Granulation of silicon in a fluidized bed reactor particle size distribution prediction

PIÑA, JULIANA; SCHBIB, SUSANA; EGE, PAUL; BUCALÁ, VERÓNICA; DE LASA, HUGO

Zacatecas, Méjico

Congreso; Chemical Reactor Engineering X: Innovations in Chemical Reactor Engineering; 2005

Engineering Conferences International

  Chemical vapor deposition (CVD) fluidized bed technology constitutes an attractive alternative to produce granular silicon (solar grade) from silane, replacing the energy- and labor-intensive Siemens process.         A CVD multizone fluidized bed reactor (FBR) model is reported. The FBR model is coupled with the population balance equation (PBE) to evaluate the Particle Size Distributions (PSDs) as the granules grow within the fluidized bed. The solution of the FBR model provides the growth rate that is used by the PBE to calculate the PSDs over the reactor time-on-stream. The method of characteristics (MOC) is selected to solve the PBE. MOC involves the shift of the control volume in the particle diameter coordinate as the operating time proceeds. Since the initial particle diameters grow as the granulation process develops, a moving grid is employed in the calculations.   Many techniques are used to experimentally determine the PSDs, among them the sieve analysis is the most popular one. Often a given set of sieves is used to measure the PSDs at different times-on-stream and therefore the experimental PSDs correspond to a fixed diameters grid. This is in contrast with the calculated PSDs characterized by the moving grid and poses a major problem for granulation modeling which involves comparison between the experimental and calculated data. To represent the calculated PSDs on the fixed grid, three criteria are proposed: numerical particles classification according to the sieves mesh, and two novel interpolation formulae. The accuracy of the criteria is discussed, and recommendations of the proper usage of the proposed formulae for different types of initial PSD are reported.