Study On Scratch Removal And Cleaning Method Of CMP Surface Of Monocrystalline Silicon Components

As the basic material in IC manufacturing,monocrystalline silicon component is the core and foundation of semiconductor industry,and its surface quality directly affects the overall performance of IC.Traditional chemical mechanical polishing(CMP)of monocrystalline silicon will produce a large number of defects such as scratches and impurities on the surface and sub surface of components,which seriously affect the surface quality of monocrystalline silicon wafer.In this paper,single crystal silicon components are studied in four parts.The mechanism of scratch formation on the surface of single crystal silicon wafer in CMP is analyzed and experimentally studied.In order to remove the scratch,the process parameters of chemical mechanical polishing of single crystal silicon wafer are optimized.Homogeneous polishing and non Newtonian fluid polishing are used to polish the single crystal silicon wafer to suppress the scratch,The scratch passivation and cleaning process of single crystal silicon wafer in CMP were studied.The specific research work is as follows:The surface scratch morphology,formation conditions and size relationship of single crystal silicon wafer CMP were analyzed.The effect of different kinds of abrasive and particle size on the scratch morphology of monocrystalline silicon wafers was analyzed.The results show that diamond and silicon carbide as polishing powder are mainly brittle scratch,while alumina and ceria as polishing powder are mainly plastic scratch.The scratch length and width increase with the increase of abrasive particle size.Based on the nanoindentation experiment,the critical load of brittle plastic transition is 35 m N.The influence of traditional CMP processing parameters(polishing disc speed,loading disc pressure,slurry concentration and slurry flow rate)on the surface scratch morphology was analyzed by single factor experiment.The mathematical model between the number of surface scratches and the four process parameters was established by response surface methodology.Based on genetic algorithm and particle swarm optimization algorithm,the mathematical model was optimized to determine the best combination of process parameters for surface scratch removal.The effects of different slurry compositions on the number of surface scratches and material removal rate were studied,and the optimal compositions of the special slurry and non Newtonian fluid slurry were determined.The experimental results show that the number and size of scratches on the surface of monocrystalline silicon can be reduced effectively compared with the traditional CMP process.Finally,the influence of different cleaning solutions and process flow on the cleaning effect of single crystal silicon wafer is studied through experiments,and the cleaning scheme of KOH+PVP+LAS new cleaning solution and HCl+H₂O₂+H₂O mixed solution is formulated,which can effectively passivate scratches and remove impurities.The results show that ultrasonic assisted cleaning can significantly improve the scratch passivation and impurity removal rate.