The evaluation and modeling of the CMP removal rate for polysilicon

Polysilicon chemical mechanical planarization (CMP) is an area of research, which has largely been ignored until recently. With the development of bipolar complementary metal oxide semiconductor (BiCMOS), polysilicon has become of increasing importance as the preferred material for filling of deep trench. In this paper, study of removal rates of Polysilicon CMP will be discussed. Traditional models for predicting the polishing rate are typically based on Preston's equation. Here we present a new empirical polishing rate (EPR) model, which is statistically evaluated by cross-validation method, which assess the predictive ability of the model. The study also reveals inadequacies of Preston's equation in describing the polysilicon removal rate in the CMP process.