(589c) Carbon in Liquid Silicon: Diffusion, Solubility, and Silicon-Carbide Nucleation

Micron-scale silicon carbide (SiC) precipitates are known to detrimentally impact multicrystalline silicon (mc-Si) ingots grown from the melt, which are used to create wafer substrates for photovoltaic device manufacture. First, the hardness and size of the SiC precipitates can damage wire-saws used for cutting mc-Si block ingots into wafers. [1] Moreover, SiC precipitates act as electrical recombination sites, lowering the efficiency of the cell [2, 3, 4]. Although the size and structure of SiC precipitates suggest that they form in molten silicon during the solidification process [5], the mechanism and conditions under which they form are not well understood at all.

Here, we employ an array of atomistic simulation tools to study the properties of carbon impurity atoms and the nucleation of SiC precipitates in liquid silicon. Given the relative paucity of atomistic simulation studies of the liquid silicon-carbon system, we consider multiple empirical potentials including different parameterizations of the popular Tersoff [6] potential: tersoff-1989 [7], tersoff-1994 [8], and Erhart-Albe [9] (EA), as well as the modified embedded atom method (MEAM) [10] potential. Using these potentials we compute several key thermodynamic and kinetic properties including carbon diffusivities, solubility limits, liquid-solid segregation coefficients, and homogeneous nucleation barriers for SiC crystallite formation.

We find that most of the potentials give a consistent description of carbon diffusion and melt-solid segregation, and provide predictions that are in line with experimental estimates [11]. However, larger deviations are observed for carbon solubility in liquid silicon; this thermodynamic property is substantially overestimated by all the potentials considered. Some possible reasons for this finding are discussed. Nonetheless, we use the derived solubilities to compute SiC nucleation barriers at known values of carbon supersaturation and undercooling. The results are discussed in the context of experimentally observed SiC particle distributions.

[1] Du, G., et al. (2008). "On-wafer investigation of SiC and Si3N4 inclusions in multicrystalline Si grown by directional solidification." Solar Energy Materials and Solar Cells 92(9): 1059-1066.

[2] Breitenstein, O., et al. (2004). "Shunt types in crystalline silicon solar cells." Progress in Photovoltaics: Research and Applications 12(7): 529-538.

[3] Breitenstein, O., et al. (2007). "Material-induced shunts in multicrystalline silicon solar cells." Semiconductors 41(4): 440-443.

[4] Möller, H. J., et al. (2009). "Improving solar grade silicon by controlling extended defect generation and foreign atom defect interactions." Applied Physics A 96(1): 207-220.

[5] Lotnyk, A., et al. (2008). "A TEM study of SiC particles and filaments precipitated in multicrystalline Si for solar cells." Solar Energy Materials and Solar Cells 92(10): 1236-1240.

[6] Tersoff, J. (1986). "New empirical model for the structural properties of silicon." Physical Review Letters 56(6): 632-635.

[7] Tersoff, J. (1988). "Empirical interatomic potential for silicon with improved elastic properties." Physical Review B 38(14): 9902-9905.

[8] Tersoff, J. (1994). "Chemical order in amorphous silicon carbide." Physical Review B 49(23): 16349-16352.

[9] Erhart, P. and K. Albe (2005). "Analytical potential for atomistic simulations of silicon, carbon, and silicon carbide." Physical Review B 71(3): 035211.

[10] Baskes, M. I. (1992). "Modified embedded-atom potentials for cubic materials and impurities." Physical Review B 46(5): 2727-2742.

[11] Luo, J., et al. (2017). "Atomistic simulations of carbon diffusion and segregation in liquid silicon." Journal of Applied Physics 122(22): 225705.