(442e) Evaluation of Bubble Size Distribution and the Effects on Silicon Particle Growth in a Silane Fluidized CVD Reactor

Photovoltaics (PV) power generation has experienced rapid growth in the past decade， yet the market share of PV power is bottlenecked by the high costs of solar grade silicon manufacture that mainly relies on the Siemens process.  Fluidized bed silicon-CVD (FBSC) is considered an important alternative technique due to numerous advantages of low energy costs, good heat and mass transfer characteristics and continuous production, however, the technology is still immature.  In a FBSC reactor, the bed is filled with silicon seed particles fluidized by hydrogen and/or reactant silane gases. The particles are heated above the silane decomposition temperature, and the silicon is deposited on the particle surface. A problem that hinders a continuous operation of the FBSC reactor is the clogging of reactor internals because of undesired fines dust production mainly due to homogeneous decomposition^[1]. Here, the homogeneous decomposition is decomposition of silane in gas phases, which is different from heterogeneous decomposition at a seed particle surface.  Therefore, a critical issue is how to control the thermal and hydrodynamics so that the crystal deposition process is well defined and monitored.  A thorough understanding of the hydrodynamics is essential for design and optimization of the FBSC process.

Experimental study had suggested that the homogeneous decompositions are directly linked to the gas bubble which evolves dynamically in the fluidized bed^[2][3][4]. A number of mechanisms have been proposed for the role of bubble dynamics in fines formation including gas temperatures and the distributions related to silane decomposition rates. The particular features of bubble dynamics, such as the bubble sizes and the evolution of the distributions are thought to be important for predicting and control of fines productions.  Computational fluid dynamics (CFD) provides a viable alternative technique for the description of the multispecies reacting fluid dynamics, the gas, and/or surface transport and the reactions. 

In this work, we present a TFM-KTGF model for the prediction of gas solid hydrodynamics and the crystal growth of particles in FBSC. A 0.5 m ID cylindrical silicon fluidized bed was simulated with computational grids of 200 times particle size, and the volume fraction of the gas phases was obtained. A Matlab image processing toolbox was used for the post-processing of the simulation data and the bubble size distributions were obtained. The classical Gidaspow drag force model and a literature reported sub grid model (SGS) were compared^[5]. The simulated bed expansion height and the average bubble size were validated against the well-accepted empirical formulation. Furthermore, the chemical vapor deposition process in the reactor was simulated. A PBM model was used to predict the growth of silicon particles.  Some well documented data sets were benchmarked.

The results showed that the simulation results of the bed expansion height and the average bubble size were in well-agreement with the Mori & Wen’s empirical formulation^[6], and the deposition rate was in consistence with Hsu’s experimental data. The deposition rates of I.D. 0.5m fluidized bed under different operating conditions were simulated. The bubble size distribution and the evolutions were predicted and the effects on the deposition rates are analyzed computationally.

[1] Filtvedt W O, Javidi M, Holt A, et al. Development of fluidized bed reactors for silicon production[J]. Solar Energy Materials and Solar Cells,2010,94(12):1980-1995.

[2] Praturi A K, Lutwack R, Hsu G. Chemical VaporDeposition of Silicon from Silane Pyrolysis[J]. 1977,77(1):38.

[3] Hsu G, Hogle R, Rohatgi N, et al. Fines in fluidized bed silane pyrolysis[J]. Journal of The Electrochemical Society, 1984, 131(3):660-663.

[4] Lai S, Dudukovic M P, Ramachandran P A. Chemical vapor deposition and homogeneous nucleation in fluidized bed reactors: silicon from silane[J]. Chemical Engineering Science, 1986, 41(4): 633-641.

[5] Wang J, Van der Hoef M A, Kuipers J A M. Coarse grid simulation of bed expansion characteristics of industrial-scale gas–solid bubbling fluidized beds[J]. Chemical Engineering Science, 2010, 65(6): 2125-2131.

[6] Mori S, Wen C Y. Estimation of bubble diameter in gaseous fluidized beds[J]. AIChE Journal, 1975, 21(1): 109-115.