The role of overlap region width in multi-laser powder bed fusion of Hastelloy X superalloy

ABSTRACT
 During the multi-laser powder bed fusion (ML-PBF) process, the quality of the overlap region is essential to ensure the integrity between different laser scanning areas. Here, six different overlap region widths (ORWs) are selected to systematically investigate the effect of ORWs on densification, metallurgical defects, crystalline microstructure, and mechanical performance of the Hastelloy X samples processed by multi-laser powder bed fusion. ORWs are set to 0, 50, 100, 150, 200, and 250 μm, respectively. With an increment in ORW, the porosity gradually increases from 0.01% to 7.08%. Inversely, microcracks are reduced with increasing the ORWs due to the smaller width-to-depth ratio of the melt pools and the increased degree of recrystallisation. All the samples display equivalent ultimate tensile strength of ∼880 MPa, while the elongations at break of the overlap samples are 3.5%−10.5% lower than that of the single-laser processed one (28.4%). The inferior ductility is ascribed to the decreased densification in the overlap regions, of which the adverse effect can be minimised when an ORW of 150 μm is utilised. This work is anticipated to provide efficient reference and theoretical guidance for the large-size nickel-based superalloy components fabricated by ML-PBF.


Introduction
Nickel-based superalloys have been widely utilised as the preferred materials for crucial hot end components in the aerospace fields because of their excellent properties such as high strength, good plastic toughness, superior thermal stability, and desirable oxidation resistance at high temperatures (Tan et al. 2020;Thakur and Gangopadhyay 2016;Zhang, Chen, and Hu 2018).The solid solution nickel-based superalloy Hastelloy X is an important structural material in the aerospace industry, which is used to manufacture aero-engine combustion chambers, afterburner components, rectifiers, and structural covers with high operating temperatures as well as complex structures (Xia et al. 2019).The demand for more complex and lighter aerospace components is gradually increasing with increased service temperatures and pressures, which presents a more serious challenge to the conventional manufacturing processes (DebRoy et al. 2018).
Laser powder bed fusion (L-PBF) has been adopted to fabricate nickel-based superalloy components with complex structures.As a promising metal processing technique in the additive manufacturing (AM) family, L-PBF is well suitable for the formation of complex components by selectively melting powder-form raw materials using a bottom-up stacking manner with the assistance of laser beams (Cai et al. 2020;Joshi and Sheikh 2015;Patterson et al. 2021;Wang et al. 2022).It is also known as direct metal laser melting (DMLM by GE Additive) and selective laser melting (SLM by SLM Solutions Group AG) commercially (Sing et al. 2021).It has the ability to simplify manufacturing patterns, reduce production costs, and shorten process flow, which makes L-PBF become an optimal alternative for manufacturing metal components with geometrical complexity in the aerospace domain.However, it is of great challenge to fabricate large-size components through the conventional L-PBF technique, limited by the scanning range and manufacturing efficiency with single laser system (Gusarov et al. 2018).
The multi-laser powder bed fusion (ML-PBF) technique is able to solve the above-mentioned challenges by dividing the forming regions into multiple individual scanning fields with multiple laser beam sources (Hinke 2017).At present, the commercial ML-PBF equipment has been extensively used to meet the urgent demand of large-size parts fabrication in aerospace fields.For example, GE Additive X line 2000R, Bright Laser Technologies BLT-S600, and Techgine Laser TS-500 have been employed with the aim of improving the maximum building size by increasing the number of lasers.Simultaneously, the maximum productivity can be enhanced with the increased laser beams working in parallel (e.g. the nominal maximum build rates are 1000 cm 3 h −1 for NXG XII 600 and 150 cm 3 h −1 for Renishaw RenAM 500Q) (Yin et al. 2021).Nevertheless, the overlap region, which is printed by the multiple laser beams, is essential to ensure the integrity between different laser scanning regions.
Owing to the re-melting of the overlap regions by multiple laser scanning, it is challenging to guarantee the homogeneity of densification, microstructure, and properties between single-laser processed regions (base regions) and multi-laser processed regions (overlap regions).Defects such as surface quality degradation (Li, Wang, and Zeng 2017), internal gas voids, and lack-of-fusion holes (Zhang et al. 2019) are prone to appear in the overlap regions, which deteriorates the mechanical properties of the entire samples.For instance, during the ML-PBF process of AlSi10Mg, there were a few small pores along the overlap boundaries and the elongation at the overlap regions was 17.4% lower than that of the base regions (Taheri Andani et al. 2017).Li et al (Li, Yang, and Wang 2021) fabricated Ti-6Al-4V samples using single, dual, and quadruple L-PBF systems, where the relative densities of overlap regions gradually deteriorated from 99% to 96.3% with an increase in the re-melting numbers.Wei et al (Wei et al. 2021) demonstrated that keyhole defects were apt to form in overlap regions when the multi-laser encountered together to trigger the interaction of laser plume.The width of overlap regions not only plays a vital role in multi-beam stitching but also is exceedingly crucial for the consistency and uniformity of forming quality for the printed components.This is suggested that further research on the role of overlap region width is required to reduce the defects and modify the inferior microstructure caused by the laser re-melting process to acquire desirable mechanical properties.However, few studies have been reported on the role of overlap region width in the literature, and it is of great significance to clarify the relationship between defect, microstructure, and mechanical properties of the samples with different overlap region widths.
In this study, a dual-laser PBF platform was applied to fabricate Hastelloy X bulk samples with different overlap region widths (ORWs).The differences in porosity, densification, microcracks, melt pool morphology, element segregation, and grains in the printed Hastelloy X samples with different ORWs were unveiled in depth.The hardness and tensile properties of the printed samples with different ORWs were investigated.The purpose of this study is expected to significantly reduce the adverse effect of the re-melting process during ML-PBF and obtain simultaneously desirable tensile strength and ductility by finding a befitting ORW to guarantee the metallurgical quality and performance consistency of the whole component.

Raw material
Gas-atomised Hastelloy X nickel-based superalloy powder particles supplied from the Beijing Institute of Aeronautical Materials (China) were nearly spherical in shape.Their diameter distribution determined by a Mastersizer 3000 laser size analyzer was in a range of 10-60 µm with an average size of 35.4 µm (Figure 1(a)).The chemical composition (wt.%) of the Hastelloy X powder was Ni-21.4Cr-1.6Co-0.7W-10.5Mo-0.3Cu-17.8Fe-0.09C.Before the ML-PBF experiments, the powder was dried at 90°C for 10 h to fully remove moisture and improve the fluidity.

ML-PBF process
A FastForm FF-M500 additive manufacturing apparatus with two pairs of laser systems was used as the ML-PBF platform.Each laser system is equipped with a continuous wave IPG YLR-500 fibre laser (focused spot size of ∼100 μm, wave-length of 1080 nm, maximum scanning velocity of 5000 mm/s).The two corresponding scanning fields (Area #1, #2) are arranged to realise a maximum scan size of 490 × 290 × 390 mm 3 (Figure 1 (b)).
Five groups of samples (D1, D2, D3, D4, and D5) with ORWs of 50, 100, 150, 200, and 250 μm were prepared by the dual-laser system, respectively (Figure 1(b)).Each group contains a 10 mm × 10 mm × 10 mm cube for metallurgical microstructure observation, and a 55 mm × 10 mm × 10 mm cuboid bulk for mechanical property evaluation.A blank group (S) with the same geometric characteristics was built within Area #1 using single-laser.Based upon a series of preliminary experiments, the optimised process parameters for all the samples were set as laser power of 240 W, laser scanning velocity of 600 mm/s, hatch spacing of 100 μm, and layer thickness of 40 μm.A progressive scanning strategy with a rotation angle of 67°was utilised during the experiments (Figure 1(b)).The building spaces were divided into two individual scanning fields and the laser beams scanned the overlap regions successively, where laser #1 scans the powder layer firstly, and then, laser #2 scans the areas again.

Microstructures and properties testing
The Image J software was used for the quantification statistic of the porosity.To calculate the porosity in each optical microscope (OM) image, five random regions with an area of 1 × 1 mm 2 were selected taken from the side surfaces of the polished metallurgical samples with different ORWs.Metallurgical microstructure characterisation of the as-fabricated samples was conducted with a DM750M optical microscope, a Nova Nano-450 scanning electron microscopes (SEM), an FEI Quanta 650 FEG integrated with an EDAX-TSL electron backscatter diffraction system (EBSD) and an EPMA-8050G electron probe microanalysis (EPMA).The overlap regions of the test pieces were highlighted by a marker to ensure that the overlap regions (including the overlap line) are in the centre when being tested for OM, EBSD, EPMA, and SEM.The metallurgical samples were ground, polished, and electrolytically etched via an A2 reagent (10 vol% perchlorate and 90 vol% ethanol) on a Lechtropol-5 machine.The voltage and time for the electropolishing processes were set as 20 V and 10 s, respectively.The hardness was tested by using a 430SVD Vickers hardness tester with a load of 0.5 Kg for 15 s.The residual stress measurements were performed using an x´pert3 powder XRD machine with a point focus Cu X-ray tube.The tensile tests were measured by an AG-IC 100 KN machine with a strain rate of 1.33 × 10 −3 s −1 .All tensile samples were extracted from the 55 × 10 × 10 mm 3 rectangle samples built along the horizontal plane and machined according to ISO 6892-1:2009 standard (Figure 1(c)).For samples from groups D1 to D5, the centre lines of the corresponding tensile specimens were controlled to be coincided with the overlap line.For reliability, at least three duplicates were tested for each group.

Changes in defect characteristics
Figure 2 shows the SEM images of the upper surface of the unpolished samples with different ORWs.There are deep ravines between the melt channels and many round unmelted powder particles can be observed on the surface.The base regions in the multi-laser processed samples are almost the same as the single-laser processed sample (Figure 2(b−f)), where there are also few manufacturing defects except the unmelted powder particles.For the overlap regions, the number of the unmelted powder particles is decreased because these particles are melted during the re-melting process to make the surface smoother (Leung et al. 2019).All the overlap region boundaries are located on the right side and show an arc-like shape.This is because laser #1 scans the powder layer firstly, and then, laser #2 scans the areas again.
According to the OM images (Figure 3), few manufacturing defects are detected in the single-laser processed sample with the porosity of only 0.01% under the optimised processing condition.For the overlap regions, a large quantity of pores with the diameter of 5-50 μm can be observed.The porosity of groups D1−5 is 0.09 ± 0.02%, 1.21 ± 0.21%, 3.67 ± 1.07%, 7.08 ± 0.38%, and 6.69 ± 0.86%, respectively.This is similar to the previous description that re-melting in the overlap regions generally deteriorates the densification of printed parts (Li, Yang, and Wang 2021;Masoomi, Thompson, andShamsaei 2017a, 2017b).
The pores in the single-laser processed sample are below 10 μm in size and belong to gas voids (Figure 3 (a)).As the ORWs increase to 100 and 150 μm for groups D2 and D3, the gas pores are retained but the major defects turn into shrinkage pores with larger sizes of 30−60 μm (Figure 3(c−d)).When the ORWs further increase into 200 μm and 250 μm, the defects are mainly composed of lack-of-fusion holes, accounting for a dramatic increase in porosity (Figure 3(e−f)).These lack-of-fusion holes appear in a form of weak bonding and they are irregularly shaped with lengths larger than 100 μm.
Figure 4 shows that the microcracks of the as-fabricated Hastelloy X samples along the build direction have a width from ∼5 μm (group S) to ∼3 μm (group D1), ∼2 μm (group D2), and ∼1 μm (groups D4 and D5) with an identical length of ∼100 μm, but the width of the microcracks decreases with the ORWs.Most microcracks are preferentially propagated along the building direction and located at grain boundaries, which is consistent with the single L-PBF fabrication of Hastelloy X (Han et al. 2018(Han et al. , 2020; Montero-Sistiaga show typical dendritic structures, this is a sign of the presence of residual interdendritic liquid when the microcrack is formed.This demonstrates a lack of liquid metal feeding in the last stages of solidification and causes poor cohesion between dendrites.This kind of microcracks can be defined as solidification microcracks (DuPont, Lippold, and Kiser 2009).

Changes in microstructures
The overlapped semi-elliptical melt pool characteristics can be observed in all samples, and it is easy to distinguish the overlap interface from the melt pools (yellow dot lines in Figure 4).As for the overlap regions, the melt pool width is reduced and the length is larger than that of the single-laser processed sample (Figure 5(a−f)).It can be understood in Figure 5(g) that a higher overlap will produce smaller melt pool width (W MP ) and greater melt pool depth (D MP ), thus leading to a smaller width-to-depth ratio (τ = W MP /D MP ).Measures of τ and the area ratio of columnar structure were taken on ten melt pools in the centre of all the samples, and the values of τ of each sample from groups S and D1−5 are 3.82 ± 0.52, 3.74 ± 0.74, 3.36 ± 0.15, 2.42 ± 0.33, 2.88 ± 0.24, and 2.23 ± 0.68, respectively.It can be deduced that the overlap regions experience a more intense thermal history during the L-PBF process compared to the base regions, which is probably due to the remelting process.Meanwhile, typic fine columnar and cellular sub-grains are discernible within the melt pools (Figure 5(a) and (f)).These sub-grains are typical solidification microstructures in L-PBF fabricated nickel-based superalloys due to the large temperature gradients and high cooling rates during L-PBF (Bouabbou and Vaudreuil 2022;Li et al. 2015;Teng et al. 2021).Figure 5(h) shows the statistical histogram of columnar structures for the six samples.Our measurements reveal the presence of a smaller percentage of columnar structures in the samples with different ORWs compared with that of the base regions.The as-built samples contain a fraction of columnar structures with values of 81.06 ± 2.31% (S), 79.33 ± 3.41% (D1), 68.31 ± 3.22% (D2), 64.32 ± 1.21% (D3), 54.62 ± 3.54% (D4), and 58.41 ± 5.66% (D5), respectively.
Cr-rich and Mo-rich carbides are detected along the microcrack in the single-laser processed sample (Figure 6 (a)).This finding confirms that element segregation occurs at the grain boundaries during the rapid solidification process, which has also been reported in our previous study (Sun et al. 2021).In contrast, Figure 6(b) implies that carbide segregation only occurs around the pores rather than the microcracks in the overlap regions.
Figure 7 presents the crystallographic orientations on the side surface taken by EBSD for all the samples.Epitaxial grains are found to be nearly parallel to the build direction due to the grain growth along the positive temperature gradient and the higher thermal gradient along the build direction in the melt pool (Guo et al. 2017).A comparison of these EBSD images shows the differences in grain morphology and grain size of the base regions and the overlap regions (marked in black lines).It can be observed that the overlap regions exhibit a larger melt pool depth and a more significant percentage of coarse grains compared to those of the base regions.As the ORWs increase, the grains become coarser and the average grain size increases from 9.55 μm (S) to 13.57 μm (D1), 11.90 μm (D2), 20.46 μm (D3), 19.55 μm (D4), and 22.46 μm (D5).This is similar to the morphology evolution characteristics of the melt pools observed above.Additionally, the grains that formed in the single-laser processed sample exhibit random crystallographic orientations, while the overlap regions display a strong <0 0 1> orientation, which previous studies have found to be the preferential grain-growth orientation for FCC structure materials (Cloots, Uggowitzer, and Wegener 2016;Guo et al. 2017;Popovich et al. 2017).The reason for the different crystallographic orientations is probably related to the larger proportion of cellular grains generated in the overlap regions.
In order to facilitate the analysis of misorientation angle distribution, low angle grain boundaries (LAGBs, 2°−15°)   in, the fraction of HAGBs gradually decreases and LAGBs become predominant, which accounts for an even higher proportion.This is in agreement with the reports of Song et al (Song et al. 2022) and Chen et al (Chen et al. 2018), that is, the greater reheating and cooling cycles during re-melting led to a slightly higher content of LAGBs.The grain orientation distribution map of D3 in Figure 9(d) indicates that the volume fraction of HAGBs is 35.1%, which is the minimum in the six samples.

Changes in mechanical properties
The side-surface Vickers hardness profiles of the asfabricated Hastelloy X samples (Figure 10) show that the average hardness of the overlap regions (251.4 ± 6.4 Hv for group D1, 248.6 ± 4.5 Hv for group D2, 235.4 ± 10.6 Hv for group D3, 240.7 ± 6.1 for group D4, and 233.1 ± 9.4 Hv for group D5) is about 8.4% lower than that of the single-laser processed sample (264.1 ± 5.4 Hv).Obviously, the better hardness feature should be ascribed to the highly densified microstructure and excessive residual stress.The measurement of the residual stress through the X-ray diffraction method reveals that the residual stress in the as-fabricated samples is reduced when multilaser processed, where the base regions and overlap regions are calculated to be ∼256 MPa and ∼236 MPa.Note that the presence of compressive residual stress will reduce the plastic deformation and therefore increase hardness (Fitzpatrick et al. 2002;Liu et al. 2012;Song and Yu 2002).
Tensile properties of the as-fabricated Hastelloy X samples are presented in Figure 11.It can be seen that the ultimate tensile strength (UTS) of the single-laser processed sample is as high as 880.5 ± 10.9 MPa, which is much higher than that of the wrought counterparts (Ni et al. 2019).It can be attributed to grain refinement (average size of 9.55 μm) in the as-fabricated material.Besides, the high residual stress within the as-fabricated Hastelloy X can also promote the UTS (Lewandowski and Seifi 2016).From the inset of Figure 11(a), the fracture occurs in the middle of the samples, which is the   overlap region.Since there is no significant difference between the microstructural morphologies of the various as-fabricated samples, the weakened overlap regions should be mainly attributed to the decrease in densification.
As for the multi-laser processed samples with different ORWs, the UTS is measured to be 887.1 ± 11.8 MPa (D1), 884.6 ± 3.1 MPa (D2), 889.4 ± 9.4 MPa (D3), 881.6 ± 11.6 MPa (D4), and 882.1 ± 20.5 MPa (D5), respectively (Figure 11(b)), indicating that the tensile strength of the overlap regions can be very close to that of the base regions.The porosity measurements reveal that the pore volume may not have an obviously significant effect on tensile properties despite the difference in densification.However, a 3.5 −10.5% reduction of elongation at break (ε ab ) of the multi-laser processed samples is detected in contrast with the single-laser processed sample (28.4 ± 0.3%), of which the ε ab of the sample with ORW of 150 μm can reach 24.9 ± 1.8%.
There are a few manufacturing defects on the fracture surfaces of both the single-laser processed sample and the multi-laser processed samples (Figure 12(a−f)).The insets of Figure 12(a) and (d) show extensive dimples with diameters of a few microns, which indicates a ductile fracture mode and agrees with the high ε ab values.Simultaneously, this indicates that the fracture mechanism of the asfabricated samples is a ductile transgranular fracture.On the contrary, a lot of near spherical pore defects, including gas voids, shrinkage pores, and lack-offusion holes, can be observed on the fracture surface of the samples from groups D1 to D5 (Figure 12(b−f)).The remarkable discrepancy in macroscopic fracture morphology matches well with the variation trends of both the densification degree (Figure 3) and the tensile strength (Figure 11) of the as-fabricated samples with different ORWs.With the ORW raises, the size and number of the pores become more extensive and destroy the mechanical properties in the overlap regions.

Discussions
Based on the above results, the melt pool and defect formation mechanism of the as-fabricated Hastelloy X alloy with different ORWs can be illustrated in Figure 13(a−f), respectively.The overlap regions undergo an additional laser melting and experience the recrystallisation process when compared to the base regions.Consequently, with the increase of ORWs, the broad and shallow melt pools (group S) turn into narrower and deeper ones (groups D1−5) with smaller widthto-depth ratio.Additionally, when the ORWs increase, most of the columnar structures turn into cellular structures.Microcrack-dominated defects are converted into microcrack-and-pore-containing defects with the incremental porosity from 0.01% to 7.08% when increasing the ORWs.Additionally, the width of the microcracks is then slightly reduced from 5 μm to 1 μm on account of the alleviated element

Role of ORWs in defects
There are two main factors that are responsible for the generation of solidification microcracks.Firstly, due to the rapid cooling rates in L-PBF, a complete homogenisation of the liquid alloy within the critical temperature range is not possible.As a result, low melting films that are caused by the constitutional cooling effect are rich in Cr and Mo, which could cover the emerging grains at the end of solidification due to their significant segregation coefficients.Thus, it will lead to brittle grain boundaries and cause a separation of two adjacent grains under residual tensile stress.
Secondly, it has been reported that HAGBs have an inherently higher grain boundary energy than LAGBs, which makes HAGBs easier to be affected by solidification microcracks according to the work of Chen et al.
(2016) and Wang et al. (2004): where g gb is the grain boundary energy, G is the shear modulus, b is the Burgers vector of the edge dislocations, v is the Poisson ratio, and u m is the angle at which g gb reaches its maximum value.Equation ( 1) is only valid for small angle boundaries, beyond u m , the equation cannot apply and the grain boundary energy g gb is assumed to be constant at its maximum value (Wang et al. 2004).It indicates that the grain boundary energy of the HAGB would be relatively higher than that of the low angle one.Therefore, the liquid phase at higher angle grain boundaries will exist at a relatively lower temperature according to the dendrite coalescence theory (Rappaz, Jacot, and Boettinger 2003): where DT b is a critical undercooling at the coalescence of the adjacent grains, DS f is the entropy of fusion per unit volume, and d is the thickness of the diffuse interface (∼1 nm).For HAGBs, the grain boundary energy is likely to exhibit high values, therefore g gb − 2g sl is expected to be >0.As a result, DT b is significant and consequently the liquid film remains stable at lower temperatures and delays the coalescence of dendrite secondary arms.Simultaneously, a lower temperature at which the liquid film exists will lead to a higher solidification cracking sensitivity.Therefore, most microcracks tend to occur at the HAGBs rather than at the LAGBs.
When operating the re-melting process during ML-PBF, the newly deposited layer will be completely heated to a single-liquid phase once the temperature exceeds its liquidus temperature, whereafter, the remelted layer forms after solidification.Meanwhile, the previously deposited material in the underlying layer will be heated to a temperature between solidus and liquidus simultaneously, thus, partially fusion and re-solidification will occur in the underlying layers.During the re-solidification process, a mild tilt will occur in the interior of grains, leading to the development of LAGBs.Since the high lattice distortion and solubility of the supersaturated solid-solution phase will significantly accumulate the contraction stress during the process of rapid solidification, dislocations will gradually pile up and are mainly distributed at the interface of polygonal grains and interdendritic region (Shakerin et al. 2019).Consequently, according to the Equations ( 1) and ( 2), the sensitivity to microcracks is reduced by the decrease of HAGBs.
In addition, in the as-fabricated Hastelloy X samples, microcracks may form when the accumulated thermal residual stresses tear the liquid films caused by segregated carbides at the grain boundary.But in the overlap regions, the re-melting of the as-fabricated Hastelloy X has the ability to re-dissolve those carbides into the surrounding matrix, which will reduce the lowmelting-point eutectic liquid films to a certain extent and finally decrease the sensitivity to microcracks.
Inversely, the pore defects in the as-fabricated Hastelloy X samples deteriorate with increasing the ORWs, which experiences a successive evolution: tiny gas voids (ORW of 0 μm) → large gas voids (ORW of 50 μm) → shrinkage pores and gas holes (ORW of 100 and 150 μm) → lack-of-fusion holes, shrinkage pores, and gas voids (ORW of 200 and 250 μm).The high solidification rate during the L-PBF process traps the gas bubbles before it rises and escapes, resulting in defect inclusions in the shape of gas voids (Drissi-Daoudi et al. 2022).These are usually small spherical pores that are driven by Marangoni flow and fail to escape from the melt pools (Yu et al. 2022).For group D1 (Figure 3(b)), sufficiently rapid cooling causes the generation of metallic vapours inside the material and makes the gas voids coarser (more than 10 μm).When ORW increases, the formation of shrinkage pores is associated with the more severe Marangoni convection caused by the higher laser energy, which then results in an incomplete flow of metal (Goel et al. 2019;Wu et al. 2020;Zafer et al. 2020).With an excess of ORW, insufficient melting for a given volume of the solidified materials will occur due to a large laser beam processed area of the underlying layer (Figure 14(a)), and finally resulting in the formation of irregular lack-of fusion holes with sharp edges (Kumar et al. 2019;Tang, Pistorius, and Beuth 2017).Moreover, as illustrated in Figure 5(h), the width-todepth ratio of melt pools in the overlap regions decreases with increasing the ORWs, which results in less overlap between the neighbourhood melt pools than that in the base regions, and thus, there exists pore/unmelted region between melt pools, as shown in Figure 14(b), resulting in lack-of-fusion holes (Koh et al. 2022).

Role of ORWs in microstructural evolution
As illustrated in Figure 14(a), the laser beam acts on the powder layer with lower thermal conductivity in the base regions, while the laser beam melts the previously printed solid parts in the overlap regions.Hence, the cooling rate in the overlap regions is higher than that in the base regions, and thus, the melt pool lifetime is shortened, which decreases the laser energy absorption and results in more profound and narrower melt pools with a coarsening trend of grain size (Yao et al. 2022).Generally, columnar grains stem from the first few layers.As the L-PBF process continues, columnar grains start to grow epitaxially along the preferred orientation of the materials.As the defects in the overlap regions are more serious, the columnar grains are blocked by the pore-and-microcrack defects in the solidification front, resulting in the formation of cellular grains (Niu et al. 2021).Moreover, the solidification of the melt pools in the overlap regions proceeds by columnar growth on the underlying layer, and the underlying layer also recrystallises.Thus, the solidification of each layer in the overlap regions is dominated by short columnar and equiaxed growth on the previously solidified layer, which blocks the development of columnar grains.When the ORW increases, the recrystallisation effect of re-melting enhances and most of the columnar structures turn into cellular structures.
With the increase in ORW, the grains are gradually enlarged from 9.55 μm to 22.46 μm.The grain size and morphology strongly depend on the temperature gradient and the cooling rate during the L-PBF process.The laser re-melting process in the overlap regions leads to the extension of the cooling time, and thus sufficient time for grain growth, leading to an increase in grain formation.Simultaneously, those defects between grains promote the recrystallisation of grains and enhance the grain sizes.The heat-affected zones introduced by re-melting can also result in coarsening of the grains and grain recrystallisation in the solidified regions (Griffiths et al. 2018).This could be the dominant factor that determines the coarsening microstructure in the overlap regions.
From Figure 7, it is seen that the overlap regions exhibit a relatively stronger <0 0 1> orientation compared with the base regions, which is the direction of the fastest crystallisation and in line with the temperature gradient orientation (Cloots, Uggowitzer, and Wegener 2016).Since the overlap regions have received successive laser energy inputs and have experienced more complex thermal cycles during the building process, the corresponding thermal stress values may be higher than that of the base regions.According to the study of Antonysamy et al (Antonysamy, Meyer, and Prangnell 2013), the 'variant selection' phenomenon, which usually happens in the case of high thermal stress, will occur during the solidification and cooling process, thus promoting the growth of grains along the preferential grain-growth orientation.Additionally, the reason for the different crystallographic orientations is also related to the larger proportion of cellular grains generated in the overlap regions.
Compared with the base regions, the fraction of HAGBs in the overlap regions decreases with increasing the ORWs (from 63.2% to 35.1%).This large fraction of LAGBs could firstly be attributed to the formation of cellular microstructures induced by the rapid cooling rate during the re-melting process in the overlap regions.Generally, the evolution of the misorientation angle follows the solidification characteristics of materials (Shakerin et al. 2019).It is well known that the distinctive process of adding material track by track and layer by layer together with the fast and directional cooling rates during the L-PBF process creates a unique microstructure in the products.During the ultra-rapid melting and solidification process, a large number of sub-grains nucleate and grow in the microstructure, and these sub-grains boundaries mainly appear as LAGBs (Wen et al. 2019).Due to the high-energy laser beam, when making a new layer, the previous layer(s) that have been solidified will be heated and partially (or completed) remelted again.This phenomenon, which is approximated to the annealing treatment, leads to transformation from LAGBs into HAGBs (Li et al. 2016).As illustrated in Figure 14(b), since the thermal conductivity of the powder bed is lower than that of the previously printed solid parts, the heat cannot be quickly dissipated in the base regions, resulting in heat accumulation.But in the overlap regions, the decreased heat accumulation shortens the time of the self-annealing effect, which in turn leads to the decrease of HAGBs.In addition, the volume fraction of HAGBs decreases with increasing the ORWs can also be attributed to the following reason.During the laser re-melting process, the previously printed solid part is re-melted and induces pore-and-microcrack defects, under which non-equilibrium condition dendrite growth is deviated (Newell et al. 2005).In the process of branching growth, the dendrite experiences plastic deformation in the mushy zone, which leads to the deviation of dendrite orientation.When the dendrite converges again during the re-solidification process in the overlap regions, the LAGBs will be generated, thus leading to the reduction of HAGBs.

Role of ORWs in mechanical properties
The influence of ORWs on the hardness and residual stress of the overlap samples has been analyzed in section 3.3, where they are about 10% lower than that of the single-laser processed sample.The hardness feature can be ascribed to the highly densified microstructure and excessive residual stress.The significant difference worth mentioning is the grain size observed in Figure 7. Based on image measurements, with increasing the ORWs, the average grain size increases from 9.55 μm to 13.57, 11.90, 20.46, 19.55, and 22.46 μm.This implies that the laser re-melting results in coarser grains, which leads to the reduction in hardness observed in Figure 10.Generally, the formation of residual stress is highly related to the temperature gradient during solidification and a higher temperature gradient leads to higher residual stress.As the thermal conductive in the overlap regions is higher than that of the base regions, the reason for the reduced residual stress obtained in the overlap region can be attributed to the lower temperature gradient within the melt pool compared to that of the base region.Moreover, the preheat effect resulted from the first laser scan per layer can significantly heat the processing specimen, which also decreases the temperature gradient during the re-melting process (Zhang, Gu, and Dai 2022).
The effect of ORWs on mechanical properties can be analyzed based on three aspects: metallurgical defects, microstructures, and misorientation angle distribution.While there is little difference in tensile strength between those samples with different ORWs, the reasons for the decreased elongation of the overlap samples are analyzed as follows.Firstly, the higher porosity of 1.21%−7.08%plays a vital role in the reduction of elongation.These pores produced by the re-melting process during ML-PBF are considered as an extremely adverse factor for tensile performance as they act as stress concentration sites and accelerate microcrack propagation under tensile loads, causing premature fractures and failures, especially the lack-of-fusion holes.With the increase of ORWs, the pore defects in the overlap samples translate from gas voids into shrinkage pores and lack-of-fusion holes, which seriously affects the elongation.The inference can be confirmed by the fracture surface of the as-fabricated samples (Figure 12(b−f)), where a large number of pores (gas voids, shrinkage pores, and lack-of-fusion holes) and microcracks are observed.
Secondly, samples with columnar grains have lower strength but higher plasticity, while those with cellular grains have higher strength and lower plasticity (Kuo et al. 2020).Most of the columnar structures transform into cellular structures by the recrystallisation effect in the overlap regions, consequently, the overlap samples observe the comparable tensile strength while there are more pore defects.It can also be a reason for the drop in ε ab of the samples with different ORWs.Thirdly, with increasing the ORWs, the fraction of HAGBs in Figure 9 gradually decreases from 63.2% (S) to 57.4% (D1), 41.3% (D2), 35.1% (D3), 36.7% (D4), and 36.4% (D5), respectively.It is noted that the local stress in grain boundary becomes more intense with the increase of grain boundary misorientation angle.With the increase of misorientation angle, the continuity of crystal deformation on both sides of grain boundary is weakened, which decreases the coordination ability of grain boundary.When the tensile stress is applied, the strain is concentrated at grain boundary, resulting in the fracture of the as-built samples.Consequently, the decreased HAGBs volume fraction of the overlap samples enhances the grain boundary coordination capacity and bonding strength, providing a benefit to improve the mechanical properties.
Based on the characterisations and analyses in the present study, the overlap sample (D3) with ORW of 150 μm observes relatively optimal mechanical properties.For one reason, the width of the microcracks decreases with the ORWs, it is harder for the microcracks to propagate under tensile loads in the samples from group D3 than that of groups D1 and 2. Simultaneously, there are only gas voids and shrinkage pores (without lack-of-fusion holes) in the sample from group D3, and the porosity is 3.67%, which is significantly lower than the porosity of 6.69% in the sample from group D5.For another, the fraction of HAGBs in the sample from group D3 is minimum, accounting for only 35.1%, which is approximately half of the sample from group D1 (63.2%).

Conclusions
In this study, a dual-laser PBF platform is utilised to process Hastelloy X alloy with different ORWs.The defect, microstructure, and mechanical properties of the overlap regions are studied and compared to those of the base regions.The primary conclusions are summarised as follows: . With more laser beams being involved during the ML-PBF process, the common defect characteristic of the overlap regions is gas voids, shrinkage pores, and lack-of-fusion holes, which are induced by the remelting process.Nevertheless, the size of microcracks in the overlap regions is smaller than that of the base regions.The re-melting of laser beam causes a smaller width-to-depth ratio and more cellular structures in the overlap regions, which is responsible for the microcrack suppression. .All the overlap regions are composed of fine columnar/cellular sub-grains, which is essentially the same as that of the base regions.However, the grains that form in the base regions exhibit random crystallographic orientations, while the overlap regions display a strong <0 0 1> orientation.And the decreased heat accumulation in the overlap regions leads to the transformation of HAGBs to LAGBs, which reduces the sensitivity to microcracks.Moreover, the alleviation of element segregation is also one of the primary contributors to the reduction of microcracks. .Samples in the overlap regions experience a deteriorated hardness and ductility than that of the base regions due to the reduction in densification degree caused by the pore defects during the re-melting process.The tensile strength of the overlap regions can be comparable to those of the base regions (∼880 MPa), while the elongation at break of the multi-laser processed samples is 3.5−10.5%lower than that of the single-laser processed one (28.4%),where the ε ab of the sample with ORW of 150 μm could reach 24.9%.And the average hardness of the overlap regions (233.1 Hv) is 10.1% lower than that of the base regions (259.2Hv). .Based on the above results, it can be concluded that to improve the performance consistence of large-size Hastelloy X components fabricated by ML-PBF, it is advised to conduct the ORW of 15 μm.

Figure 1 .
Figure 1.Experiment details of the ML-PBF process: (a) SEM micrograph and particle size distribution of the as-received Hastelloy X powder particles, (b) layout of the samples with different ORWs on the platform, (c) laser scanning strategy and two-dimensional drawings of tensile test samples.

Figure 2 .
Figure 2. SEM images of the as-fabricated L-PBF samples from groups S and D1−5.

Figure 3 .
Figure 3. OM images and porosity of the as-fabricated L-PBF samples from groups S and D1−5.

Figure 4 .
Figure 4. SEM images and microcracks of the as-fabricated L-PBF samples from groups S and D1−5.

Figure 5 .
Figure 5. High-magnified view of microstructures and schematic illustration of the melt pools in different ORWs: (a−f) groups S and D1−5, (g) schematic diagram of the definition of the melt pool width-to-depth ratio, (h) statistical histogram of columnar structure.

Figure 6 .
Figure 6.EPMA element mapping profiles of microcracks and pores on the side surface of the as-fabricated samples: (a) group S and (b) group D5.

Figure 7 .
Figure 7. Crystallographic orientations on the side surface of the as-fabricated L-PBF samples from groups S and D1−5.

Figure 8 .
Figure 8. EBSD image quality maps on the side surface of the as-fabricated L-PBF samples from groups S and D1−5.

Figure 9 .
Figure 9. Misorientation angle distribution of the as-fabricated L-PBF samples from groups S and D1−5.

Figure 10 .
Figure 10.Hardness of the as-fabricated L-PBF samples from groups S and D1−5.

Figure 11 .
Figure 11.Tensile performance of the as-fabricated L-PBF samples from groups S and D1−5: (a) stress-strain curves, (b) summary of the YS, UTS, and elongation at break.
segregation and the transformation from HAGBs to LAGBs.

Figure 12 .
Figure 12.Typical fracture surfaces of the as-fabricated L-PBF samples from groups S and D1−5.

Figure 13 .
Figure 13.Schematic illustration of the differences between the as-fabricated samples.

Figure 14 .
Figure 14.Schematic illustration of (a) the laser beam processed region, and (b) the melt pool size and spacing in the base region and overlap region.
This work is financially supported by the Project funded by the National Natural Science Foundation of China [grant number 52201040], the China Postdoctoral Science Foundation [grant number 2021M701291], the Academic Frontier Youth Team Project of Huazhong University of Science and Technology [grant number 2017QYTD06], the National Natural Science Foundation of China [grant number 52275333], and the Stabilization Support Project of AVIC Manufacturing Technology Institute [grant number KZ571801].