Shear-driven solid-state additive manufacturing of aerospace aluminum on impurity contaminated surfaces
We explore the environmental resilience of a next-generation solid-state additive process, additive friction stir deposition, highlighting its ability to print on surfaces contaminated by impurity particles. This is demonstrated by depositing a non-weldable aerospace alloy AA7075 onto Fe2O3 contamin...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-08-01
|
| Series: | Materials & Design |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S0264127525007324 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | We explore the environmental resilience of a next-generation solid-state additive process, additive friction stir deposition, highlighting its ability to print on surfaces contaminated by impurity particles. This is demonstrated by depositing a non-weldable aerospace alloy AA7075 onto Fe2O3 contaminated substrates, with the equivalent initial impurity thickness of several hundred microns. On the mesoscopic level, the extensive flow of AA7075 drives vertical migration of Fe2O3 particles, which are dispersed along the build direction. On the microscopic level, shear-induced deagglomeration reduces the Fe2O3 particle size from tens of microns to ∼ 100 nm, enabling their dispersion within fine, recrystallized AA7075 grains. These multi-scale dispersion mechanisms can prevent the formation of concentrated impurity layers or interfacial voids, rendering the best interface tensile strength of 269.05 MPa and elongation of 10.97 % along the vertical direction. With increasing impurities and local particle accumulation, however, both values decrease, resulting in the lowest interface strength of 96.94 MPa (with the maximum impurity inclusion) and the lowest elongation of 0.30 % (with the maximum void formation). Using high-throughput experimental design and mesoscopic feature extraction from X-ray computed tomography, a five-layer neural network is shown to effectively predict the interface strength with only a dozen training data. |
|---|---|
| ISSN: | 0264-1275 |