Ti-6Al-4V microstructural orientation at different length scales as a function of scanning strategies in Electron Beam Melting in additive manufacturing

Additive manufacturing has been around for many years, yet the underlying physics of thermal gradients, local pressure environment, and other non-steady state manufacturing conditions are not fully understood. A Multi-University Research Initiative (MURI) is currently ongoing to measure liquid/solid and solid/solid interface stabilities in AM Ti-6Al-4V. Samples were produced with different beamscanning strategies in order to study the role of thermal gradients on the resulting microstructure. The motivation is to determine which beam-scanning strategy leads to desired grain size and texture. Orientation at different length scales (from mm to nm) can be quantified and compared with a combination of techniques including Precession Electron Diffraction (PED), Electron Backscatter Diffraction (EBSD) and Neutron diffraction. This new information will help predict properties of additively manufactured parts. Disciplines Manufacturing | Materials Science and Engineering | Structural Materials Comments This proceeding is published as Agrawal, Priyanka, Maria J. Quintana, Matt Kenney, Sabina Kumar, Alec Saville, Amy Clarke, and Peter C. Collins. "Ti-6Al-4V microstructural orientation at different length scales as a function of scanning strategies in Electron Beam Melting in additive manufacturing." In MATEC Web of Conferences 321 (2020): 03031. DOI: 10.1051/matecconf/202032103031. Posted with permission. Creative Commons License This work is licensed under a Creative Commons Attribution 4.0 License. Authors Priyanka Agrawal, Maria J. Quintana, Matthew Kenney, Sabina Kumar, Alec Saville, Amy Clarke, and Peter C. Collins This conference proceeding is available at Iowa State University Digital Repository: https://lib.dr.iastate.edu/ mse_conf/50 MATEC Web of Conferences 321, 03031 (2020) https://doi.org/10.1051/matecconf/202032103031 The 14 World Conference on Titanium © The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (http://creativecommons.org/licenses/by/4.0/). Ti-6Al-4V microstructural orientation at different length scales as a function of scanning strategies in Electron Beam Melting in additive manufacturing Priyanka Agrawal1, Maria J. Quintana1, Ma� Kenney1, Sabina Kumar2, Alec Saville3, Amy Clarke3, Peter C. Collins1 1Iowa State University, Materials Science and Engineering Department, Ames, IA, United States 2University of Tennessee, Department of Materials Science and Engineering Department, Knoxville, TN, United States 3 Colorado School of Mines, Metallurgical and Materials Engineering Department, Golden, CO, United States


Introduc on
A principal advantage of manufacturing parts by Addi ve Manufacturing (AM) system is the ability to obtain complex geometries, different types of composites/alloys or even gradients of composi on in the same part, as it follows a layer-by-layer build up approach. Therefore, when each layer is been printed, it is possible to melt 4-5 layers below it, as per the thermal conduc vi es and thermal diffusivi es of the material used. This has thusly created the necessity of understanding the rela onships between process, composi on, microstructure and proper es [1][2][3][4], as well as crea ng mathema cal and computa onal models that can predict the proper es of the final printed parts [4]. However, in order to do so, it is necessary to fully understand the fundamentals that dictate the microstructure (defects, phase frac ons, morphologies, texture, etc.) and mechanical proper es as a func on of thermal gradients, local pressure environments and solidifica on strategies ( Figure 1) [5][6][7][8][9]. The Rosenthal equa on [10] gives the three-dimensional steady state temperature field for a point source. While it is a straight forward analy cal solu on, it does not account for complex boundary condi ons and non-steady state condi ons, nor for latent heat of fusion or convec on. A recent semi-analy cal heat model has been developed at Oak Ridge Na onal Laboratory Manufacturing Demonstra on Facility (ORNL MDF), to tackle the above-men oned complexi es [11], based on a transient solu on, to construct a temperature field for an arbitrary beam path, from which informa on can be extracted regarding thermal gradients and interphase growth veloci es of a layer of a build, both essen al in understanding the microstructural features of the build such as phase selec on (or variant selec on in Ti-6Al-4V), morphological features of the phase.
Electron-Powder Bed Fusion systems are known to create complex heat and mass transfer conditions that vary throughout the build as the heat source creates a melt pool that moves following a given scanning strategy [12]. When each layer is been printed, it is possible to melt 4-5 layers below it, as per the thermal conductivities and thermal diffusivities of the material used. https://doi.org/10.1051/matecconf/202032103031 The 14 th World Conference on Titanium The proper es of AM parts, as with any other manufacturing process, are a func on of composi on, microstructure (phases, grain size and morphology, etc.), orienta on, texture, defects (created during the manufacturing process) and superficial characteris cs [7]. Each one of these parameters can be analyzed using different equipment and techniques, as shown in Figure 2. The goal of this Mul disciplinary University Research Ini a ve (MURI) funded by the Office of Naval Research is to understand how the varia on of composi on and thermal gradients, both spa ally and temporally, will result in differences in liquid-solid interface veloci es, thermal gyra ons and elas c-plas c stress/strain gradients as a func on of geometry and energy deposi on while manufacturing the part; in other words understanding how the build parameters influence local condi ons that result in specific microstructural features, defects and heterogenei es.

Materials and Experiments
ORNL MDF additively manufactured Ti-6Al-4V samples with 3 different scan strategies (shown schematically in Figure 3), namely raster (or line scan) (L) and two spot melt scans, random fill (R) and Dehoff fill (D), in order to evaluate and understand the complexities of AM fundamentals previously mentioned, including the spatial and temporal thermo-mechanical distributions in AM parts. The chosen design was to fabricate cuboids that have a wide range of thermal gradients and interface velocities. These cuboids were built on an ARCAM Q10 Plus machine as: 15 x 15 x 25 mm with a layer height of 50μm. The builds were made under a vacuum of 4.5x10 -2 mBar and maintained at a preheat temperature of 470°C. In this multi-university project, twelve samples of each melting strategy were built concurrently, Iowa State University was provided with samples designated as number 5 (L5, R5 and D5).  The samples for microscopy were sec oned using wire cu ng -EDM (electrical discharge machining). The cuts were designed to provide access to different cross-sec onal planes of the build that would be necessary for imaging using scanning electron microscopy (SEM) and transmission electron microscopy (TEM), electron backsca�er diffrac on (EBSD), precession electron diffrac on (PED), and, in the future, for spa ally resolved acous c spectroscopy (SRAS) analysis. A por on of the samples' XY plane (perpendicular to build direc on) and the complete XZ plane (parallel to build direc on) were ground and polished using tradi onal techniques and an overnight polishing on Vibromet (Buehler VibroMet 2 Vibratory Polisher).
Scanning electron micrographs were taken with a FEI Teneo LoVac SEM in BSE mode. BSE images were taken for all the three samples at different loca ons from bo�om to top of the build in XZ plane at different magnifica ons. EBSD analysis helped confirm the qualita ve results of the SEM images using Oxford EBSD detector. 350 x 350 µm area maps with a 0.5 µm step size and 8 x 8 binning were collected at the top, middle and bo�om of each sample on the XZ plane.
TEM samples were prepared with a FEI Helios DualBeam TM (focused ion beam/SEM) from the bottom, middle, and top of the samples in the XZ plane. PED scans were carried out to obtain crystallographic information at nanoscale and quantify the dislocation density (density of dislocations) affected by the three different scan strategies using a MatLab code developed previously [13]. Scans for PED [14] with 2.5 x 4 µm area were run on the TEM foils using ASTAR on FEI Tecnai G2-F20 STEM with 10 nm step size, 0.917 precession angle, 10 precessions per frame and 0.65 gamma. The camera length was set to 77 mm at smallest C2 aperture and smaller spot size.

Results and Discussion
Scanning electron microscopy was used to characterize microstructures from the builds made with three scan strategies along and across the build direc on. A qualita ve comparison of these samples from SEM showed that a columnar β morphology was present in all of the builds. The raster scanning strategy (L5) has a finer morphology when compared to the two-point mel ng strategies (R5 and D5): the center of the builds have the following mean grain sizes: 6.36 µm for L5, 9.81 µm for D5 and 16.7 µm for R5. All samples have both colony and basket-weave type microstructures; however, the frac on of colonies is different: 5% for L5, 30% for D5 and 50% for R5 (all measured at the center of the builds in the XZ plane). An example of change in microstructures from process type and posi on is presented in Figure 4. From SEM imaging and textures studies, it is evident that the bo�om of all three samples is primarily basket-weave and it transi ons to a more colony microstructure as we travel higher up the build. By the middle of the sample, the hea ng/cooling cycles appear to reach some steady state and as a result, the microstructure of the top and middle are largely the same.
Although comprehensive microstructural gradients were imaged across the different length scales using SEM, the crystallographic informa on is missing. As men oned earlier, the crystallographic informa on is necessary to understand the effect of different scan strategies (or spa al varia on of thermal gradients) on the solid-solid phase transforma ons. Therefore, we obtained micro-texture informa on is obtained using SEM-EBSD and nano-texture from PED ASTAR using TEM.
An analysis of texture for a grains from EBSD ( Figure 5) shows changes along the build direc on. Since the texture of α is caused by two factors (i.e., the parent β grain orienta on and whether the microstructure is colony or basket-weave), a certain amount of local varia on is to be expected. For example, when comparing the bo�om and top of sample D5 shown in Figure 5(a), the difference in the IPF maps corresponds with the differences in the microstructural observa ons made in the SEM images. While this visual correspondence is useful, one advantage of conduc ng diffrac on experiments is the representa on of the crystallographic informa on as pole figures. In this texture analysis, two different types of diffrac on experiments have been conducted -EBSD and neutron diffrac on. The differences are that EBSD is a localized diffrac on experiment, while neutron diffrac on can be used to assess both bulk and local texture. While there are differences in the pole figures obtained from EBSD data in Figure 5(b) when compared with those from neutron diffrac on data represen ng the bulk specimen in Figure 5(c), both sets of results are nominally consistent with a parent β grain orienta on whose (001) plane is nominally parallel with the z direc on and whose α precipitates obey a Burger's Orienta on Rela onship with the parent β grains. For example, when looking at the (0002) for the neutron diffrac on data, most of the informa on lies at ~45° away from a center posi on. We note that the center of this (0002) data is offset slightly from the center of the pole figure, corresponding with a " lt" in the parent β grains, consistent with observa ons previously made on op cal images. The EBSD data, though represen ng the data from a different perspec ve (i.e., the y perspec ve) is also consistent, with the (0002) laying at 45 and 90° to the center of the pole figure, and consistent with the aforemen oned grain orienta on and orienta on rela onship. PED ASTAR experiments give crystallographic informa on at nanoscales but for the current work they are performed on thin TEM foils to obtain the quan ta ve informa on of the defect density (the density of disloca ons in the printed samples). The PED scans are carried out for the TEM foils obtained from all the three samples from top, middle and bo�om of the build are discussed below. The foils milled from XZ plane will lead to a foil containing the XY plane. Figure 6 shows preliminary data from one of the foils in PED ASTAR. The scans obtained are directly from the raw data without any cleaning steps with the op mized parameters men oned in the previous sec on.

Conclusion
There is an interes ng correla on between micrographs and texture informa on at different length scales and different build posi ons for all the three samples. The microstructure (both grain size and colony frac on) varies as a func on of build height and scanning strategy, however a more detailed analysis is required, which is underway, to fully explain the results observed.