Statistical Analysis and Sensor Based Modeling of Electrochemical Mechanical Polishing (Ecmp) of Cu Blanket Wafer for Semiconductor Applications

Abstract

Copper electrochemical mechanical polishing (Cu ECMP) has attracted many scientists in the field of semiconductor manufacturing industry, as they are investigating whether to replace chemical mechanical polishing (CMP) with Cu ECMP after the introduction of porous low - k dielectric materials because of its increased efficiency and better quality. An approach to locating real time variations in Cu ECMP processes has been identified. The various regions in Cu ECMP, active, passive, trans-passive, and transient regions have been explored through current and voltage signals obtained from a data acquisition setup hooked up with the process. Extensive work has been done in characterizing and optimizing the process, but little work has been done in correlating the key process input variables (KPIV) such as voltage, platen speed and wafer carrier speed with the key process output variables (KPOV) such as the material removal rate (MRR) and surface roughness (Ra) using time-frequency patterns of the current-voltage signals and statistical models. Based on these analytical methods, optimal surface value was realized in the transient/trans-passive region. A mathematical model depicting ECMP was implemented in COMSOL to understand the characteristics of the process, which will be a great breakthrough in the field of semiconductor manufacturing.