Fuzzy Logic Based Model to Predict Surface Roughness of Si(100) Wafer Using Preliminary Experimental Data Obtained From Single Pole Magnetic Abrasive Finishing Process

The monocrystalline silicon wafer, Si (100), in the present manuscript has been polished with the single pole magnetic abrasive finishing (SPMAF) technique using the oxidizing agents, namely Sodium Hydroxide (NaOH, 1.5% wt/wt) Marshall’s acid salt (K₂S₂O₈, 1.5% wt/wt), Caro’s acid salt (KHSO_5, 1.5% wt/wt) and hydrogen peroxide (H₂O₂, 0.5% wt/wt) in alumina slurry (Al₂O₃) with a mesh size of 1200. The effect of process variables like polishing speed, working-gap and feed-rate was investigated on the surface roughness of monocrystalline silicon wafer with the help of eight experiments based on factorial design. The experimental data obtained from the experiment was used to develop a fuzzy model available with the MATLAB toolbox for predicting the surface roughness of Si (100). The fuzzy model developed had a close resemblance to the experimental data, with a deviation of 8.70%. The study was further extended to examine the surface morphology of unpolished and polished samples using SEM and AFM images. The SEM image shows that defect-free polishing was achieved at optimum parameters and the peak-to-valley height has been reduced to 18 nm from 0.22 µm.