Mechanisms of abnormal grain growth, oxygen sensitivity elimination, and intergranular crack in SLM alloys via quasi-in-situ EBSD

• Main Author: Chen, Kewei
• Other Authors: Li Hua
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Materials::Material testing and characterization
• Online Access: https://hdl.handle.net/10356/160612
• Institution: Nanyang Technological University

Typically, there are three commonly-used techniques for microstructural characterization of selective laser melted (SLM) alloys, namely post-mortem, in-situ and machine learning (ML) based characterizations. Compared with post-mortem characterization, in-situ characterization allows the microstructural changes observed in real-time. However, ML extracts key microstructural information from the results of post-mortem and in-situ characterization. Currently, the microstructures behind the three issues of SLM alloys remain understood poorly, namely (a) abnormal grain growth (AGG) during annealing, (b) elimination mechanism of oxygen sensitivity, and (c) prediction of intergranular crack in annealed SLM alloys. However, the characterization methods mentioned above have some drawbacks to solve the three issues. For example, in-situ heating electron backscatter diffraction (EBSD) cannot follow the large-scale microstructural evolution at high temperature. Therefore, novel quasi-in-situ EBSD/scanning electron microscope (SEM) methods are proposed to address the three issues. Formation mechanism and avoidance of AGG in recrystallization of SLM commercially pure titanium (CP-Ti) and Inconel (IN)718 Usually, recrystallization annealing is used to weaken the texture, reduce anisotropy, and thus improve the mechanical properties. However, extensive AGG is often observed in the annealing process of SLM alloys, which is detrimental to the mechanical properties. The underlying formation mechanism of AGG is unclear. In order to explore the formation mechanism of AGG, the recrystallization process of SLM CP-Ti is followed by a vacuum sealing-based quasi-in-situ EBSD method, and the recrystallization process of SLM IN718 is followed by an argon protection-based quasi-in-situ EBSD method. It is found that AGG, primarily caused by strain-induced boundary migration (SIBM), initially emerges at the beginning of annealing of SLM CP-Ti at 800 °C. However, at the beginning of annealing of SLM CP-Ti at 700 °C, many small equiaxed grains are formed due to more recrystallization nucleation, followed by the formation of AGG resulting from the secondary recrystallization. As for SLM IN718, SIBM also dominates recrystallization at 1,150 and 1,075 ℃. AGG at 1,150 ℃ is attributed to a higher grain boundary (GB) velocity, by quickly consuming high-strain regions while fine grains at 1,075 ℃ is due to slow GB velocity, by allowing activation of more SIBM grains. Elimination of oxygen sensitivity in SLM Ti Although the addition of oxygen greatly increases the strength of Ti, the ductility would decrease significantly, which becomes a concern, briefly termed the oxygen sensitivity. It is recently reported that the oxygen sensitivity could be eliminated in SLM Ti, but no scientific explanation is provided. In order to get insight into the fundamental mechanism for elimination of oxygen sensitivity, the tensile processes of 0.13%O-Ti and 0.24%O-Ti samples are followed by quasi-in-situ EBSD/SEM method. It is found that the elimination of oxygen sensitivity is probably due to the oxygen segregation, promoting wavy slips, and intertwining microstructures, promoting the formation of multiple slip systems and preventing the propagation of intergranular crack. ML-based prediction of intergranular crack of annealed SLM CP-Ti Previous studies successfully predicted intergranular crack of conventionally produced alloys with equiaxed grains, through experiments and/or simulation. However, compared with conventionally produced alloys, annealed SLM alloys possess non-equiaxial grains with large different shapes and sizes. Analysis and prediction are still poor for the intergranular crack in annealed SLM alloys. In order to predict intergranular crack via ML-based models, a quasi-in-situ EBSD/SEM technique is used to follow the compression of annealed SLM CP-Ti to develop database for training and validation of ML-based models. It is found that, support vector machine (SVM) and random forest (RF) exhibit the best prediction of intergranular cracks with an accuracy of 98.7%. Additionally, the geometric compatibility factor m’ is identified as the most significant feature for the prediction. However, other features are also important and cannot be ignored for prediction, such as those regarding grain shape and size.