Modeling Of Surface Roughness Of Silicon Wafer Rotating Grinding And Experiment Research

Single crystal silicon wafer has been widely used in the semiconductor industry, as the basic substrate material for the integrated circuits. Ultra-precision grinding technology is the core technology of a semiconductor chip substrate planarization processing and back thinned processing. The surface roughness is an important quality evaluation indicator of wafer processing. In order to get lower surface roughness, a large number of experiments to explore the complex grinding parameters were needed to carry out. However, the ultra-precision grinding technology studied in this thesis is quite different from the traditional grinding technology in terms of the grinding motion. The mechanical property of single crystal silicon material is also different from metallic materials. There are still a lot of technical problems to be overcome, since the grinding machine for ultra-precision grinding development is in the prototype stage in our country. Grinding experiments cannot be fully carried out. For these reasons, the establishment of the wafer surface roughness model R_a based on self-rotating grinding principle is significant.Firstly, two different material removal mechanisms were explained including brittle fracture removal mechanism and plastic removal mechanism according to the material properties of silicon. It was shown in the self-rotating grinding experiment that the material removed by the600#grinding wheel was mainly in brittle fracture model and the material removed by2000#grinding wheel was mainly in ductile model.Secondly, based on the principle of rotating grinding movement, the kinematics equation relative to the single diamond grits was built up. The average cutting depth of single diamond grit was established according to the principle of volume invariably. Based on it, the normal force model of the single diamond grits was established. The surface roughness model applied to the rotating grinding was built up, which respectively under the single crystal silicon brittleness removal way and the ductile removal way. Thirdly, the influence rules of the grinding wheel speed, the silicon wafer speed and the spindle to feed speed on the silicon wafer surface roughness were investigated through some related experiments in this paper, respectively. The theoretical model has been validated by the experimental data. The relationship between R_aand the diameter of the diamond abrasive or time of No-Feeding was analyzed.Finally, based on BP artificial neural network, which used the grinding wheel speed, the silicon wafer speed and the spindle feed speed as input value, the prediction model of the output surface roughness Ra was successfully established. This model realized the reliable prediction of surface roughness Ra.