Statistical description of Preston's coefficient for loose abrasive processes

Chemical mechanical planarization and lapping are critical processing technologies in a number of precision manufacturing operations. In spite of the critical nature of these processes, the process variability can be quite high. While recent efforts have focussed on reducing process variability, little has been done in to expand existing models of the material removal rate to make it possible to predict process uncertainty and yield. This paper uses a Monte-Carlo simulation to predict the final thickness of a part produced in a standard soft plate planarization process. This model is based on Preston's equation, but is explained in terms that account for the reduction in material removal rate that occurs as the time since plate charging increases. Data are presented that compare experiments for one material, NI-Zn ferrite, with predictions obtained using the model. The simple model is shown to be capable of reducting process uncertainty and may help to bring the process under control without procedural changes or capital expenditures.