Sensor Based Modeling of Copper Chemical Mechanical Planarization

The variations in slurry chemistry parameters §including pH, concentrations of complexing and §corrosion inhibiting agents, and slurry flow rates §as well as average MRR in Cu-CMP process are modeled §using features from two vibration sensors (one wired §and the other wireless accelerometer). The study is §perhaps one of the first efforts to use wireless §vibration sensors for real-time monitoring of MRR §and the chief slurry chemistry parameters in CMP. We §hope that this work will lead to the advent of §physics-based process-machine interaction (PMI) §models that can be used to delineate the various §process behaviors, which in turn can be used to §select vibration signal features that are more §sensitive to variations in MRR and other quality and §performance variable of the CMP process. The study §has delineated the joint (i.e., statistical §interaction) effects of various slurry components on §MRR in Cu-CMP. In specific, the joint effects of the §complexing agent with other input parameters, such §as pH and flow rate (i.e., two-way interactions) on §MRR have been found to be significant.