Microstructural characterization and hardness prediction of AlCuCrFeNi high entropy alloys using Transformer-based Physics-Informed Neural Networks (T-PINN)
This research investigated a novel approach to generating dual-phase high entropy alloys (HEAs) from recycled waste materials, with the objective of attaining sustainable manufacturing while maintaining material performance. The AlCuCrFeNi composition was synthesized using vacuum arc melting (VAM) t...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-09-01
|
| Series: | Journal of Materials Research and Technology |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S223878542501974X |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This research investigated a novel approach to generating dual-phase high entropy alloys (HEAs) from recycled waste materials, with the objective of attaining sustainable manufacturing while maintaining material performance. The AlCuCrFeNi composition was synthesized using vacuum arc melting (VAM) to achieve the desired equiatomic composition. The HEAs were characterized using X-ray diffraction (XRD) and field emission scanning electron microscopy (FESEM). Synthesized alloys were heat-treated at different temperatures to enhance phase stability, induce microstructural evolution and hardness performance. The findings indicate that the AlCuCrFeNi HEAs have a dual-phase composition, characterized by a Body-Centered Cubic (BCC) and Face-Centered Cubic (FCC) structure, with a dendritic and interdendritic microstructure. Microstructural alterations were detected at each heat treatment temperature. Grain formation commenced at 950 °C, and the BCC phase dissolved around 1100 °C. The highest microhardness value of 353.4 HV was achieved for the alloy subjected to heat treatment at 950 °C. To predict the hardness properties quantitatively, a Machine Learning (ML) model using a Transformer-based Physics-Informed Neural Network (T-PINN) was developed. The T-PINN model demonstrated high accuracy, with a Root Mean Square Error (RMSE) - 0.048, Mean Absolute Error (MAE) of- 0.036, and R-square (R2) - 0.97 values, respectively. This research has highlighted the potential of ML to advance materials with enhanced mechanical properties. |
|---|---|
| ISSN: | 2238-7854 |