Tailored microstructure in laser-based powder bed fusion of IN718 through novel beam shaping technology

Unlike conventional manufacturing methods, Laser-based Powder Bed Fusion of Metal (PBF-LB/M) alloys entails highly localized heating and rapid cooling during the melting and solidification processes. In this additive manufacturing (AM) technique, a high-energy laser beam scans a powder bed along a 2D path generated from a sliced CAD model, selectively melting the powder to construct a volumetric part layer by layer.¹ Using a Gaussian intensity distribution, solidification occurs directionally perpendicular to the melt pool boundaries. The thermal gradient (G) and solidification growth rate (R) in PBF-LB/M do not favor the nucleation and growth of equiaxed grains, which would lead to isotropic mechanical properties.² Therefore, a typical microstructure of metal alloys like Inconel718 (IN718) is a dendritic columnar structure with coarse grains elongated along the building direction (BD), persisting through several layers due to epitaxial grain growth.³ While coarse, elongated grains with a preferred orientation along the BD are ideal for high-temperature applications requiring creep resistance due to reduced grain boundary percentage, fine equiaxed grains are generally preferred for applications demanding high ductility, as they also offer better tensile strength and good fatigue properties. Developing solutions to achieve customized localized microstructures along with increasing productivity is an ongoing challenge in PBF-LB/M, addressed by various research activities and technological advancements such as laser beam shaping.^4,5

Using alternative laser spatial distributions has shown more stability in melt pool dynamics by producing less spatter⁶ and creating a wider process window.⁷ Recent studies demonstrate that changing the laser beam profile from Gaussian to shapes like ring-shaped,⁸ core-ring,⁹ top-hat,¹⁰ and elliptical (parallel or perpendicular to the scanning direction)^4,5 can significantly affect microstructural properties such as grain size and orientation in both single tracks and volumetric parts.

Pérez-Ruiz et al.¹¹ studied the effect of different beam shapes, such as Gaussian, ring-shaped, and top-hat, with various beam diameters ranging from 50 to 150 μm. Considering that the beam intensity varied across different beam shapes, they concluded that each beam shape develops a unique microstructural characteristic in IN718. This is attributed to the impact of process parameters, including the laser beam profile, on melt pool morphology and the derived correlation between texture index and the radius of curvature of beam shapes. Additionally, a shallow, wide melt pool morphology promotes epitaxial grain growth along the BD, while a nail-shaped melt pool favors equiaxed grain nucleation and growth, leading to grain refinement and subsequently improved tensile properties. Moore et al.¹² developed a coupled thermal transport-Monte Carlo model to predict the evolution of temperature fields and grain microstructures during PBF-LB/M using Gaussian, ring-shaped, and Bessel beam profiles. The simulation results indicate that the ring-shaped beam generates lower temperatures than the Gaussian beam. Additionally, the Bessel beam produces a smaller melt pool, leading to smaller and more equiaxed grains compared to those formed with the Gaussian and ring beams. However, in this study, the beam intensity varied between the different beam shapes due to the change in the diameter of the studied beam shapes, which affects the conclusions. Roehling et al.⁴ investigated the effect of elliptical laser beam profiles and their thermal distribution on grains morphology. They found that, compared to the Gaussian shape, volumetric parts fabricated with elliptical profiles exhibited finer grains and a higher area fraction occupied by equiaxed small grains. Shi et al.¹³ investigated the transition from columnar to equiaxed microstructures during single-track PBF-LB/M of 316L using Gaussian and elliptical laser beam profiles. They examined the effects of different scanning directions with a rotationally asymmetrical elliptical beam shape on laser-powder interaction time, nucleation rate, and resulting microstructure. Their findings indicate that a wider beam, created by an elliptical transverse shape, promotes equiaxed grain structures by increasing the nucleation rate. A cellular automaton simulation was used to model the nucleation rate under the applied process parameters. Loh et al.¹⁴ explored the use of a uniform top-hat laser beam to alter melt pool morphology. Their research demonstrated the feasibility of combining two beam shapes, Gaussian, and top-hat, to enhance manufacturing efficiency. Cloots et al.¹⁵ compared the microstructure of PBF-LB/M parts produced using Gaussian and ring-shaped beam profiles. Their results indicate that ring-shaped beam profiles create smaller and shallower melt pools, which can help prevent hot cracking in nickel-based alloys. However, it is important to note that in this study, the laser diameter and consequently the laser intensity varied for the different beam shapes. As a conclusion of these research activities, increasing building rate can be realized by increasing the beam diameter and unlocking faster scanning speed with non-Gaussian beam profiles. However, this will lead to flat-like shallow melt pool leading to high anisotropic properties by promoting elongated grains with a high texture index.

In this study, we present the results of a systematic layerwise investigation on the effect of laser beam profiles on microstructure anisotropy of IN718 material. We start at a μm level with hatch packages on single and five-layer samples, then extends to cubic parts and tensile bars. All laser beam profiles were generated with a spatial diameter of 400 μm to enable larger hatch distances and achieve a build rate exceeding 16 mm³/s. The new rotational asymmetrical, simulation-based beam profiles, named as chair shape, demonstrated the potential by maintaining the same build rate and process stability while reducing the texture index from 4.6 to 1.4 approaching isotropic properties. Finally, we tested the tensile properties.

This study employs MetcoAdd 718C as a working material from Oerlikon Metco AG with the chemical composition of Ni (53–55 wt. %), Fe (18 wt. %), Cr (18 wt. %), Nb + Ta (5 wt. %), Mo (3 wt. %), Ti (1 wt. %), and <0.5 wt. % of other elements and particle size ranging from 18 to 46 μm.¹⁶ 

To manufacture hatch packages and volumetric cubic parts, an EOS M290 machine equipped with a beam shaping module, a 2 kW laser source with a wavelength of 1050 nm is used. As illustrated schematically in Fig. 1, the laser beam emitted from the source has a planar phase and Gaussian amplitude distribution. Upon entering the beam shaping device, a beam splitter within the module divides the beam into two linearly polarized partial beams. Each partial beam is directed toward its respective beam shaping elements, based on liquid crystal on silicon (LCOS1 and LCOS2). The LCOS elements can alter the wavefront of the incident beams by locally modulating the phase and/or amplitude using an imported computer-generated holography (CGH) mask. The CGH mask in the LCOS acts as a phase modulator, manipulating the phase of light waves to create specific interference patterns and beam shapes. The resultant beam shapes from each LCOS are combined in the beam combiner component to form the final beam shape on the target plane. The laser beam with the defined shape is then transferred to a scanner, which moves along the exposure path, while a focusing device ensures the focal plane aligns precisely with the building platform.²² 

Five laser beam profiles named as ring-shaped, core-ring, line-shaped beams parallel (Line//SD) and perpendicular (Line⊥SD) to scanning direction (SD), and chair-shaped beam profile are used for this study. The width and length of line-shaped beam profiles are 90 × 400 μm. Prior to the experiments, beam profiles were measured by using caustic measurements with a CINOGY Focus Beam Profiler 2KF. If necessary, the measured beam shapes were adjusted to the desired target shapes by modifying the corresponding Zernike coefficients. The beam diameter of shapes is determined with the second momentum method. All shown caustic profiles in Fig. 2 are measured at a laser power of 100 W. There is an unknown uncertainty regarding beam deviation at higher laser power due to rising temperature in optical components. The chair-shaped laser beam profile is calculated inversely by Holla et al.¹⁷ developed a method to determine the beam shape necessary to achieve a desired melt pool geometry, a rectangular weld seam, with a uniform temperature distribution, suitable for heat conduction welding.

The process parameters used to manufacture one- and five-layer hatch packages, as well as cubes for different beam shapes and tensile bars, are listed in Table I. All parts were scanned bidirectionally with a layerwise rotation angle of 67° (shown in Fig. 3). The listed process parameters in Table I are selected after extensive process parameter study.

Figure 4 illustrates the measurement of melt pool morphology in a single layer experiment. The depth is defined as the vertical distance from the top surface down to the deepest point of the melt pool. The width is defined as twice the half-width, where the half-width is the horizontal distance from one corner of the melt pool to its deepest point.

The results of Archimedes’ relative density test for the manufactured cubes are shown in Fig. 5 with relative density above 95%. The reference density used to calculate relative densities of cubes is 8.25 gr/cm³.

All hatch packages and cubes were analyzed using an OLYMPUS optical microscope after being cross-sectionally cut, ground, and polished. To reveal the grain structure, the samples were etched using a Royal Water etchant solution (a 3:1 ratio of HCL to HNO3). For quantitative analysis of grain orientation and grain sizes, the electron backscatter diffraction (EBSD) technique was applied. The EBSD analysis was conducted with a field of view set to 2500, resulting in a measurement area of 2.5 × 1.87 mm². All EBSD data are analyzed and postprocessed using the MTEX MATLAB library. For grain size estimation, all EBSD measurements were conducted with a step size of 2 μm. Data cleaning utilized the Neighbor CI Correlation (NCIC) clean-up procedure, with the “minimum confidence index” parameter set to 0.3. To plot the stereographic pole figures, a generalized spherical harmonic expansion smoothing was applied, using a Gaussian half-width of 5°.¹⁸ 

To test the mechanical properties, tensile testing was conducted on samples oriented vertically, horizontally, and diagonally at a 45° angle (as demonstrated in Fig. 6) according to the standard ISO 6892-1:2017. To ensure statistically validated results, each tensile test was repeated three times.

To gain understanding of the influence of laser beam shape on microstructure through different layers, a one-layer and a five-layer hatch package were built, cross-sectionally cut, and etched. The temperature gradient (G) and solidification rate (R) are crucial thermal factors influencing grain growth in the melt pool. The direction of grain growth is primarily governed by the thermal gradient (G), while the grain size is controlled by the solidification rate (R). By combining G and R, a solidification map can be constructed using G * R and G/R (Fig. 7). The slope (G/R) on the graph determines the grain morphology of the solidified structure, while the product G * R controls the grain size.¹⁹ The laser beam profile affects both the thermal gradient (G) and solidification rate (R). By altering the spatial distribution of laser intensity can change the melt pool morphology from nail-shaped to flat. It has been demonstrated that the melt pool morphology significantly impacts grain orientation.^4,11,13

In principle, grains grow in the opposite direction of heat flow (or heat dissipation). Therefore, within a melt pool, grains grow perpendicular to the fusion boundary. As shown in Fig. 8, grain growth normal to the melt boundary occurs within a melt pool or a single hatch layer. However, in five-layer hatch packages or cubes, other process parameters, such as laser beam shape and scanning strategy, can alter the heat dissipation direction and cooling rate, resulting in different microstructures. This result demonstrates why conclusions about microstructure formation cannot be based on single track or even single layer experiments. Figure 8 shows the single layer and five-layer experiments for different laser beam profiles. Significant differences between the shapes are evident in the five layers. After three layers, with a theoretical layer thickness = 240 μm, the promotion of columnar grains by using ring-shaped beam profile is apparent. In contrast, the line-shaped and chair-shaped beam profile show that the epitaxial growth of grains is mostly prevented by the subsequent layers within these five layers. The thermal gradient combined with cooling rates for the chair-shaped and line laser beam shapes promotes more equiaxed grains than columnar ones. Figure 9 illustrates the etched image of the grains in cubic parts for five different shapes. The epitaxial growth of the grains for the 400 μm ring-shaped and core-ring beam profiles is clearly visible, leading to an anisotropic microstructure with elongated large grains along the BD. In contrast, the Line//SD and chair-shaped profiles significantly deviate from epitaxial growth, as seen in Fig. 8 from the five-layer experiment.

As demonstrated in previous research, melt pool morphology plays a crucial role in determining grain orientation and elongation.¹¹ Wide, shallow melt pools, such as those produced by the ring-shaped laser beam profile, tend to form columnar grains. In contrast, nail-shaped melt pools result in smaller, more equiaxed grains. This variation in melt pool morphology is attributed to the spatial distribution of laser intensity (laser beam shape), while other process parameters, such as volumetric energy density (VED), scanning speed, hatch distance, and scanning strategy, are kept constant. The quantitative analysis of melt pool morphology, including melt pool depth and width, is shown in Fig. 10. The measurements represent the average of 20 melt pools over a single layer. The results indicate that both the chair-shaped and the Line//SD profile exhibit greater depth (around 50 μm) and width (maximum 100 μm) compared to the other shapes. Even though Line//SD and Line⊥SD expose different laser lengths to SD, a smaller melt pool width for Line//SD than Line⊥SD was expected. Possible reasons for observing similar widths (∼450 μm) could be beam profile deviation at higher laser power or longer laser-powder interaction time for Line//SD which widen melt pool. However, as demonstrated by Shi et al.,¹³ a wider melt pool leads to more nucleation events, while a shallower melt pool promotes the formation of more columnar grains. Higher depth of Line//SD and chair-shaped increased the melt pool curvature reducing grain size and J-Index.¹¹ 

In the equation $d A W ¯$ which represents the average area-weighted grain diameter in micrometers (μm), N_i denotes the total number of grains, d_i is the circular equivalent grain diameter for the ith grain, S_i is the area of the ith grain, and S_total is the total measurement area.

As illustrated in Fig. 12, both chair-shaped and Line//SD laser beam shapes are drastically reduced the grain sizes, which means these intensity distributions are triggering more grains nucleation than ring-shaped beam. Chair-shaped laser beam profiles are designed to produce an asymmetrical intensity distribution, resulting in a wide and shallow melt pool. In addition to the desired melt pool morphology, achieving a homogeneous temperature distribution across the melt pool was also considered. These simulation results have been evaluated experimentally.¹⁷ However, it is important to note that the process parameters used in the experimental evaluation differ from those used in this study. Figure 10 shows that chair-shaped-generated melt pools have the greatest width, which, according to Shi et al.,¹³ leads to more nucleation events. Despite this, the chair-shaped-generated melt pools exhibit a nail-like shape with a relatively large depth. This morphology hinders the formation of columnar grains from the bottom to the top of the melt pools. The flat shape morphology is inhibited, and grains tend to grow normal to the boundary, resulting in varying orientations due to the increased melt pool curvature.

Elliptical laser intensities are studied and compared with circular (Gaussian) laser beam shapes by Roehling et al.^4,5 Roehling et al. demonstrated that longitudinal (LE, parallel to SD) and transversal (TE, perpendicular to SD) elliptical laser beam profile, which is quite like Line//SD and Line⊥SD, have higher ratio of equiaxed to columnar grains.⁵ Using simulation in this, it has been shown that both TE and LE show higher melt pool flow velocities compared to Gaussian profile which leads to fragmentation of dendrite tips and redistribution, allowing them to act as intrinsic nucleation sites ahead of the growth front in the melt zone.⁵ Roehling et al. continued study from single tracks to volumetric parts and change in the microstructure using TE and Gaussian intensity distribution. The increased number of equiaxed grains due to reduction in thermal gradient, supported by simulation, is shown for 316L material. However, in that study, the other process parameters were not kept constant.⁴ In our research with both line shapes along and perpendicular to SD, we could observe a completely different microstructure. The measured width of both line shapes is similar as it is shown in Fig. 10 but Line//SD has deeper depth meaning more curvature on the melt pool morphology.

Grain size, orientation, and morphology have a significant impact on the tensile properties and anisotropy of the as-built parts. To evaluate this effect, tensile properties were assessed in different orientations: vertical, horizontal, and diagonal (45° to BD) as shown in Fig. 6. According to Hosseini and Popovich,²⁰ the ductility and strength of as-built IN718 parts are lower than those of heat-treated samples due to the absence of γ′ and γ″ phases, which contribute to precipitation hardening. Table II summarizes the state of the art of as-built and heat-treated mechanical properties of IN718 manufactured by EOS M290.¹⁶ According to Fig. 13, the tensile properties of samples manufactured using beam shaping are superior to those reported for as-built properties with standard process parameters of EOS. The most significant differences between various beam shapes are observed in the vertically built samples. These vertical samples exhibit the lowest elastic modulus, tensile yield strength (YS), and ultimate tensile strength (UTS), while showing the highest elongation compared to specimens built in other directions. Ghorbanpour et al. have investigated the anisotropy in the tensile response of additively manufactured IN718.¹⁸ Their observation is mainly indicating higher ductility, lower yield/ultimate strength and elastic modules for the samples loaded parallel to BD (vertical samples). The main reason for this difference is the strong texture along ⟨100⟩ and the columnar morphology of the grains. A pronounced ⟨100⟩ texture leads to a lower elastic modulus, as the ⟨100⟩ direction has the lowest elastic modulus.²⁰ 

In Fig. 13, among the vertically built samples, those with a ring-shaped beam profile exhibit the highest texture index (4.6) along the ⟨100⟩ direction, indicating a greater number of grains with fiber texture oriented in this direction. This results in the lowest Young’s modulus, tensile strength, and UTS. Conversely, the Line//SD beam profile, with the lowest texture index (1.37), leads to a higher Young’s modulus.

Additionally, according to the Hall–Petch equation and grain boundary strengthening mechanisms, smaller grains result in higher yield strength. However, no significant difference in tensile strength was observed in vertically built samples with different beam shapes, and only slight differences were noted in other directions. According to the measured grain size (Fig. 12), the difference in ultimate tensile strength can be attributed to the higher work hardening rate in finer-grain microstructures. The effect of grain boundary strengthening with smaller grains is noticeable in the Line//SD and chair-shaped vertical samples. Taylor factor mappings based on the preferential slip system of {111} ⟨110⟩ for Inconel 718 by Deng et al.²³ indicate that the Taylor factor is lower for specimens loaded parallel to the building direction (i.e., vertically built specimens) compared to those loaded perpendicular to the building direction (i.e., horizontally built specimens). This difference accounts for the lower strength observed in the vertically built samples. The variation in ductility can be attributed to distinct cracking mechanisms that occur when tensile loads are applied parallel versus perpendicular to the building direction. Columnar grain boundaries serve as preferred routes for accumulating damage, ultimately leading to material failure.²¹ The orientation of the tensile load relative to these grain boundaries significantly affects how cracks propagate. Specifically, when the tensile load is applied perpendicular to the columnar grain boundaries in transverse samples, it triggers Mode I fracture. This fracture mode facilitates crack propagation by promoting crack opening along the grain boundaries, which consequently reduces the material’s ductility. This explains the slightly lower UTS for horizontally built samples with a ring-shaped beam profile compared to those with Line//SD and chair-shaped profiles, considering the grain morphology and size as shown in Figs. 11 and 12.

This study focuses on systematically understanding the layer-by-layer microstructure evolution during PBF-LB/M of IN718 and examining the effects of newly calculated beam shapes, such inverse calculated shape (chair shape) and Line//SD, on microstructural features in hatch packages (one and five layers) and volumetric parts. This analysis extends to tensile mechanical properties and sample anisotropy. The results demonstrate that using high-productivity process parameters—including a large beam diameter of 400 μm, high layer thickness, high laser power, and fast scanning speed—the beam shapes can significantly alter the melt pool morphology by changing the spatial laser intensity distribution. This, in turn, affects the thermal gradient and the number of nucleation events during solidification, reducing the texture index from 4.6 to 1.36 and the grain diameter from 521 to 155 μm. These changes lead to substantial improvements in tensile properties, such as elastic modulus, yield strength, UTS, and elongation. Additionally, the anisotropy percentage decreases, indicating more homogeneous properties in different directions.

The authors thank the European Health and DigitaL Executive Agency (HADEA) for funding the project InShaPe—Green Additive Manufacturing through Innovative Beam Shaping and Process Monitoring, under Project No. 101058523.

The authors have no conflicts to disclose.

Narges Mirzabeigi: Conceptualization (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Validation (equal); Visualization (equal); Writing – original draft (equal). Peter Holfelder-Schwalme: Conceptualization (equal); Supervision (equal); Writing – review & editing (equal). Yu He: Data curation (equal); Investigation (equal); Methodology (equal); Visualization (equal). Katrin Wudy: Conceptualization (equal); Funding acquisition (equal); Project administration (equal); Supervision (equal); Writing – review & editing (equal).