Multi-objective optimization of tungsten CMP slurry for advanced semiconductor manufacturing using a response surface methodology

In this study, a response surface methodology (RSM) coupled with a face center cube design (FCD) was used to optimize the three principal components (i.e., Fe(NO₃)₃, H₂O₂, and SiO₂ abrasives) in polishing slurries for a W barrier chemical mechanical planarization (CMP) process. The experimental ranges of the three components were 10–50 ppm of Fe(NO₃)₃, 0.3–0.9 wt% of H₂O₂, and 1–5 wt% of SiO₂ abrasives. Based on the experimental data from the FCD, the second-order models for the material removal rate (MRR) of the W and Oxide films were fitted; these were determined to be statistically valid and reliable. We have achieved the optimal conditions for the three components where the MRR is maximized and the selectivity between the W and Oxide MRRs is ~ 1. The predicted MRR and selectivity at the optimal conditions were well correlated with the results of a confirmation run, which was conducted by using the W barrier CMP process with W-patterned wafers. In addition, we employed a particular RSM called dual-response optimization in order to investigate the tradeoff between the MRR and selectivity. Based on the tradeoff information, process engineers can conduct the optimization of the three components more flexibly.