Carbon solubility in liquid silicon: A computational analysis across empirical potentials

Area of Science:

• Materials Science
• Computational Chemistry
• Semiconductor Physics

Background:

• Silicon carbide (SiC) precipitation in liquid silicon is crucial for silicon crystallization in the photovoltaic industry.
• Accurate modeling of SiC nucleation and growth requires understanding carbon solubility and phase equilibria.

Purpose of the Study:

• To calculate the solubility of carbon in liquid silicon relative to the beta-SiC phase.
• To evaluate the suitability of empirical potentials for simulating SiC precipitation.
• To identify limitations of existing potentials and suggest improvements.

Main Methods:

• Utilized statistical thermodynamic techniques for solubility calculations.
• Employed and assessed various empirical potentials, including Tersoff-type potentials.
• Performed chemical potential calculations for pure silicon.

Main Results:

• Some Tersoff-type potentials accurately model carbon solubility in liquid silicon.
• These potentials overestimate the silicon melting point, limiting their use to high temperatures irrelevant for typical solidification.
• The identified weakness is confined to the liquid phase description.

Conclusions:

• Existing empirical potentials have limitations for simulating SiC precipitation under typical industrial conditions.
• Recent silicon potential models show promise in addressing the liquid phase description issues.
• Improved potentials could enhance the computational analysis of SiC precipitation in silicon crystallization.