All the reported thermodynamics analysis implied that ammonia is a promising reagent for removing boron from silicon, but efforts to employ it in silicon refining have failed in practice. As such, there are few reports detailing this process. In this study, this process was analyzed experimentally. Various concentrations of ammonia were introduced to remove boron from silicon in the temperature range of 1200–1550 ∘C with a total pressure of 1 atm. Boron containing nitrides precipitates were detected in the furnace tube. The effect of ammonia content in the feeding gas was explored. It implied that higher ammonia partial pressures promote the boron removal. The experimental results have suggested that ammonia could remove boron from silicon in the form of volatiles, such as BHx (x = 1, 2, 3) and B3 H 6 N 3, in practice. The reaction-rate constant was limited to 10−6– 10−7 m/s in pure ammonia at 1450 ∘C. Moreover, a higher ammonia flow rate resulted in lower boron removal ratio. It was indicated that the rate determining steps of boron removal and silicon loss in this process were the chemical reaction at the surface of the melt and the transport of ammonia from gas phase to the surface, respectively. The relationship of the boron-removal rate with temperature followed a “V”-shaped curve, which implied the limit of thermodynamic factors at high temperature and the limit of kinetic factors at temperatures lower than 1300 ∘C. Based on the analysis results, temperature-programed reaction was designed to promote the boron-removal efficiency doubled. Cu was used to decrease the liquidus temperature of Si based alloy in the process. As a result, more than 80% boron in Si-Cu alloy could be removed.
Read full abstract