wafer-level spatial uniformity. In particular a single-wafer rapid thermal oxidation of silicon process is used as the experimental platform. Response surface methodology consisting of design of experiments and regression techniques was used to construct the models. Results show that spatial uniformity metric with rational data subgrouping is needed to provide a more complete picture on the thickness uniformity. With different subgroupings of the oxide thickness data, we can offer better strategies for process optimization.