Material removal model of ultrasonic elliptical vibration-assisted chemical mechanical polishing for hard and brittle materials

The chemical mechanical polishing (CMP) is widely used to polish hard and brittle materials. However, it is difficult for conventional CMP to achieve high material removal rate (MRR) and high surface quality while polishing of hard and brittle materials such as monocrystalline silicon. Therefore, ultrasonic elliptical vibration (UEV) is employed to aid conventional CMP in our research, which combines the functions of conventional CMP and ultrasonic machining. In the ultrasonic elliptical vibration-aided chemical mechanical polishing (UEV-CMP) experimental setup developed by us, the workpiece attached on the ultrasonic vibrator can vibrate simultaneously in both horizontal and vertical directions during CMP. It is found that the ultrasonic elliptical vibration can effectively increase the MRR while maintaining surface quality in conventional CMP. The possible mechanism in UEV-CMP is firstly analyzed chemically to establish a reasonable material removal rate model. The effects of the ultrasonic elliptical vibration on the interaction among the abrasive particles, polishing pad, and workpiece are investigated to explain why the MRR of UEV-CMP is higher than that of the conventional CMP. A mathematic model, which includes polishing variables such as morphology and material properties of polishing pad; abrasive size; and material properties of particles, frequency, and amplitude of the ultrasonic vibration as well as polishing process parameters, is set up to interpret the increase in MRR for UEV-CMP. The results from the MRR model show that the ultrasonic elliptical vibration can improve material removal by increasing both the chemical reaction efficiency of polishing solution and mechanical impact efficiency of the abrasive particles on the workpiece surface and also by increasing vibration amplitude in vertical direction as the horizontal vibration contributes less towards the increment of MRR. Experiments are conducted for model verification, which should be that the experimental results agree well with model predictions.