Accelerated First-Principles Calculations Based on Machine Learning for Interfacial Modification Element Screening of...

Area of Science:

• Materials Science
• Computational Materials Science
• Nanotechnology

Background:

• Silicon Carbide (SiCp)/Aluminum (Al) composites are valuable for aerospace and instrumentation due to their lightweight and high strength.
• High-temperature interfacial reactions between SiC and Al degrade composite properties.

Purpose of the Study:

• To investigate the effect of alloying elements on the interfacial properties of SiCp/Al composites.
• To develop a machine learning model for accelerated computation of interface energies.

Main Methods:

• First-principle calculations were used to compute interface segregation and binding energies.
• Feature engineering was applied to refine machine learning descriptors.
• Six machine learning models (RBF, SVM, BPNN, ENS, ANN, RF) were trained, with ANN selected based on R² and MSE.
• Machine learning-accelerated computation was performed for 89 elements.

Main Results:

• Elements such as B, Si, Fe, Co, Ni, Cu, Zn, Ga, and Ge were identified as dual-functionality elements.
• These elements inhibit interfacial reactions and enhance interfacial binding.
• Atomic-scale mechanisms for interfacial modulation by these elements were elucidated.

Conclusions:

• The study provides a theoretical basis for designing SiCp/Al composite compositions.
• Optimizing alloying elements can significantly improve the high-temperature performance and stability of SiCp/Al composites.