Machine-Learning-Based Atomistic Model Analysis on High-Temperature Compressive Creep Properties of Amorphous Silicon...

Area of Science:

• Materials Science
• Computational Materials Science
• Nanotechnology

Background:

• Silicon carbide (SiC) ceramic matrix composites (CMCs) are vital for high-temperature applications, particularly in turbine engines.
• Accurate prediction of mechanical properties at elevated temperatures is critical for the reliability and lifespan of SiC-CMC components.

Purpose of the Study:

• To develop a novel interatomic potential function for silicon-carbon systems using artificial neural networks (ANNs).
• To investigate the high-temperature mechanical properties of SiC materials, with a focus on amorphous SiC relevant to CMC applications.

Main Methods:

• Development of an ANN-based interatomic potential for Si-C systems.
• Validation of the ANN potential against first-principles calculations for SiC, Si, and C single crystals.
• Simulation of amorphous SiC using the ANN potential, including radial distribution function analysis and creep tests.

Main Results:

• The ANN potential accurately reproduced material properties of SiC, Si, and C single crystals.
• The potential demonstrated applicability to amorphous SiC simulations.
• Creep tests on amorphous SiC revealed two distinct creep behaviors governed by different atomistic mechanisms, dependent on strain rate.

Conclusions:

• The developed ANN potential is a reliable tool for simulating SiC materials, including amorphous phases.
• The study highlights the significant role of amorphous regions in the creep behavior of SiC composites.
• Lower evaluated activation energies suggest amorphous SiC influences creep dynamics more than previously understood.