3D characterization of microstructural evolution and variant selection in additively manufactured Ti-6Al-4 V

Abstract

Ti-6Al-4 V is a popular alloy in additive manufacturing (AM) due to its applications in the biomedical implants and aerospace industries where the complex part geometries allowed by AM provide cost and performance benefits. Ti-6Al-4 V goes through a b ? a' transformation after solidification which is known to experience variant selection, e.g., through the formation of clusters of variants which, when situated together, partially accommodate the strain of the phase transformation. During electron beam powder bed fusion AM, an in situ decomposition of a' martensite occurs during the cyclic reheating caused by melting successive layers, resulting in a ? b microstructures. How variant selection influences the evolution beyond the initial rapid cooling remains an open question. Using 3D electron backscatter diffraction, we provide a clearer understanding without ambiguity from sectioning effects of how a' decomposes into microstructures with distinct morphologies and variant/intervariant distributions. We extract quantitative 3D information on the various intervariant boundaries networks formed in samples printed using three different electron beam scanning strategies. This shows that differing mechanisms during the decomposition result in a shift from self-accommodating clusters in an acicular microstructure, to either the preferred growth of six variants in a basketweave microstructure, or to a colony microstructure where variant selection is determined by prior-b grain boundaries. We propose a new representation of the misorientations arising from the Burgers orientation relationship, which we refer to as intervariant network diagram, to reveal how variant selection during the martensitic transformation and subsequent decomposition leads to the intervariant boundary networks observed. This holistic understanding of the microstructural evolution has the potential to allow tailoring of microstructures and properties for specific applications.

A new approach to three-dimensional microstructure reconstruction of a polycrystalline solar cell using high-efficiency Cu(In,Ga)Se2

“…X-ray methods such as 3D X-ray diffraction (3DXRD) 15 and diffraction computed tomography (DCT) 16 are non-destructive techniques and allow for the measurement of 3D structures. However, the resolution of 3D X-ray CT methods are in microns and much poorer compared to electron microscope serial sectioning 3D EBSD that are in tens of nanometers 15 – 19 . Despite the development of various imaging methods, 3D structure analysis is not widely used because of the complexity of 3D data acquisition and processing 9 .…”

The use of defects and compositional variations to elucidate physical phenomena in electron beam melted Ti-6Al-4V across scan strategies

“…Qualitatively, the random scan strategy sample has the largest α lath sizes as well as the highest fraction colony (Figure 4.9e and f), while the raster scan strategy sample has the smallest α laths and least fraction colony (Figure 4.9c and d). While a quantitative assessment of the microstructure was not conducted as part of this work, previous analyses have been published on these samples [27] as well as other samples from the same build [27,[135][136][137][138][139][140][141]155,156] supporting the qualitative conclusions. The difference in microstructure is primarily a result of cooling rate differences, with the raster sample cooling the fastest, enabling the formation of a basketweave structure, and the random sample cooling the slowest [27].…”

“…In addition, parent β grain reconstruction (Figure 3.21b and Figure 3.22b) using the MTEX Toolbox in MATLAB can help enhance understanding of the interaction of defects and grain growth. Firstly, as expected in raster AM parts [27,[135][136][137][138][139], the build direction and the columnar growth direction are not perfectly parallel, as there is a slight tilt of the growth axis to the left (for Figure 3.21) as a result of the beam (and energy) rastering in the XY-plane, creating thermal gradients that are not aligned with the Z-axis. Secondly, the presence of overhanging partially melted particles, like defect 1 from Figure 3.21a, and the larger defect in Figure 3.22, can nucleate new small grains, and will have an influence on the local texture.…”

“…The lower Al regions, being held above the local β-transus, experienced the α→β transformation. Upon cooling, the β phase transformed back into α in such a manner (controlled by the cooling rate) that the subsequent variant selection [135,136,139,140,164] enabled significant colony growth. The higher Al regions, which did not exceed the local β-transus, did not experience the α→β transformation and thus retained the primarily basketweave microstructure observed before the heat treatment.…”

“…The microstructure (and texture) of the individual scan strategies of the batch one samples has already been extensively characterized in publications from the MURI team at large: [27,[135][136][137][138][139][140][141]155,156]. Note that while these cited analyses were performed on different individual samples, in each case the samples were from the same batch one build, and, thus, were created under the same conditions at the same time.…”