Comprehensive study on the mechanical properties of Ti-doped Si3N4 ceramics: Experimental analysis, DFT calculations, and machine learning

• Published in: Ceramics international Vol. 50; no. 9; pp. 16352 - 16360
• Main Authors: Wu, Jing, Xu, Zhenru, Liu, Xunyong, Tang, Huaguo, Li, Tongyang, Qiao, Zhuhui
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01-05-2024
• Subjects: First principle; Fracture toughness; Machine learning; Si3N4; TiN; TiN; Si3N4; First principle; Machine learning; Fracture toughness
• ISSN: 0272-8842; 1873-3956
• DOI: 10.1016/j.ceramint.2024.02.119

This study involved the synthesis of Si3N4 ceramics, focusing on adjusting TiN content and sintering temperatures to enhance mechanical properties. Utilizing Density Functional Theory (DFT), we examined the microstructure and elastic properties of Ti-doped β-Si3N4, providing insights into how the Ti atom influences mechanical behavior. Additionally, a machine learning approach was applied to establish a predictive model for Si3N4 fracture toughness, factoring in material composition and the sintering process. Remarkably, TiN incorporation led to a substantial increase in hardness, reaching 16.63 GPa. At 1800 °C, the resulting material at 10 wt% and 5 wt% TiN content exhibited impressive flexural strength (924 MPa) and fracture toughness (8.48 MPa m1/2). The theoretically calculated elastic properties confirmed the benefits of Ti atom doping in enhancing Si3N4 ceramics. By employing the relu activation function in combination with the rmsprop optimizer, we achieved a coefficient of determination (R2) for the test set no less than 0.824, ensuring reliable predictions for TiN–Si3N4 fracture toughness. This study provides valuable guidance for swiftly predicting high-performance Si3N4 ceramics and introduces a novel research avenue for the development of advanced materials.