Machine-Learning-Assisted Design of Powder-Metallurgy Ti-Al-Nb Alloys Through Rare-Earth Microalloying: Microstructure and Mechanical Properties
Materials · 29.10.2026 · 19 (21) · 4568
DOI: 10.3390/ma19214568 ↗Kurz gefasst
Fachartikel in der Zeitschrift Materials, online erschienen am 29.10.2026. Verfasst von Xinze Li und 8 weiteren Autor:innen. Beteiligte Einrichtung: School of Materials Science and Engineering, Beijing Institute of Technology, Beijing 100081, China. Die Arbeit nennt 48 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Science and Technology Project of Hebei Education Department.
Abstract (Originalsprache)
Powder-metallurgy (PM) Ti-Al-Nb alloys are attractive for high-temperature structural applications but remain limited by poor room-temperature ductility. This study combines an exploratory machine-learning (ML) analysis with rare-earth (RE; La and Ce) microalloying to examine relationships among composition, thermal response, microstructure, and tensile properties. The ML dataset comprised 30 in-house Ti-Al-Nb experimental records. Four differential scanning calorimetry (DSC) descriptors—the peak areas and peak temperatures in the 800–900 °C and 1000–1100 °C ranges—were used to predict room-temperature yield strength. Six regression algorithms were compared using a 70/30 train–test split and five-fold cross-validation during hyperparameter selection. The gradient boosting model achieved the best test performance (R2 = 0.8752; MSE = 0.0089 on the normalized response). The model was used as a screening and interpretation aid rather than as a direct optimizer of the final sintering condition; the 1400 °C/3 h sintering schedule was an established process from prior laboratory work. For Ti-22Al-(25 − x)Nb-xRE alloys (x = 0–2 at.%), RE addition produced RE2O3-containing microstructures, reduced the mean grain size from 51.6 to 18.1 μm, and narrowed the O-phase laths from about 0.55 to 0.19 μm. At room temperature, the highest peak true stress was 572 MPa at 1.0 at.% RE, whereas the largest true fracture strain was 0.63% at 1.5 at.% RE. At 600 °C, the highest strength was 634 MPa at 0.5 at.% RE, while the largest true fracture strain was 0.98% at 1.5 at.% RE. Accordingly, 1.5 at.% RE provides the best overall strength–ductility balance across both test temperatures, although it does not maximize every individual property. The observed trends are consistent with a proposed RE–O redistribution and particle-pinning mechanism, but direct quantitative oxygen analysis and high-resolution phase mapping were not available; mechanistic statements are therefore treated as evidence-supported interpretations rather than definitive proof.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- School of Materials Science and Engineering, Beijing Institute of Technology, Beijing 100081, China.
- Tianjin Key Laboratory of Integrated Design and On-Line Monitoring for Light Industry & Food Machinery and Equipment, School of Mechanical Engineering, Tianjin University of Science and Technology, Tianjin 300457, China.
- Material Science Research Center, Xingtai University, Xingtai 054001, China.
ORCID-Profile: Qi Cai
Förderung
- National Natural Science Foundation of China
- Science and Technology Project of Hebei Education Department
Zitieren
APA
Li, X., Yu, Q., Xia, G., Yan, X., Li, Z., Luo, C., Wang, Q., Zhang, Y., & Cai, Q. (2026). Machine-Learning-Assisted Design of Powder-Metallurgy Ti-Al-Nb Alloys Through Rare-Earth Microalloying: Microstructure and Mechanical Properties. Materials, 19(21), 4568. https://doi.org/10.3390/ma19214568
Vancouver
Li X, Yu Q, Xia G, Yan X, Li Z, Luo C, et al. Machine-Learning-Assisted Design of Powder-Metallurgy Ti-Al-Nb Alloys Through Rare-Earth Microalloying: Microstructure and Mechanical Properties. Materials. 2026;19(21):4568. doi:10.3390/ma19214568
BibTeX
@article{Li2026,
author = {Li, Xinze and Yu, Qilin and Xia, Guoqing and Yan, Xifeng and Li, Zhu and Luo, Chaoqun and Wang, Qi and Zhang, Yaran and Cai, Qi},
title = {Machine-Learning-Assisted Design of Powder-Metallurgy Ti-Al-Nb Alloys Through Rare-Earth Microalloying: Microstructure and Mechanical Properties},
journal = {Materials},
year = {2026},
volume = {19},
number = {21},
pages = {4568},
doi = {10.3390/ma19214568}
}
EUVIMED