Modelling And Parameters Optimization Of The Czochralski Silicon Growth Process

Mono-crystalline silicon is the most important fundamental material in semiconductor industry.The requirements are pushed to larger diameter and lower defects by the industry.Under such backgrounds,research is focused from microscopic of defects,to marcoscopic view of control parameters.Detailed researches are done on the melt-solid' phase interface shape formation,the relationship between pulling rate and temperature gradient,and outer shape regulation of silicon ingot.Interface shape formation is studied with p.rocess modelling and process parameters are optimized for production impovement.1.Numerical simulation plays an important role in the researches of Czochralski silicon growth process.Finite element method(FEM)is used.to build the 2-D themal field process model with latent heat taken into account.This also helps th e study of the interface shape formation mechanism during each stage of growth,and how the process parameters affect the interface shape.Modeling of CZ-Si process under magnetic field is done with 2D/3D mixed boundries conditions,wh ich has the effects of heat radiation,heat conduction and Ar gas flow taken into consideration.Simulation helps the study of the temperature fluctuation near the melt-solid interface in large diameter hot zone.Result noted-inadequate of the horizontal magnetic field,and a new quad-pole magnetic flux is proposed.Simulation results of growth process under quad-pole magnetic flux shows it has significantly improved the temperature distributaion near the melt-solid interface and the melt silicon.After optim ization of the magnetic flux density under such structure,it provides a stable environment for the growth process.2.In the CZ-Si process,the efficiency and yield rate is partly depents on the crown and cone stage.For the phenomenon of melt fluctuation touches the crystal which lead to the intruption of mono-crystalline formation,theoretical research and numerical simulation are used to explain this phenomenon.Tuning of crystal rotation speed is proposed to eliminate the strong convection in the melt during crowning process,which ensures the smooth and stable transition to the body growth stage.During the cone growth process,quick diameter shrink and sudden stop of growth process may lead to the reverse extension of dislocation,simulation helps explain the cause of cone growth failure,adjusting the heating power trajectory shows improved result.During the experimental runs,both crown and cone growth process achieved good result which improves the efficiency and yield rate.3.Max pulling rate is highly correlative with vertical temperature gradient formed by the hot zone components.A structure improvement is made to the heat shield in the hot zone components.Different heat shield design's impact on the growth process is studied and compound heat shield structure shows higher growth efficiency.Furthermore,method for max pulling rate estimation is proposed and verified by the experiment.The number of micro defects is compared under different pulling rate.Analysis provides constraints on pulling rate which is for improvements on crystal's microscopic quality.4.Current growth controller shows weak ability on the control of crystal quality.Based on the analysis of diameter fluctuation after crownning growth,the model predictive algorithm is proposed in the growth process.A Generalized Predictive Controller(GPC)is used in the pulling rate feed-forward path which improves the trajectory following accuracy.Experimental runs shows good result.The new controller improved the usable length by surpressing the diameter fluctuation after crown stage,and at the same time ensured the microscopic quality of the crystal.The results correctness is verified by simulation research and engineering experiments.