Silane Pyrolysis CVD Reaction Model And Its Application In The Simulation Of Fluidized Bed Reactor

Utilization of solar energy is an effective solution to solve the problems of resource shortage and environmental pollution.High purity poly-silicon is an important material of solar energy conversion.With low energy consumption,low cost and low pollution,silane fluidized bed CVD reactor is one of the key technologies of manufacturing poly-silicon.Due to the complexities of silane CVD pyrolysis,and multiphase flow in the system,the design and scale-up of the fluidized bed CVD reactor are facing great difficulties.In this paper,the computational fluid dynamic(CFD)method is used in cooperation with the population balance model(PBM)and reaction kinetic model,to predict the flow pattern and the particle size growth rate.Firstly,silane pyrolysis reaction models are analyzed and compared.Based on a critical review of reaction kinetics of silane decomposition,Ho’s gas and surface reaction mechanisms are used for simulating a horizontal single-substrate reactor with the 2-D boundary-layer reaction model of CHEMKIN software.The predicted growth rates are comparable with experimental data,indicating the above mechanisms can be further applied in CFD simulation of silane fluidized bed CVD reactors.Secondly,the couling process of flow and reaction in a small fluidized bed CVD reactor is simulated.With the CFD-PBM coupling model,the modified Ho’s reaction model is used to simulated Hsu’s fluidized bed experiment of poly-silicon.In the fluidized bed reactor,the scavenging effect of cluster precursor in Ho’s model is very significant.When silane mole concentrations are 0.2 and 0.5,the calculated particle growth rates are slightly higher than the experimental results.When silane mole concentrations are 0.57 and 0.8,with high concentrations of cluster precursor transferring stable powder,the scavenging effect of cluster precursor is weaken evidently.The pre-exponential factor of H2SiSiH2 deposition reaction rate equation is decreased 10 times,and the simulated particle growth rates fit well with the experimental data.