A simple transferable adaptive potential to study phase separation in large-scale xMgO-(1-x)SiO2 binary glasses

Area of Science:

• Materials Science
• Computational Chemistry
• Geochemistry

Background:

• Simulating phase separation in binary systems like MgO-SiO2 is crucial for understanding material properties.
• Fixed-charge models often struggle to accurately represent the diverse compositions of separated phases.
• Previous models were limited to pure cluster formation or the modified random network model.

Purpose of the Study:

• To develop a simple, transferable adaptive model for simulating phase separation in the MgO-SiO2 system.
• To overcome limitations of fixed-charge models in representing varied phase compositions.
• To investigate the formation of amorphous nanoparticles in melts.

Main Methods:

• Development of a novel transferable adaptive molecular dynamics model.
• Simulation of the MgO-SiO2 binary system from a melt composition of 0.1MgO-0.9SiO2.
• Tracking the formation and composition of separated phases.

Main Results:

• The adaptive model successfully simulates large phase separation in MgO-SiO2, consistent with experimental and phase diagram predictions.
• The model accurately reproduces known crystalline and glassy structures.
• Formation of Mg-rich, Si-poor amorphous nanoparticles within an Mg-poor, Si-rich matrix was observed.

Conclusions:

• The developed adaptive model offers a significant improvement over fixed-charge models for simulating phase separation.
• This model enables the study of complex phase behaviors and nanoparticle formation in binary melts.
• The findings provide new insights into the microstructural evolution of MgO-SiO2 materials.