Determination of the effective distribution coefficient (K) for silicon impurities

For the production of photovoltaic cells, the silicon purity can be intermediate between metallurgical grade silicon (MG-Si, 98%-99.9% pure) and electronic grade silicon (>99.9999% pure). This silicon, with intermediate purity and that still meets solar cell requirements, is called upgraded metallurgical grade silicon (UMG-Si). One method of producing UMG-Si is applying a controlled solidification process, like unidirectional solidification (heat exchange method), zone melting (or zone refining), or Czochralski growth to MG-Si. These processes use the impurities solubility difference in solid and liquid silicon known as effective distribution coefficient (K). For these reasons, to study the solidification process, it is necessary to determine K for silicon impurities, which is the objective of this study. MG-Si (99.85% purity or 1500 ppm of impurities) was processed by 1 pass of zone melting at 1 mm/min using an electron beam furnace with water cooled copper crucible. The effective distribution coefficient (K) for impurities with Ko <= 10(-1) was found to follow the relation K = 0.03 Ko(-0.063). For boron, K = 0.8. Impurities with Ko between 10(-3) and 10(-8) presented similar effective distribution coefficients (K = 0.07 +/- 0.02), meaning that the effective distribution coefficient of a specific impurity depends on the total amount of impurities. The measured impurities profiles in silicon were compared with those obtained by Pfann's equations using the effective distribution coefficients and showed comparative results. (C) 2012 American Institute of Physics. [http://dx.doi.org/10.1063/1.4739759]