Materials Characterization 145 (2018) 29–38 Contents lists available at ScienceDirect Materials Characterization journal homepage: www.elsevier.com/locate/matchar Solidiﬁcation pattern, microstructure and texture development in Laser Powder Bed Fusion (LPBF) of Al10SiMg alloy T ⁎ Hong Qina, Vahid Fallaha, , Qingshan Donga, Mathieu Brochub, Mark R. Daymonda, Mark Gallerneaulta a b Department of Mechanical and Materials Engineering, Queen's University, Kingston, ON K7L 3N6, Canada Department of Materials, McGill University, Montreal, QC H3A 0C5, Canada A R T I C LE I N FO A B S T R A C T Keywords: Additive manufacture Laser Powder Bed Fusion AlSi alloys Solidiﬁcation microstructure Crystallographic texture Cellular structure A comprehensive analysis of solidiﬁcation patterns and microstructural development is presented for an Al10SiMg sample produced by Laser Powder Bed Fusion (LPBF). Utilizing a novel scanning strategy that involves counter-clockwise rotation of the scan vector by 67° upon completion of each layer, a relatively randomized cusp-like pattern of protruding/overlapping scan tracks has been produced along the build direction. We show that such a distribution of scan tracks, as well as enhancing densiﬁcation during LPBF, reduces the overall crystallographic texture in the sample, as opposed to those normally achieved by commonly-used bidirectional or island-based scanning regimes with 90° rotation. It is shown that, under directional solidiﬁcation conditions present in LPBF, the grain structure is strictly columnar throughout the sample and that the grains' orientation aligns well with that of the α-Al cells. The size evolution of cells and grains within the melt pools, however, is shown to follow opposite patterns. The cells'/grains' size distribution and texture in the sample are explained via use of analytical models of cellular solidiﬁcation as well as the overall heat ﬂow direction and local solidiﬁcation conditions in relation to the LPBF processing conditions. Such a knowledge of the mechanisms upon which microstructural features evolve throughout a complex solidiﬁcation process is critical for process optimization and control of mechanical properties in LPBF. 1. Introduction Laser Powder Bed Fusion (LPBF) is one of the most popular metal Additive Manufacturing (AM) processes that is based on a layer-bylayer manufacturing consisting of re-melting/solidiﬁcation of the layered powder by a high energy input [1–4]. This technique has attracted increased interest in the production of high value, low production volume parts where it has strong advantages in creating complex geometries [1,2], particularly in applications where thermal and cooling ﬂuid management would be impossible to create using conventional casting and machining operations . Al10SiMg is commonly used in aerospace and transportation applications on account of its low cost, excellent (speciﬁc density) mechanical properties and good thermal conductivity . Applying LPBF to these materials increases the cooling rate during processing yielding a ﬁner microstructure [4–7] and thus an improvement in the as-produced (as-cast) mechanical properties compared to the same alloy cast conventionally . Cooling rates as high as 106 Ks−1 have been found to exist during LPBF [5,8], while the cooling rates attainable via conventional casting processes are ⁎ normally about 3–4 orders of magnitude lower, e.g. ~300 Ks−1 in strip casting of Al alloys . A complex thermal cycle develops within the LPBF part being built, which is caused by a highly concentrated heat source (generated by a focused laser beam) that rapidly moves along a ﬁnely resolved scanning pattern in each layer [4,10]. Such a heat cycle may give rise to development of an unsteady heat ﬂow, a condition that is further exacerbated in LPBF of metals with higher thermal conductivity and reﬂectivity, such as Al . As a result, melt pools generated in LPBF of Al alloys are known to ﬂuctuate substantially in size throughout the process [5,11]. Due to such uncertainty in melt-pool sizes as well as the lower absorption of laser light due to high reﬂectivity, LPBF of Al alloys is normally performed at relatively high laser powers, under which condition the melt pools penetrate occasionally as deep as four layers underneath the current track . Also, the selected scanning regime can further enhance such melt pool instabilities. Due to the repetitive nature of commonly-used scanning regimes within successive layers along the build direction, local heat concentrations normally occur as a result of coinciding/overlapping scanning vectors and their start/end Corresponding author. E-mail address: firstname.lastname@example.org (V. Fallah). https://doi.org/10.1016/j.matchar.2018.08.025 Received 17 May 2018; Received in revised form 24 July 2018; Accepted 15 August 2018 Available online 18 August 2018 1044-5803/ © 2018 Elsevier Inc. All rights reserved. Materials Characterization 145 (2018) 29–38 H. Qin et al. in this study based on maximizing the interval number of printed layers, N (as deﬁned and explained in ref. ), with the same deposition direction. It is shown by Guan et al.  that the larger N value the lower the residual stresses and higher the densities within the build; this results in improved mechanical properties. Recently, the hatch angle of 67° has been adopted in the standard operating parameters of some commercial SLM systems including the Renishaw AM400. Due to such successive rotation in subsequent layers, as opposed to a simple bidirectional regime, the local heat concentration will be reduced across the part being built. Such local heat concentrations are rather unsteady  and normally occur due to coinciding/overlapping scanning vectors and their start/end points within successive layers along the build direction. Such conditions give rise to the formation of occasional key-hole melt pools during the LPBF processing thus promoting formation of microscopic solidiﬁcation/shrinkage porosity . The relative density was measured using Archimedes principle by weighing the sample in air as well as in methanol. The measured density, ~99%, competes with the highest Archimedes densities reported for LPBF of Al10SiMg alloy, such as those reported by Thijs et al. . Referring to the specimen coordinates shown in Fig. 1(b), X-Y (top view), Y-Z (side view) and X-Z (front view) sections were cut from the as-fabricated part and used for the microstructural examinations. The samples were prepared using common metallographic techniques for aluminum alloys. As a reference, one specimen was etched in Keller's reagent while the other samples were ion-milled using a PECS II system. Ion-milling was used rather than electropolishing in order to avoid the possibility of silica contamination. Microstructural analysis of the asfabricated Al10SiMg sample cross-sections was performed on a FEGSEM (FEI Nova NanoSEM 400), using backscatter and secondary electron imaging. Grain sizes and crystallographic orientation information were obtained using ECC imaging (under 10 keV and 7 mm working distance) and EBSD analysis was carried out with a step size of 0.5 μm. EBSD was carried out using a Bruker e-FlashHR detector and the acquired data was analyzed in an Oxford Instruments HKL Channel 5 software. points [5,12]. Such heat concentrations, that are rather unsteady , give rise to the formation of occasional key-hole melt pools during LPBF processing thus enhancing porosity formation [5,12,14]. This is commonly observed in LPBF of Al10SiMg under highly repetitive scanning strategies such as the so-called uni- or bi-directional regimes [5,12,14–16]. To boost densiﬁcation in LPBF, researchers have rather utilized island-based scanning regimes where each layer is divided into several islands. The scan vectors are then alternated by 90° in orientation amongst neighbouring islands in each layer, while also being slightly shifted in location with respect to the corresponding islands within the previous layer. Particularly, Thijs et al.  have shown that utilization of such an island-based scanning strategy signiﬁcantly enhances densiﬁcation, as well as inducing an overall reduction in the modes and intensities of crystallographic texture within the LPBF part. To further explore the eﬀect of scanning regimes on enhancing heat ﬂow stability and uniformity during LPBF, in this study, we utilize a new strategy where the bidirectional scan vectors are rotated counterclockwise by 67° with respect to the previous layer. Furthermore, thanks to the unique combination of analytical, sample preparation and electron microscopy techniques utilized in this study, we have been able to conduct a comprehensive and highly detailed analysis of cells/grains structure and texture development as well as the evolution of various types of Si phase during LPBF. It is of a great importance to understand the solidiﬁcation patterns and the resultant microstructural features in relation to the processing conditions, as such features are the major contributors to the ﬁnal strength in the LPBF parts [5,15]. Obtaining such a knowledge can be a crucial factor in tailoring the LPBF process design in order to eﬀectively control the microstructure, residual stresses and defects formation, and thus to engineer the ﬁnal physical and mechanical properties. This study presents a fundamental approach to accurately resolve the solidiﬁcation and phase transformation patterns that lead to the microstructural development during one of the more complex LPBF processing conditions; i.e. an LPBF process that utilizes a more isotropic scanning pattern throughout the part, presented here for the ﬁrst time. The focus of the present study is to obtain a solid knowledge of microstructure development in LPBF of Al10SiMg alloy, while the mechanical properties of printed parts will be investigated in a future study. The test specimens, prepared using a variety of techniques from chemical etching to ion-milling, are characterized using high-resolution Scanning Electron Microscopy (SEM) for a detailed observation of cells/ grain structure development. Moreover, Electron Channelling Contrast (ECC) imaging and Electron Backscattered Diﬀraction (EBSD) techniques are eﬀectively used in conjunction with analytical investigation of solidiﬁcation process in order to link the microstructural development to the LPBF processing conditions employed. 3. Results Here, we present our microstructural analysis and observations of the evolution of cellular/dendritic network and grain structure/texture development across the LPBF part being built under the above described processing conditions. 3.1. Cellular/dendritic structure SEM examination of a Keller's solution etched cross-sections along the side and front view, Fig. 2(c–d), revealed a repetitive cusp-like pattern that has been observed before in LPBF processed Al alloys [15,18–20]. Unlike those developed in LPBF using a simple unidirectional scan strategy , the cusp-like pattern seems to be distributed rather randomly across the side and front views. Such a randomized pattern is the result of the 67° rotation in the scanning direction of each subsequent layer and the deep melt-pools that penetrate into the previous layers with a variable depth. The melt-pools during LPBF of aluminum alloys are known to be normally deeper than the layer thickness , while also being unsteady and sometimes varying up to four times the layer thickness . This is in accordance with the top view microstructure, shown in Fig. 2(a), where a variety of melt pool orientations (angled at ~67°, as annotated) appear in the image, indicating the interception of melt pools from multiple scanned layers in a single cross-section through the sample. Fig. 3(a) shows the cellular microstructure within a re-solidiﬁed melt-pool. Three diﬀerent zones are distinguishable across the meltpool: ﬁne, coarse and the heat aﬀected zone, HAZ, as shown in a higher magniﬁcation in Fig. 3(a1-a2-a3), respectively. It is noted that Keller's 2. Experimental Procedure A rectilinear sample (100 mm × 20 mm × 10 mm) was printed using Al10SiMg powder starting stock with a size range of 20–63 μm and an Mg content of approximately 0.3 wt%. Fig. 1(a) shows a secondary SEM surface shot of the powder stock with a cross-sectional view of the internal microstructure captured in the optical micrograph shown in Fig. 1(b). The high magniﬁcation secondary SEM image, shown in Fig. 1(c), and the corresponding inset (Fig. 1(ci)) reveal the cellular microstructure and the intercellular Si network including a lamellar eutectic structure as well as Si particles (as labeled in Fig. 1(ci)). The sample, shown in Fig. 1(d), was printed in an Argon atmosphere using a Renishaw AM400 operated with standard processing parameters provided by the manufacturer. The build plate temperature was 35 °C and a rod-style support structure of 3 mm was utilized underneath the sample. As demonstrated in Fig. 1(b), to enhance densiﬁcation in the part, the bidirectional scan vectors were rotated counter-clockwise by 67° upon completion of each layer. The hatch angle of 67° is adopted 30 Materials Characterization 145 (2018) 29–38 H. Qin et al. Fig. 1. (a) Secondary SEM surface shot of Al10SiMg powder starting stock, (b) optical micrograph and (c) high-magniﬁcation secondary SEM image of the crosssectional view of the internal microstructure with the corresponding inset shown in (ci); (d) the printed sample and schematic representation of scanning strategy; The bi-directional scan vectors in Layer n + 1 are rotated by 67° counter clockwise with respect to those at Layer n. solution aggressively etches the α-Al matrix and leaves behind a continuous network of Si-rich inter-cellular phase, similar to those revealed by Thijs et al. . This, however, limits the ability to resolve more detail about the nature of such a Si-rich network and its building blocks, including a eutectic α-Al/Si structure as well as Si particles. Therefore, other means of sample preparation and examination were subsequently employed. Fig. 3(b) is from a similar region as in Fig. 3(a) but was prepared by ion milling, which appears to be less preferential in its removal of the αAl matrix and overall less reactive to the surface, when compared to the eﬀect of the Keller's etch. At low magniﬁcation (Fig. 3(b)), the revealed microstructure depicts a cellular/dendritic structure similar to that revealed by the Keller's etch (Fig. 3(a)). However, at higher magniﬁcation, the ﬁne, coarse and HAZ regions (shown in Fig. 3(b1-b2-b3), respectively) exhibit more microstructural details than their Keller's etched counterparts, thus allowing more precise description of the inter-cellular microstructure within the melt pools (as will be described below). The morphology of the cellular/dendritic structure revealed, as can be seen in Fig. 3(a–b), varies from directional to non-directional inside and across the melt pools. These two regions are denoted as “D” (Directional) and “ND” (Non-directional) in Fig. 4(a), in which the same side-view microstructure as in Fig. 3(a) is shown at a higher magniﬁcation. Here, we explain that these two morphologies are, in reality, diﬀerent views of the same cellular structure. Solidiﬁcation during laser material treatment normally occurs under constraint conditions where the laser beam is scanned at high velocities with respect to the heat diﬀusion rate, α/rL, where α is the thermal diﬀusivity (K/ ρcp), rL the laser beam radius, K the thermal conductivity, ρ the density and cp the speciﬁc heat capacity [21,22]. Under such conditions, it is shown that the solidiﬁcation direction is controlled primarily by the heat ﬂux constraint and is usually along the normal to the solid/liquid interface, i.e. the melt-pool boundary. In this case, the local growth velocity, Vs, can be determined knowing the angle θ between the normal to the melt-pool boundary and the laser scan direction, as demonstrated in Fig. 4(b) [23,24]. For a cellular/dendritic structure with developed side branches, however, not only the heat ﬂux constraint but also the preferred crystallographic directions may aﬀect the resultant growth orientations [22,24]. The local growth velocity, Vs, is then related to the laser scan velocity, V, by: Vs = V cos θ cos φ (1) where φ is the angle between the normal to the melt pool boundary and the cell/dendrite axis. Due to the normally large imposed temperature gradients during laser processing of metals and particularly in LPBF Fig. 2. Secondary Electron SEM micrographs showing the microstructures revealed by Keller's etch for cross-sections at (a) top, (b) side and (c) front views. 31 Materials Characterization 145 (2018) 29–38 H. Qin et al. Fig. 3. Secondary Electron SEM micrographs showing the side-view microstructures revealed by (a) Keller's etch and (b) ion milling. High magniﬁcations of a1, a2, a3 and b1, b2, b3 areas referring to ﬁne, coarse and HAZ zones in (a) and (b), respectively; the term “MP” in (b) denotes Melt-Pool. (~106 Km−1 ), the cells/dendrites are often free of side branches thus growing primarily along the normal to the melt-pool boundary, i.e. φ = 0 [21,24]. Therefore, as Wu et al.  has also pointed out, the cells appearing non-directional in Fig. 4(a) are simply a cross-section of the directional cells within the microstructure which have grown relatively perpendicular to a melt pool boundary which is orientated on an angle to the side view cross-section. The rotation of scanning direction by 67° in each subsequent layer explains the relatively random location of melt pool boundaries with respect to the viewing crosssection. To further explain the variety of cells orientations in Fig. 4(a), the front view microstructure (shown in Fig. 4(b)) is annotated to depict three potential cell orientations with respect to a given side-view crosssection (the dotted line). Fig. 5(a & b) show more examples of regions containing a mixture of cross-sectional and longitudinal views of cells/ dendrites, with magniﬁed images shown in Fig. 5(c & d), respectively. The cellular morphology revealed by deep Keller's etching, as shown in Fig. 6, further reaﬃrms its viewing-angle dependency, i.e. the cells' directional appearance gradually fades as the specimen is tilted from −15° to +20°. Also, a 3D assembly of optical micrographs of microstructures revealed by Keller's etch is shown in Fig. 10(a) to help visualize the distribution of melt-pool boundaries as it relates to the utilized scanning strategy. At higher magniﬁcation, as shown in Fig. 5(c–d), the high clarity ion-milled surfaces reveal a great deal of microstructural detail. It can be seen that the intercellular regions are decorated with a discontinuous network of nano-sized Si-rich particles, as well as what appears to be AleSi eutectic lamella which form predominantly at occasional “triple points” (denoted by an arrow on Fig. 5(d)). The regions within the primary α-Al cells also contain a very ﬁne network of Si-rich particles, i.e. appearing as black dots across all cells in Fig. 5(c–d). It is then suggested that, upon the last stage of solidiﬁcation, the Si-rich liquid trapped in-between the cells solidiﬁes into an inter-cellular mixture of AleSi eutectic and a Si-rich α-Al phase. The latter phase, upon cooling from solidiﬁcation temperatures, further undergoes a precipitation of Si-rich particles. Similarly, the primary α-Al phase cells that are solidiﬁed under high cooling rates are supersaturated with Si and undergo a precipitation of ﬁne, intra-cellular, Si-rich particles upon post-solidiﬁcation cooling. It is noteworthy that the Si-rich network within the HAZ region (as shown in Fig. 3(b3)) has rather coarsened/ ripened due to the heat from an adjacent passing molten pool. It has been shown that short periods of time at high temperature can lead to Si-rich particle coarsening and a reduction in their number . Under these conditions, agglomeration and diﬀusion of Si is enhanced . To further conﬁrm the variety of Si-rich phases identiﬁed above, their distribution, crystallographic and morphological nature will be studied in detail in an upcoming Transmission Electron Microscopy (TEM) investigation. 3.2. Grain Structure In order to reveal and analyze the grain structure that contains the primary α-Al cells, ECC imaging and EBSD analysis were performed. This allows us to carry out an orientation comparison of the diﬀerent cellular regions in relation to the larger grain domains. Smooth polished surfaces created by ion-milling allowed ECC imaging at high resolution, contrast and clarity, as can be seen in Fig. 7. Particularly, the contrast from ECC was found to be suﬃcient to delineate both grain and cell boundaries, as demonstrated in the high magniﬁcation images of the grains shown in Fig. 7(b & d). Fig. 7(a & c) shows the grain structures revealed by ECC imaging of two diﬀerent 32 Materials Characterization 145 (2018) 29–38 H. Qin et al. 3D printed component. Highly smooth ion-milled surfaces allowed EBSD analysis to be performed with high indexing eﬃciency (> 95%) thus resulting in the production of exceptionally high resolution maps. Fig. 8(a–c) show the orientation maps obtained for top, side and frontview microstructures, respectively. The corresponding grain boundary maps, shown in Fig. 8(d–f), were obtained for a minimum of 15° misorientation. The high-resolution grain boundary maps reveal both the morphology of the Al grains and the melt-pool boundaries. Similarly to the grain structures revealed in the ECC images, one can observe both the cross-sectional and longitudinal views of the same columnar grains within the melt-pools. The examples of these two diﬀerent views of columnar grains are labeled with dotted lines within the front-view grain boundary map in Fig. 8(d). It can be seen that the grain structure coarsens towards the center of the melt-pools. Also, the epitaxial growth from the previous layers can be observed mainly in the regions closer to the melt-pools centerline. To analyze the texture development within the melt pools and across the LPBF part, as demonstrated in Fig. 9, we obtained (100), (110) and (111) pole ﬁgures from top-, side- and front-view microstructures as well as the inverse pole ﬁgures for various sample coordinates. As can be seen in the pole ﬁgures shown in Fig. 9, ⟨100⟩ is the predominant texture in the sample while there are also weak ⟨110⟩ and ⟨111⟩ texture components present in either top, side or front view microstructures. The inverse pole ﬁgures for all views demonstrate strong intensities for ⟨100⟩ texture along the build direction and to a lesser extent along the side- and front-view directions. Thus, the ⟨100⟩ texture has developed predominantly along the build direction. Weak ⟨110⟩ and ⟨111⟩ textures have developed primarily on a ~20–70° angle with respect to the build direction, as can be seen from their radial distribution in the top-view pole ﬁgures demonstrated in Fig. 9(a). These textures are distributed rather uniformly with respect to the build direction, which is due to the successive rotation of scanning direction in each subsequent layer, as opposed to the four-fold anisotropy that normally develops in bidirectional or island-based scanning strategies with 90° rotation (as demonstrated by Thijs et al. ). Thus, the scanning strategy employed in this study further reduces the overall texture in the LPBF parts by cutting back the anisotropy. A 3D assembly of the orientation and grain boundary maps from various orthogonal views is shown in Fig. 10(b–c) to assist with the visual perception of the overall texture within the LPBF part. Fig. 4. Secondary Electron SEM micrographs of (a) the side view and (b) the front view microstructures; On the front view image in (b), the potential variations in the growth direction of cells are demonstrated along an imaginary side-view cross-section (the dotted line), assuming a local scan vector, V. “D” and “ND” refer to Directional and Non-directional, respectively. regions within the side view cross-section. While the image in Fig. 7(c) is predominantly along a longitudinal view of columnar grains, the image in Fig. 7(a) shows a cross-sectional view of the same-type columnar grains. Under normal solidiﬁcation conditions (the range of Vs and G values) in LPBF and using the Hunt's criterion of columnar to equiaxed transition (CET) , as depicted for Al10SiMg alloy by Chou et al. , a fully columnar grain structure must develop throughout the melt pools. At higher magniﬁcation, looking at the cellular structure within these grains, one can observe that the grain and cell trajectories are largely parallel to each other; i.e., within the cross-sectional view of the grains (Fig. 7(b)) one can see a cross-sectional view of the cells and, similarly, for the longitudinal view of the grains (Fig. 7(d)), the cells appear with the same columnar morphology and orientation. Very ﬁne grains can be observed only along the melt-pool boundaries where a competitive growth mechanism has been active (as denoted by an arrow in Fig. 7(a)). As can be seen in Fig. 7(c), the columnar grains have grown predominantly perpendicular to the melt-pool boundaries and towards the center of the melt-pool. Also, as indicated by arrows in Fig. 7(c), a number of grains have continually grown through the meltpool boundary suggesting some degree of growth epitaxy. Although the ECC imaging of ion-milled surfaces was able to resolve (at varying resolution) the grain structure, EBSD analysis is required for delineation of the grains' orientation and the overall crystallographic and morphological texture within the melt-pools as well as across the 4. Discussion Below, we investigate the mechanisms upon which the size and morphology of the cellular network and grain structure develop during LPBF processing, as well as the evolution of the crystallographic texture in the part. Such mechanisms are discussed below as they relate to the solidiﬁcation conditions normally present during LPBF. 4.1. Cell Size Evolution It was pointed out in our observations in the previous section that the cells are coarser near the melt-pool boundaries and are increasingly reﬁned towards the center of the melt-pools, as can be seen in Fig. 3(b1) vs. Fig. 3(b2). The cell size evolution within the melt-pools can be explained by means of analytical formula that incorporate the eﬀect of local solidiﬁcation conditions, i.e. by accounting for variations of growth velocity, Vs, and temperature gradient, G, along the solid-liquid (S/L) interface (or the melt-pool boundary). A well-known model of such kind is the geometrical model of cellular growth presented by Hunt  which correlates the evolution of primary dendrite arm (or cell) spacing, λ1, with the local solidiﬁcation conditions along the advancing S/L interface as well as the alloy thermophysical properties: λ1 = [2.83(k∆T0 DΓ )0.25] Vs−0.25G−0.5 (2) The segment in the brackets represents the thermophysical 33 Materials Characterization 145 (2018) 29–38 H. Qin et al. Fig. 5. Secondary Electron SEM images of microstructures (a–b) at two diﬀerent areas within the side view cross-section revealed by ion-milling. Higher magniﬁcation images showing (c) the cross-sectional and (d) longitudinal view of the cellular/dendritic structure. Fig. 6. Cellular morphology revealed by deep Keller's etching of the front-view microstructure; The same area is imaged by varying the specimen tilt from (a) −15° to (b) 0° and then (c) +20°. the S/L interface favors an increase in cell spacing towards the top/ center of the melt-pool, while the increasing Vs tends to reﬁne it. However, since normal melt-pool sizes in LPBF are about an order of magnitude smaller than those in LMD, the variations in G within the melt-pools are rather negligible and often reported in the literature as a single value [5,8]. The cell spacing is thus primarily controlled by Vs variations along the S/L interface, meaning a gradual reﬁnement from the melt-pool boundary towards the top/center. This is consistent with the observations in this study, e.g. Fig. 3. properties of the alloy, where k is the partition coeﬃcient in the Al-rich portion of AleSi phase diagram, ΔT0 the freezing range, D the liquid diﬀusion coeﬃcient of the solute atom (Si in this study) and Γ the Gibbs–Thomson parameter. Furthermore, the above model has been quantitatively veriﬁed by Fallah et al.  against the experimental observations and phase-ﬁeld simulations of solidiﬁcation microstructure in Laser Metal Deposition (LMD) of TieNb alloys. According to Eq. (2), the cell spacing is inversely related to both growth velocity and temperature gradient along the S/L interface. As explained earlier in Section 3.1 and according to Eq. (1), the local growth velocity along the S/L interface for a cellular structure can be approximated as a function of the laser scan velocity and the morphology (curvature) of the melt-pool boundary, i.e. Vs = V cos θ. Therefore, considering the general morphology of the melt pools in LPBF (as demonstrated in Fig. 11), the growth velocity varies from nearly zero at the bottom of the melt-pools to a maximum of V (the laser scan velocity) at the top of the melt-pools. Using ﬁnite element simulations of heat transfer during LMD, Fallah et al.  have shown that the temperature gradient, G, is maximum at the bottom of the melt-pool and decreases along the S/L interface towards the top (see the schematic representation shown in Fig. 11). Thus, the reduction of G along 4.2. Crystallographic Texture The evolution of crystallographic texture is controlled by the following main factors: overall heat ﬂow direction within the part being built, growth direction and preferred growth orientation of the cells. While the ﬁrst two factors are primarily governed by the processing conditions, the third important contribution is the alloy characteristics. As discussed in Section 3.1, the growth direction along the S/L interface is imposed by the local orientation of the melt-pool boundary which itself is a function of the laser processing conditions . The overall heat ﬂow direction is mainly dictated by the build direction, which is 34 Materials Characterization 145 (2018) 29–38 H. Qin et al. Fig. 7. (a & c) ECC images of the side-view microstructure within the melt pools at two diﬀerent regions; Higher magniﬁcation images of (b) the cross-sectional and (d) the longitudinal view of the grains; “MPB” refers to Melt-Pool Boundary. Fig. 8. (a–c) Orientation maps and (d–f) the corresponding grain boundary maps for top, side and front view microstructures, respectively; The grain boundaries were identiﬁed with a minimum of 15° misorientation. occurs speciﬁcally when the heat ﬂow direction is aligned with the growth direction of the cells. Such a condition is more likely met in regions near the melt-pool centerline where the cells' growth direction is more closely aligned with the overall heat ﬂow direction. Under such due to the relatively small melt-pool size compared to the part size as well as the successive remelting of the previous layers during LPBF. In cellular solidiﬁcation of cubic materials, it is well-known that the cells preferentially grow along the ⟨100⟩ crystal direction . This 35 Materials Characterization 145 (2018) 29–38 H. Qin et al. Fig. 9. Pole ﬁgures and inverse pole ﬁgures obtained from (a) top-, (b) side- and (c) front-view microstructures. The (100), (110) and (111) pole ﬁgures (left) and inverse pole ﬁgures (right) along the build direction (Z,), side-view direction (Y) and front-view direction (X) are given. The relative intensity of the diﬀraction peaks is indicated by the color scale. Average Misorientation (KAM) maps shown in Fig. 12, the higher average misorientation values around the melt-pool centerlines suggest the presence of such a crystallographic texture. The exact microstructural nature of such high KAM values around the melt-pool centerline and their relationship with the crystallographic and morphological texture as well as with the grain size evolution within the part will be the subject of a future study. Weak ⟨110⟩ and ⟨111⟩ textures appear to be more randomized in the rotating scanning strategy (employed in this study) than what has been reported for the well-known bidirectional and island scan strategies with 90° rotation ; this refers to the circular distribution of such orientation intensities in the top-view pole ﬁgures (Fig. 9(a)) as opposed to a four-fold anisotropy of intensity distribution reported by Thijs et al. . Thus, the 67° rotating scanning strategy that is chosen to conditions, the growth near the melt-pool centerline is more likely to occur epitaxially, and the absence of competitive growth gives rise to the formation of larger columnar grains along the melt-pool centerline and closer to its center (as can be seen in Fig. 7(c)). In addition, the successive remelting of previous layers from various angles (due to the successive rotation of scanning direction) promotes directionality of heat ﬂow and thus the alignment of columnar grains of ⟨100⟩ orientation along the build direction. The conditions described above, together, give rise to the development of a moderately strong crystallographic texture along the build direction; i.e., columnar grains of predominantly ⟨100⟩ orientation. This is evident from the high intensities of (100) diﬀraction peaks in the inverse pole ﬁgures obtained along the build direction (for all three views), as can be seen in Fig. 9. Also, as demonstrated in the Kernel Fig. 10. 3D representation of microstructure in the LPBF part compiled from three orthogonal views manifesting (a) the melt-pool boundaries distribution revealed by Keller's Etch and optical microscopy, (b–c) the orientation and grain boundary maps, respectively, revealed by EBSD analysis of ion-milled surfaces. 36 Materials Characterization 145 (2018) 29–38 H. Qin et al. protrude/overlap one another leave behind a relatively randomized cusp-like pattern across the part being built. Under such conditions, we show that, in addition to an enhanced densiﬁcation, the overall crystallographic texture is reduced, as compared with the commonly used bidirectional or island-based scanning regimes with 90° rotation. Incorporating a unique combination of sample preparation and electron microscopy techniques along with an in-depth data analysis, the microstructural development is analyzed in detail as follows: - A fully columnar grain structure has developed across the melt pools possessing an orientation similar to that of the cells. - It appears that nano-sized Si-rich phases have solidiﬁed in three separate stages leading to formation of: (1) inter-cellular eutectic Si lamella, (2) an inter-cellular Si-rich phase precipitated within the inter-cellular α-Al phase upon cooling from solidiﬁcation and (3) an intra-cellular Si-rich phase precipitated within the α-Al primaries upon cooling to room temperature. - The cellular structure, developed under strictly directional solidiﬁcation conditions, is increasingly reﬁned as moving from the melt pool boundary towards the melt pool centerline. Such reﬁnement is attributed to the dominant eﬀect of growth velocity, Vs, in small melt pools such as in LPBF, that changes from nearly zero to a value close to the scanning speed near the melt pool center. - The grain size evolution follows an opposite pattern to the cells and increasingly coarsens moving away from the melt pool boundary towards its centerline, where the overall heat ﬂow direction (which is along the build direction) is more closely aligned with the growth direction (i.e. perpendicular to the melt pool boundary). Moreover, under such conditions, the grains near the melt pool centerline tend to grow epitaxially into large columnar grains creating a morphological texture along the build direction. - A moderately strong ⟨100⟩ crystallographic texture has developed along the build direction. This is due to the fact that, in regions near the melt pool centerline, the overall heat ﬂow direction is closely aligned with the growth direction as well as with the preferred growth orientation in Al, i.e. ⟨100⟩. - Weak ⟨110⟩ and ⟨111⟩ textures are present on a ~20–70° angle to the build direction while being distributed radially and fairly uniformly, as opposed to a four-fold anisotropic distribution that normally develops under commonly used bidirectional and island-based scanning regimes with 90° rotation. The scanning strategy utilized in this study further reduces the overall crystallographic texture in the Fig. 11. Schematic representation of variation of Vs and G along the melt-pool boundary. Tm denotes the alloy melting temperature; The isotherm Tm represents the estimated locus of the melt-pool boundary; The temperature gradient along the melt-pool boundary is then deﬁned as G = ΔT/dx, estimated by the locus of the two isotherms shown above, i.e. Tm and Tm + ΔT. enhance part densiﬁcation, also reduces the strength of the overall texture and promotes a more isotropic LPBF part beyond what could be achieved by 90° island-based scanning strategies. 4.3. Morphological Texture and Grain Size Evolution Larger grains form towards the melt-pool center (Fig. 8(e)) which is attributed to the successive remelting of previous layers as well as the alignment of heat removal direction with the growth direction [31,32]. Particularly, the latter enhances the alignment of cells growing near the melt-pool center thus minimizing the competitive growth and grain boundary formation. In addition to the above described ⟨100⟩ crystallographic texture, this also creates a morphological texture featuring the large columnar grains which extend mainly along the build direction. 5. Conclusions The detailed microstructural characterization and analysis presented in this study allows for an in-depth understanding of the unique cellular and grain structure development under a novel scanning strategy that aims at enhancing densiﬁcation during LPBF. Under a scanning regime that incorporates a counter-clockwise rotation of the scan vector by 67° upon each successive layer, the scan tracks that Fig. 12. Kernel Average Misorientation (KAM) maps obtained from (a) top-view, (b) side-view and (c) front-view microstructures. The high-angle grain boundaries are imposed on the KAM maps. 37 Materials Characterization 145 (2018) 29–38 H. Qin et al. LPBF part.  Acknowledgements This research was supported by the NSERC Discovery Grant program and the Queen's University Faculty of Engineering and Applied Science Dean's Research Fund.  References     I. Yadroitsev, I. Smurov, Selective laser melting technology: from the single laser melted track stability to 3D parts of complex shape, Phys. Procedia 5 ( (2010) 551–560.  H. Schleifenbaum, W. Meiners, K. Wissenbach, C. Hinke, Individualized production by means of high power Selective Laser Melting, CIRP J. Manuf. Sci. Technol. 2 (3) (2010) 161–169.  M. Gebler, A.J.M.S. Uiterkamp, C. Visser, A global sustainability perspective on 3D printing technologies, Energy Policy 74 (2014) 158–167.  D. Herzog, V. Seyda, E. Wycisk, C. Emmelmann, Additive manufacturing of metals, Acta Mater. 117 (2016) 371–392.  L. Thijs, K. Kempen, J.P. Kruth, J.V. Humbeeck, Fine-structured aluminium products with controllable texture by selective laser melting of pre-alloyed AlSi10Mg powder, Acta Mater. 61 (5) (2013) 1809–1819.  E. Brandl, U. Heckenberger, V. Holzinger, D. Buchbinder, Additive manufactured AlSi10Mg samples using Selective Laser Melting (SLM): microstructure, high cycle fatigue, and fracture behavior, Mater. Des. 34 (2012) 159–169.  N. Read, W. Wang, K. Essa, M.M. Attallah, Selective laser melting of AlSi10Mg alloy: process optimisation and mechanical properties development, Mater. Des. (1980–2015) 65 (2015) 417–424.  M. Tang, P.C. Pistorius, S. Narra, J.L. Beuth, Rapid solidiﬁcation: selective laser melting of AlSi10Mg, JOM 68 (3) (2016) 960–966.  V. Fallah, D.J. Lloyd, M. Gallerneault, Processing and characterization of continuous-cast AlMgSc (Zr) sheets for improved strength, Mater. Sci. Eng. A 698 (2017) 88–97.  A.B. Spierings, K. Dawson, T. Heeling, P.J. Uggowitzer, R. Schäublin, F. Palm, K. Wegener, Microstructural features of Sc-and Zr-modiﬁed Al-Mg alloys processed by selective laser melting, Mater. Des. 115 (2017) 52–63.  A.B. Spierings, K. Dawson, M. Voegtlin, F. Palm, P.J. Uggowitzer, Microstructure and mechanical properties of as-processed scandium-modiﬁed aluminium using selective laser melting, CIRP Ann. 65 (1) (2016) 213–216.  K. Kempen, L. Thijs, J.V. Humbeeck, J.P. Kruth, Processing AlSi10Mg by selective laser melting: parameter optimisation and material characterisation, Mater. Sci. Technol. 31 (8) (2015) 917–923.  J.C. Ion, Laser Processing of Engineering Materials: Principles, Procedure and Industrial Application, Butterworth-Heinemann, Oxford, 2005.  M. Cloots, P.J. Uggowitzer, K. Wegener, Investigations on the microstructure and              38 crack formation of IN738LC samples processed by selective laser melting using Gaussian and doughnut proﬁles, Mater. Des. 89 (2016) 770–784. R. Chou, A. Ghosh, S.C. Chou, M. Paliwal, M. Brochu, Microstructure and mechanical properties of Al10SiMg fabricated by pulsed laser powder bed fusion, Mater. Sci. Eng. A 689 (2017) 53–62. N.T. Aboulkhair, N.M. Everitt, I. Ashcroft, C. Tuck, Reducing porosity in AlSi10Mg parts processed by selective laser melting, Addit. Manuf. 1 (2014) 77–86. K. Guan, W. Zemin, G. Ming, L. Xiangyou, Z. Xiaoyan, Eﬀects of processing parameters on tensile properties of selective laser melted 304 stainless steel, Mater. Des. 50 (2013) 581–586. T.B. Sercombe, X. Li, Selective laser melting of aluminium and aluminium metal matrix composites, Mater. Technol. 31 (2) (2016) 77–85. K.G. Prashanth, S. Scudino, H.J. Klauss, Kumar Babu Surreddi, L. Löber, Z. Wang, A.K. Chaubey, U. Kühn, J. Eckert, Microstructure and mechanical properties of Al–12Si produced by selective laser melting: eﬀect of heat treatment, Mater. Sci. Eng. A 590 (2014) 153–160. C. Weingarten, D. Buchbinder, N. Pirch, W. Meiners, K. Wissenbach, R. Poprawe, Formation and reduction of hydrogen porosity during selective laser melting of AlSi10Mg, J. Mater. Process. Technol. 221 (2015) 112–120. W. Kurz, R. Trivedi, Microstructure and phase selection in laser treatment of materials, J. Eng. Mater. Technol. 114 (4) (1992) 450–458. M. Zimmermann, M. Carrard, W. Kurz, Rapid solidiﬁcation of Al-Cu eutectic alloy by laser remelting, Acta Metall. 37 (12) (1989) 3305–3313. V. Fallah, M. Alimardani, S.F. Corbin, A. Khajepour, Temporal development of meltpool morphology and clad geometry in laser powder deposition, Comput. Mater. Sci. 50 (7) (2011) 2124–2134. V. Fallah, M. Amoorezaei, N. Provatas, S.F. Corbin, A. Khajepour, Phase-ﬁeld simulation of solidiﬁcation morphology in laser powder deposition of Ti–Nb alloys, Acta Mater. 60 (4) (2012) 1633–1646. J. Wu, X.Q. Wang, W. Wang, M.M. Attallah, M.H. Loretto, Microstructure and strength of selectively laser melted AlSi10Mg, Acta Mater. 117 (2016) 311–320. C. Yan, L. Hao, A. Hussein, P. Young, J. Huang, W. Zhu, Microstructure and mechanical properties of aluminium alloy cellular lattice structures manufactured by direct metal laser sintering, Mater. Sci. Eng. A 628 (2015) 238–246. Y. Birol, Microstructural evolution during annealing of a rapidly solidiﬁed Al–12Si alloy, J. Alloys Compd. 439 (1–2) (2007) 81–86. J.D. Hunt, Steady state columnar and equiaxed growth of dendrites and eutectic, Mater. Sci. Eng. 65 (1) (1984) 75–83. J.D. Hunt, Cellular and primary dendrite spacings, Solidiﬁcation and Casting of Metals, the Metals Society, London, 1979, pp. 3–9. M.E. Glicksman, Principles of Solidiﬁcation, Springer, New York, 2011. X.P. Li, X.J. Wang, M. Saunders, A. Suvorova, L.C. Zhang, Y.J. Liu, M.H. Fang, Z.H. Huang, T.B. Sercombe, A selective laser melting and solution heat treatment reﬁned Al–12Si alloy with a controllable ultraﬁne eutectic microstructure and 25% tensile ductility, Acta Mater. 95 (2015) 74–82. M.M. Kirka, P. Nandwana, Y. Lee, R.R. Dehoﬀ, Solidiﬁcation and solid-state transformation sciences in metals additive manufacturing, Scr. Mater. 135 (2017) 130–134.