Theoretical analysis and modeling of the obtaining of polycrystalline silicon in a fluidized-bed reactor

Abstract

A theoretical analysis has been performed and basic problems of the technology of obtaining polycrystalline silicon by pyrolytic decomposition of monosilane in a fluidized-bed reactor have been considered. A one-dimensional, two-phase model of this process and basic results of mathematical modeling have been presented.The problem of development of new low-cost technologies of obtaining polycrystalline silicon for photoelectric transducers is closely related to the prospects for development of solar power engineering. The main factor determining these prospects is the level of scientific and technological advances in the field of silicon technology.It has historically evolved that silicon raw material for production of solar cells is mainly the waste of the semiconductor industry. Clearly, the possibilities of obtaining such an expensive waste are limited and do not meet the demand for photoelectric transducers.In the known effort to obtain inexpensive polycrystalline silicon for photoelectric transducers, one either failed to attain the required output, or the purity of the silicon was insufficient for the high efficiency of solar cells, or the price of the silicon obtained was unacceptable for production of photoelectric transducers.The traditional technology of production of polycrystalline silicon [1] suggests that one initially obtain metallurgical silicon by direct carbon reduction of high-purity silica in arc furnaces at 1800 o C. To obtain higher-purity silicon it is chlorinated to SiCl 3 or SiHCl 3 , which are subjected to deep purification by rectification, sorption, or special heat treatment and then are hydrogen-reduced at 1100-1200 o C. One deposits the silicon reduced on silicon bars of length up to 150 cm and diameter 5-10 mm, increasing it to 150 mm (Siemens process) (Fig. 1). The production is rather energy-consuming: the total expenditure of energy in obtaining a marketable product attains 250 kW⋅h/kg.Although more than 80% of the worldwide production of semiconductor silicon is based on this method, it has a number of substantial drawbacks: the energy consumption and discontinuity of the process, the low yield of silicon (the maximum extraction of silicon possible in theory is about 60%; in actual practice it amounts to 15-30%), the release of hydrogen chloride as a by-product, causing corrosion of the equipment, and the need for hydrogen dilution of the reaction mixture to prevent homogeneous nucleation. Furthermore, one is not necessarily able to achieve a high degree of cleaning of the impurity hydrides.The use of monosilane as a working gas in a modified Siemens process enables one to reduce the temperature of the process to 800-850 o C, to increase the silicon yield to 94-98%, and to ensure additional cleaning of the impurity hydrides. At the same time, serious drawbacks persist; among them are the long duration of the process and thus the high energy consumption and hydrogen dilution of the monosilane fed to the reactor.Thus, the current processes of obtaining polycrystalline silicon are c...

Manufacture of Granular Polysilicon from Trichlorosilane in a Fluidized‐Bed Reactor

“…The purity level of the granular materials produced is slightly lower than that obtained with the Siemens process, but it meets the requirement of the photovoltaic market [13]. In recent years, many research groups have studied silane pyrolysis in a fluidized-bed reactor by experiments and simulations [14][15][16][17]. However, the homogeneous reaction creates fines during the silane decomposition which is highly problematic in a fluidized-bed reactor [18,19].…”

Microstructure and grain growth of polycrystalline silicon grown in fluidized bed reactors

“…The most common source of polycrystalline silicon (polysilicon) has traditionally been via chemical vapor deposition (CVD) in Siemens-style reactors [1], in which silane or trichlorosilane is decomposed on the surface of a seed rod in which the surface temperature is controlled by internal resistance heating of the rod. This technique requires expensive capital equipment, may have high operating costs, often produces low silicon yield (up to 60% from the gas in theory, but only 15-30% in practice), and generates the release of highly corrosive hydrogen chloride, further adding to capital expense [1].…”

“…The process has been described in detail elsewhere e.g., [2,3] and begins with silicon seed particles ranging from 100 to 2000 mm in diameter [1,4] that are fluidized with a carrier gas such as hydrogen. An important and significant aspect of the FBR is the lower temperature range that is used to produce silicon.…”