This study investigated the in-situ gaseous (under 150 bar) hydrogen embrittlement behaviour of additively manufactured (AM) Inconel 718 produced from sustainable feedstock. Here, sustainable feedstock refers to the Inconel 718 powder produced by vacuum induction melting inert gas atomisation of failed printed parts or waste from CNC machining. All Inconel 718 samples, namely AM-as-processed, AM-heat-treated and conventional samples showed severe hydrogen embrittlement. Additionally, it was found that despite its higher yield strength (1462 ± 8 MPa) and the presence of δ phase, heat-treated AM Inconel 718 demonstrates 64% lower degree of hydrogen embrittlement compared to the wrought counterpart (Y.S. 1069 ± 4 MPa). This was linked to the anisotropic microstructure induced by the AM process, which was found to cause directional embrittlement unlike the wrought samples showing isotropic embrittlement. In conclusion, this study shows that AM Inconel 718 produced from recycled feedstock shows better hydrogen embrittlement resistance compared to the wrought sample. Furthermore, the unique anisotropic properties, seen in this study for Inconel 718 manufactured by laser powder bed fusion, could be considered further in component design to help minimise the degree of hydrogen embrittlement.

Recycling steel at scale is hindered by tramp elements such as Cu and Sn, which degrade material properties. Atomistic simulations using foundational machine-learned interatomic potentials (MLIPs) trained on large databases, such as Materials Project, Alexandria, and OMAT, offer a promising approach to study the effects of these impurities. However, fine-tuning these models to specific systems can lead to catastrophic forgetting–the loss of general chemical knowledge acquired during pretraining. Here, we evaluate forgetting in three foundational MLIPs: CHGNet, SevenNet-O, and MACE, by fine-tuning on a data set of bcc-based structures, with Fe atoms only. When evaluated on a subset of the Materials Project data set with a learning rate of 0.0001, the fine-tuned MLIPs of CHGNet and SevenNet-O exhibited only a minor increase in RMSE of 0.047 and 0.022 eV/atom, respectively, indicating markedly minor forgetting. In contrast, fine-tuned MACE exhibited catastrophic forgetting, despite a range of additional strategies such as layer freezing and data set replay. We attribute the catastrophic forgetting to architectural sensitivity. These results highlight the importance of fine-tuning hyperparameters, model architecture, and data set design, with fine-tuned models of CHGnet and SevenNet-O showing some potential for efficient and transferable modeling of recycled steels.