Research On Flattening Of Semiconductor Wafer Based On Multi-pole Arrangement In Magnetorheological Plane Finishing

With the rapid increase in the development of integrated circuits and other fields,more and more stringent requirements have been placed on the processing quality and accuracy of the wafer surface used to manufacture chips,requiring extremely high smoothness,cleanliness and flatness.However,it is very difficult to perform high-quality,high-efficiency polishing planarization of wafers.At present,this technology is still monopolized by a few countries or regions.Traditional polishing methods are easy to cause scratches,sub-surface damage and fragmentation on the wafer surface.Magnetorheological polishing technology has the advantages of low normal pressure and less surface damage,controllable material removal,etc.,which can achieve a high wafer surface,but it is difficult to achieve the ideal surface flatness due to the characteristics of the overall flexible contact.In the magnetorheological polishing experiment,it was found that the multi-pole arrangement of the magnetic field generator has a huge impact on the flatness of the wafer surface.In this thesis,a systematic study on magnetorheological polishing planarization of multi-pole layouts is carried out,and the mapping relationship between magnetic field characteristics and material removal is deeply studied.A prediction model for wafer surface flatness is established,and a magnetic field generator with a multi-magnetic pole arrangement realizes flattening polishing.The main research contents and results of this thesis are as follows:First of all,the mapping relationship between the magnetic field characteristics and the polishing positive force is clarified,and the wafer point movement trajectory is modeled and simulated.The force state of the magnetic particles and abrasive particles is analyzed in the magnetorheological polishing solution.Then the mapping relationship between the magnetic field characteristic quantity and the polishing positive force of the wafer is determined.The polishing positive force model is established,and the correctness of the mathematical model is verified through experiments.In addition,the movement trajectory model is established to obtain the change process of the polishing positive force and the relative speed of the wafer surface polishing point.Subsequently,based on the combination of the Preston equation and the micro-element material removal mechanism,a material removal model facing the wafer surface was established,and a prediction model of wafer surface flatness was established on the basis of this model.The prediction model of wafer surface flatness during the polishing process was verified by magnetorheological polishing experiment.A white light interferometer was used to measure the surface morphology of the polished wafer,and the accuracy of the flatness prediction model was verified by comparing the surface morphology and flatness of the wafer obtained by simulation and experiment.On the basis of this model,the effect of process parameters on surface flatness is studied by simulation,and the analysis results show that with the change of process parameters,the change law of surface flatness of wafers with different initial topologies is quite different.In addition,studying the influence of polishing process parameters on surface flatness provides a better combination of process parameters for the study of determining the arrangement of multiple magnetic poles.Finally,the influence of the multi-magnetic pole layout on the magnetic field distribution was studied in the polished contact surface area,and the influence of the typical multi-magnetic pole layout on the flatness of the wafer surface was analyzed.Considering both polishing efficiency and quality,the comprehensive flatness is defined,and then the advantages and disadvantages of different multi-pole layouts are compared.In order to obtain the best combination of magnetic pole parameters,a method for solving the problem of magnetic pole parameter combination is proposed in the multi-pole arrangement.The BP neural network optimized by the genetic algorithm is used to predict the comprehensive flatness under different magnetic pole parameters,and then the best combination of magnetic pole parameters is obtained,and a magnetic field generator with multi-pole arrangement of polishing planarization is realized.