Error Analysis And Correction Method Of Grinding Surface Roughness Measured By White Light Interferometry

With the rapid development of national defense industry,aerospace industry,automobile manufacturing,semiconductor and microelectronics industries,the quantity and quality demand of ultra-precision machined parts have increased dramatically,especially for the surface shape accuracy,surface processing quality and surface roughness,the requirements are increasing day by day.Correspondingly,the surface topography measurement technology in the field of ultra-precision machining is also facing new challenges.On the one hand,the measurement accuracy must be improved to match the accuracy of the surface quality,on the other hand,the measurement speed should also be improved to meet the needs of modern rapid production and online measurement.The white light scanning interferometer is a non-contact three-dimensional surface topography measurement method with high speed and high accuracy,which does not damage the surface to be measured and is widely used in the field of ultra-precision machining.However,due to the limitation of the diffraction limit,the surface roughness measurement results of ultra-precision machined parts are lower than the real value,so it is hard to make a correct evaluation on the surface machining quality,which affects the closed-loop feedback of the machining process,and ultimately leads to a decrease in the yield.Therefore,it is necessary to study the measurement error and correction method of the white light scanning interferometer,and reveal the mechanism of the influence of diffraction error,which is conducive to improve the measurement accuracy and lay the foundation for the realization of white light interference in online measurement applications.In this paper,based on investigating the research status of surface topography measurement method and the measurement error and correction method of white light scanning interferometer,the surface topography measurement results of self-rotation grinding silicon wafer was analyzed,and a power spectral density based high-frequency topography error correction algorithm was proposed,which significantly reduces the surface roughness measurement deviation to less than 10%.The main research contents of this thesis are as follows:(1)Considering the surface gradient error and optical lateral resolution,a white light scanning interferometric simulation measurement model was established based on the MATLAB platform.Based on the analysis of response results of the three basic signal(simple harmonic wave,square wave and triangular wave)and the grinding surface profile signal,the influence mechanism of surface gradient error and diffraction limit error was analyzed.It is concluded that the measurement error of the ultra-precision grinding surface is mainly due to the high-frequency distortion caused by insufficient optical lateral resolution.(2)The measurement results and roughness of the grinding surface morphology measured by the atomic force microscope and white light scanning interferometer were compared and analyzed.And the power spectrum density of the grinding surface profile obtained by experimental measurement and the profile generated by a wafer rotating grinding surface topography prediction model were analyzed.Combined with the mathematical principle of fractal theory,the result shows that the grinding surface is fractal.(3)A power spectral density based high-frequency topography error correction algorithm was proposed to correct the two-dimensional profile measurement results of the grinding surface.The study found that the proposed correction algorithm can effectively compensate for the high-frequency distortion of the topography caused by insufficient optical lateral resolution and reduce the profile roughness deviation from 56.5% to 4.9%.(4)The decomposition and stacking correction algorithm and the two-dimensional power spectral density correction algorithm were respectively used to extend the proposed correction algorithm from two-dimensional geometry to three-dimensional geometry,and the three-dimensional surface topography of the grinding surface was corrected with these two algorithms.By comparing the correction results,it is concluded that the correction result of the decomposition and stacking correction algorithm is better than the two-dimensional power spectral density correction algorithm,and the deviation of the three-dimensional surface roughness can be reduced from 56.0% to 7.4%.