Influence of heat treatment on microstructure, mechanical and corrosion behavior of WE43 alloy fabricated by laser-beam powder bed fusion

Magnesium (Mg) alloys are considered to be a new generation of revolutionary medical metals. Laser-beam powder bed fusion (PBF-LB) is suitable for fabricating metal implants with personalized and complicated structures. However, the as-built part usually exhibits undesirable microstructure and unsatisfactory performance. In this work, WE43 parts were firstly fabricated by PBF-LB and then subjected to heat treatment. Although a high densification rate of 99.91% was achieved using suitable processes, the as-built parts exhibited anisotropic and layered microstructure with heterogeneously precipitated Nd-rich intermetallic. After heat treatment, fine and nano-scaled Mg24Y5 particles were precipitated. Meanwhile, the α-Mg grains underwent recrystallization and turned coarsened slightly, which effectively weakened the texture intensity and reduced the anisotropy. As a consequence, the yield strength and ultimate tensile strength were significantly improved to (250.2  ± 3.5)  MPa and (312  ± 3.7)  MPa, respectively, while the elongation was still maintained at a high level of 15.2%. Furthermore, the homogenized microstructure reduced the tendency of localized corrosion and favored the development of uniform passivation film. Thus, the degradation rate of WE43 parts was decreased by an order of magnitude. Besides, in-vitro cell experiments proved their favorable biocompatibility.

Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Introduction
Magnesium (Mg) alloys are considered to be a new generation of revolutionary medical metals and demonstrate great potential for absorbable bone implant application [1,2].They can be degraded in the form of self-corrosion, and the produced Mg ions are beneficial for promoting bone tissue growth [3].Meanwhile, their elastic modulus (∼45 GPa) is close to that of human bone tissue (10 ∼ 30 GPa), thus avoiding the stress shielding effect [4].Among various Mg alloys, rare earth (RE) Mg alloy, like WE43 alloy (Mg-4Y-3(Nd,Gd)-0.5Zr,wt.%), has drawn extensive attention owing to its comprehensive mechanical properties and corrosion behavior [5].It is reported that the solubility of RE in the Mg matrix is relatively high, which enables remarkable solid solution strengthening [6].RE can also form stable intermetallic with Mg, further achieving precipitation strengthening [7].On the other hand, RE could induce the formation of a dense product film, thereby improving the anti-corrosion ability of the Mg matrix [8,9].There are already numerous researches that have evaluated the feasibility of WE43 for potential bone repair applications and proved its favorable biocompatibility and mechanical performance [10].
Despite the aforementioned advantages of WE43, the fabrication of WE43 devices with complicated structures, particularly porous scaffolds, remains a huge challenge.The primary obstacle is that Mg alloys have poor room-temperature deformability and traditional processing techniques like casting or extrusion are difficult to produce customized shapes or internal pores that are required for implants [11,12].As one typical additive manufacturing technology, laser-beam powder bed fusion (PBF-LB) utilizes a high-energy laser beam as a heat source to achieve the complete fusion of metal powder layer-by-layer for rapid three-dimensional prototyping [13][14][15].Thus, it is recognized to be suitable for the highprecision and high-efficiency manufacturing of personalized implants [16][17][18][19].Currently, PBF-LB has been successfully applied in several medical fields like tissue engineering scaffolds, oral implants, and vascular stents [20,21].A fact is that some researchers have already fabricated WE43 implants using PBF-LB technology, and studied the microstructure, mechanical, and corrosion behavior, aiming to explore their potential tissue repair applications [5,22,23].
Although PBF-LB provides the possibility of fabricating complex geometries based on a bottom-up construction approach, the performance of PBF-LB processed parts is usually unsatisfactory in an as-built state.It is well accepted that the as-built part is constructed by layer-bylayer and every layer undergoes a fast cooling, thus usually showing unfavorable microstructure features, like chemical inhomogeneity, presence of undesirable precipitates, forming defects like pores, or deformation caused by residual stress [24,25].Thus, the as-built parts usually demand posttreatment like heat treatment, hot isostatic pressure, and surface modification [26,27].Among them, heat treatment is a cost-effective process to modify microstructure heterogeneity and eliminate defects that are intrinsically induced by rapid solidification or layer-wise building process [28,29].Qi et al [30] investigated the influences of heat treatment on the microstructure of Al-based alloy prepared using PBF-LB and found that high-density needle-like ultrafine Al 2 CuMg precipitates presented after artificial aging, which significantly improved the yield strength.Kong et al [31] revealed that the heat treatment promoted the homogeneous distribution of alloying elements in PBF-LB processed 316 L stainless steel, which led to the formation of an inert and compact passive film, and consequently enhanced the anti-corrosion ability.Accordingly, homogenization treatment is used to dissolve unwanted phases, while aging treatment is usually followed for the final strength enhancement by precipitation strengthening [24].Nonetheless, there are few studies concerning the post-heat treatment strategy and corresponding influence mechanism on the properties of as-built WE43 parts.
In this work, WE43 parts were firstly prepared using PBF-LB, and then subjected to different heat treatment methods.The major efforts were devoted to elucidating the microstructure feature including grain morphology, texture and precipitates.Particularly, the influence mechanism of heat treatment on microstructure evolution was clarified in-depth.The relationship between microstructure characteristics, mechanical properties and corrosion behavior was systematically studied.

Original materials
The feedstock used in the present study was gas-atomized WE43 powder (Tangshan Weihao Magnesium Powder Co., Ltd.).It could be observed that the majority of powder had a spherical shape (figure 1

PBF-LB process
A commercial PBF-LB system (BLT S210) equipped with a fiber laser system (IPG, 500 W) was used to fabricate parts.
A scanning pattern with a rotation angle of 90 • was used for consecutive layers.Before experiments, the building chamber was filled with high-purity argon.As shown in figure 1(c), a small amount of fume appeared during PBF-LB, due to the Mg evaporation.The PBF-LB parameters were optimized with scanning speed (v) ranging from 200 to 1200 mm•s −1 , and laser power (P) ranging from 20 to 100 W, as listed in table 1.The layer thickness (t) and hatching space (h) was determined at 30 µm and 70 µm, respectively.The laser spot was ∼75 µm.Generally, the forming quality was affected by multiple parameters.Therefore, volume energy density (E v ), which was defined as follows: E v = P/(vht), was adopted to evaluate the comprehensive influence of various laser parameters on forming quality.Finally, typical WE43 parts with high forming quality were obtained at optimal parameters, as shown in figure 1(d).

Heat treatment
To determine the appropriate heat treatment parameters, thermodynamic simulation was necessary to predict the variation of phase evolution.Sun et al [32] studied the thermodynamic model of WE43 and revealed that a single α-Mg phase could be obtained as the heating temperature exceeds ∼510 • C. Besides, Mg 41 RE 5 , Mg 12 RE 5 and α-Zr trended to precipitate within the range of 300 • C ∼ 500 • C during aging treatment, while the Mg 24 RE 5 phase would be precipitated as the temperature below 300 • C. Thus, the as-built part was solution treated at 525 • C for 2 h, which was designated as PBF-LB/T4 part.Subsequently, water quenching was employed.Then, PBF-LB/T4 part was isothermally aged at 210 • C and 250 • C, respectively.The corresponding specimens were designated as PBF-LB/T6-1 and PBF-LB/T6-2 parts, respectively.

Microstructure characterization
Phase constitution was analyzed by an x-ray diffractometer (XRD, D8 Advance, Bruker) at a rate of 5 • min −1 .The microstructure was studied using scanning electron microscopy (SEM, Sigma 300, ZEISS, Germany) equipped with an electron back-scattering diffraction detector (EBSD, Nordlys Nano, Oxford, UK).For SEM observation, the samples were mechanically grounded and polished using a diamond polishing compound, then etched through a 3% nitric acid alcohol solution.With regard to the EBSD test, samples were electrolytically polished with an acidic solution.EBSD patterns were obtained by using Atec software with a step size of 1 µm to capture desirable images.The results were further analyzed by using Channel 5 software to obtain corresponding inverse pole figures (IPF) and pole figure maps.The precipitates were further investigated using transmission electron microscopy (TEM, Talos F200X, FEI).Before TEM analysis, samples were cut into a small disc, and then grounded to a thickness of ∼30 µm.Subsequently, the small discs were subjected to ion-beam thinning under cryogenic conditions at 5 kV with a degree of 5 • .

Mechanical properties tests
Vickers hardness was determined using a hardness tester with a load of 2000 g for 25 s.Tensile experiments were conducted using a universal testing machine (MTS C45.105) at 5 × 10 −4 s −1 .Since no significant yielding was observed, the yield strength was determined by a 0.2% offset of elongation.

Soaking tests
To simulate the degradation process of samples in the human body, in-vitro immersion tests were performed in simulated body fluids (SBF).Prior to the experiment, specimens were encapsulated with epoxy resin, leaving an exposed area of 1 cm 2 .Subsequently, the surface of the samples was sanded to 4000 # , ultrasonically cleaned with alcohol for 5 min, followed by drying.A self-built hydrogen collection device was utilized to record hydrogen release volume.After soaking for 7 d, specimens were immersed in a chromic acid solution at 40 • C for 5 min to remove surface degradation products, then blown dry and weighed.The degradation rate was calculated based on hydrogen volume (P H ) and weight loss (P W ), and relevant details were described in our previous works [33].

Electrochemical experiments
In the present study, electrochemical experiments were carried out on an electrochemical workstation (PARSTAT 4000A), which was equipped with a three-electrode system.In detail, the specimens acted as working electrodes, with the platinum electrode and saturated Ag/AgCl electrode serving as counter electrode and reference electrode, respectively.Before testing, samples were immersed in SBF for 30 min to reach a potential stable state.Then, polarization curves were tested over a range of 500 mV of open circuit potential and at a scan rate of 5 mV•s −1 .In addition, electrochemical impedance spectroscopy (EIS) was performed in the frequency range of 10 MHz-100 kHz with a perturbation voltage of 10 mV.The equivalent circuits were fitted by Nyquist and Bode plots in ZSimpwin software.

Cell tests
In the present study, a mouse osteoblast precursor cell line (MC3T3-E1) was adopted to assess cell behavior.Minimum Essential Medium Alpha containing 10% fetal bovine serum and 4% penicillin-streptomycin was used as a cell culture medium.The cells were co-cultured with sterilized scaffolds in a 24-well plate at a density of 5 × 10 4 cells per well.After culturing for 7 d, the cells were stained with a Live/Dead Double Staining Kit (Beyotime, China), and then observed using a confocal laser scanning microscopy (LSM710, Zeiss, Germany).
An indirect method was used to evaluate the biocompatibility.Firstly, the scaffolds were immersed in the culture medium for 3 d to obtain the extracts.Then, the cells were cultured in the extract at a density of 5 × 10 4 cells per well.After 1, 4, and 7 d, the cells were washed with phosphatebuffered saline, and then further incubated with 10% Cell Counting Kit-8 solution for 2 h.The cells incubated in the original culture medium were defined as a control group.Then, the absorbance was determined using a microplate reader at 450 nm.The cell viability was calculated as follows: Cell viability = OD test /OD Control × 100%.

Statistical analysis
All the experiments were performed at least three times.Statistical analysis was performed in SPSS 24.0 software.Data were expressed as mean ± standard deviation.* p < 0.05 was regarded as statistically significant.

Forming quality
PBF-LB parameters optimization was performed by employing laser scanning speed (v) in a range of 200 ∼ 1200 mm•s −1 and laser power (P) in a range of 20 ∼ 120 W, as presented in figure 2(a).Three typical surface morphologies and corresponding cross-section at different E v were shown in figures 2(b) and (c).At E v of 28.6 J•mm −3 , the as-built part exhibited a sandy surface morphology with intermittent scanning tracks like worms, owing to the incomplete powder melting under insufficient energy input.As a result, lack-of-fusion pores with hundreds of micrometers were observed in the cross-section, which undoubtedly caused a low-level densification rate of 96.80%.Using a suitable E v of 53.6 J•mm −3 , the part exhibited smooth and continuous scanning tracks, although a small number of sputtered particles appeared on the surface.In this case, no obvious pores were observed in the matrix, thus achieving an extremely high-level densification rate of 99.91%.As E v further increased to 119.0 J•mm −3 , a few open pores and serious splash were observed on the surface, which was induced by the severe Mg vaporization under excessive energy input.Accordingly, the densification rate dropped to 98.37%.Finally, an effective processing window was identified to obtain near fully-dense (>99.5%)WE43 parts, which was fallen in the range of 60 W ⩽ P ⩽ 80 W and 600 mm•s −1 ⩽ v ⩽ 800 mm•s −1 .That was to say, the reasonable E v for the fabrication of WE43 parts with desirable forming quality varied within 53.6 J•mm −3 ∼ 95.2 J•mm −3 .
In general, the interaction between the laser beam and powder, namely the molten pool evolution feature, directly determined the forming quality of the final parts.At low E v , the powder bed was incompletely melted, and could not obtain sufficient liquid to form a stable molten pool.In this case, the molten pool with high viscosity was difficult to flow and spread, consequently triggering the formation of lackof-fusion pores (figure 2(c)).Using elevated E v , the highenergy laser beam not only guaranteed the complete melting of the powder bed, but also guided the evaporation of molten pool.Under the action of metal vapor recoil, typical keyhole characteristics were formed, as schematically depicted in figure 2(d).In this case, the vapor-induced recoil force was able to promote the Marangoni convection within the molten pool, thereby achieving abundant heat and mass transfer [34].As the laser beam is left, the keyholes would be closed under the driving force of surface tension, thereby finally achieving the complete densification behavior.However, using excessive E v , severe evaporation aggravated the keyhole feature.During subsequent solidification, a large amount of vapor, especially at the bottom of the molten pool, could not escape in time and then trapped in the solidification layer, thus forming typical keyhole defects, as shown in figure 2(c).Nevertheless, keyhole-induced pores in high E v were smaller than the lack of fusion-induced pores in low E v , owing to the different pore-formation mechanisms as discussed above.It should be concerned that Mg alloys owned relatively low boiling points and high saturated vapor pressure, as compared to other metals like Ti alloy.Thus, the keyhole feature in PBF-LB of Mg alloy was difficult to obtain a perfect equilibrium state owing to the serious evaporation of Mg.By optimizing the process parameters, the negative impact of Mg vapor on formability could be effectively reduced, and high-density parts could still be obtained.

Microstructure feature 3.2.1. Microstructure of as-built parts.
The etched crosssection of the as-built part was presented in figure 3(a), which revealed typical signs of fish-scale-shaped molten pools, due to the characteristic layer-by-layer fashion of PBF-LB.Upon close observation, three typical precipitate morphologies could be distinguished in different areas.Clearly, a large number of coarse precipitates (2 ∼ 5 µm) decorated the whole molten pool boundaries, since it underwent a continuous heat exchange process with surrounding powder, substrate, and previously solidified layers that promoted the growth of precipitates.Within the molten pool, mutually parallel banded precipitates extended along the normal direction of the molten pool boundary, and gradually transformed into discrete particles as they approached the center of the molten pool.Distinctively, in the heat-affected zone, the banded precipitates extended nearly parallel with the molten pool boundaries, which were strictly aligned to the normal direction of the temperature gradient.Besides, a small portion of RE oxide particles (figure 3(b)) was presented, which might be originated from the oxide shell of pre-alloyed powder.The EDS maps (figure 3(c)) indicated that the banded precipitates with a spacing distance of ∼500 nm were decorated by Nd-rich particles.
EBSD results of the as-built part were displayed in figure 3(d).In general, the IPF map was dominated by red color, which indicated that the as-built part was anisotropic and owned a strong basal texture along [0001] direction.Furthermore, a typical bi-modal microstructure consisting of columnar and equiaxed grains was observed.Particularly, columnar grains usually initiated at the molten pool boundaries and extended along the normal direction.As a comparison, fine and uniform equiaxed grains were observed within the molten pool.This characteristic could be partially explained by the nature of localized heating during PBF-LB [35].Upon melting, the solidification process occurred directionally from the molten pool boundary due to a high ratio of thermal gradients (G) and low growth rate (R) [36,37].In this case, the nucleated grains tended to grow epitaxially along the direction opposite to heat flux, namely the normal to the molten-pool boundary, which led to the formation of columnar grains [38,39].During subsequent solidification, with the decreasing G/R ratio from the molten-pool border to the center, equiaxed grains tended to nucleate in the undercooled liquid ahead of the columnar zone [40,41].In other words, a columnar-toequiaxed transition occurred in the molten pool, as shown in figure 3(c).However, due to the layer-by-layer fashion of PBF-LB, the microstructure of the previous molten pool was inevitably affected by heat treatment through subsequent laser-scan tracks.Thus, the coarse and elongated grains could extend hundreds of microns in the BD and persist epitaxially through consecutive layers [42].

Microstructure observations of post-heat treatment
parts.The age hardening response curves of the PBF-LB/T4 part were depicted in figure 4(a).Obviously, the hardness increased first and then decreased with the prolongation of aging time.Besides, PBF-LB/T4 part aged at 250 • C reached peak-aged status faster than 210 • C, while the peak-hardness value was significantly lower than 210 • C. In this work, peak-aged status samples were selected for investigation.The phase evolution of all parts was revealed by XRD analysis, as shown in figure 4(b).For the as-built part, it exhibited strong peaks corresponding to α-Mg and minor peaks corresponding SEM analysis of post-treatment parts on transverse crosssections was carried out to reveal the distribution and composition of phases, as presented in figure 5. Different from as-built parts, heat-treated parts exhibited no obvious pattern of laser scanning track.For PBF-LB/T4 part, there were still several flake-shaped phases in the matrix.Based on the EDS analysis of point A, these white compounds could be identified as MgO and Y 2 O 3 .As for PBF-LB/T6-1 part, a large number of fine and rectangular particles with sub-micrometer were precipitated.Combining the EDS and XRD results, it was deduced that these particles were Mg 24 Y 5 intermetallic.With regard to PBF-LB/T6-2 part, abundant coarse and reticulate precipitates emerged along grain boundaries, which could be identified as Y, O, and Zr-enriched intermetallic by EDS mapping (marked by the red box).This meant that Zr and Y elements tended to preferentially segregate at grain boundaries during hightemperature aging.In addition, there was also a great quantity of fine and needle-like precipitates that penetrated the whole grains, which might be the β 1 phases (Mg 3 Nd).
TEM analysis was applied to further identify the phase composition of the PBF-LB/T6-2 part.Network precipitates were observed at the grain boundaries, as the brightfield image shown in figure 6(a).As reported by previous studies, the network precipitates were the equilibrium phase of Mg 41 Nd 5 .Notably, needle-like precipitates and rectangular particles were observed within the grains, as shown in figure 6(b).Based on the interplanar spacing of ∼0.3027 nm (figure 6(c)), needle-like phases were determined as β 1 phases (Mg 3 Nd), which was also proved by the corresponding SAED pattern.Besides, the rectangular particles also presented in PBF-LB/T6-1 part, were identified as Mg 24 Y 5 intermetallic by the inserted SAED pattern (figure 6(d)).
EBSD analysis was adopted to investigate the texture of heat-treated parts, as shown in figure 7.For PBF-LB/T4 part, it still showed a strong texture with a privileged grain orientation along <0001>.The columnar grains were eliminated in place of equiaxial grains, manifesting that the grains of the as-built part experienced recrystallization [43].At the same time, the grain anisotropy was reduced significantly.However, its grain size inevitably turned coarse during this high-temperature heat treatment.As for PBF-LB/T6-1 part, it exhibited random orientations along three basic grain orientations, which proved a considerable influence of aging treatment on the texture evolution [44].Usually, the precipitates were able to change the orientation of grains, leading to a reduction in the intensity of certain crystallographic planes.Besides, the microstructure remained equiaxial grains with slightly coarsened grain size.It was believed that the precipitation phase acted as a nucleation site for new grains during recrystallization [45].With the increasing of aging temperature, the grain size tended to enlarge further, which was due to the partial dissolution of Mg 24 Y 5 intermetallic and re-precipitation of the β 1 phase [46,47].

Mechanical properties
The Vickers microhardness evolution on transverse crosssections was measured, as shown in figure 8(a).For the as-built part, the microhardness exhibited a cyclical fluctuation (74.7 ∼ 95.2 HV) along BD, owing to the microstructure inhomogeneity caused by layer-by-layer fashion.After solution treatment, the detected microhardness curves varied within a very small scope.However, the average microhardness was considerably decreased from 86.1 to 71.7 HV, which might be attributed to the grain growthinduced softening and elimination of residual stress.After subsequent aging treatment (PBF-LB/T6-1), the microhardness was significantly enhanced by homogeneous fine Mg 41 Nd 5 phases.With increasing the aging temperature (PBF-LB/T6-2), the microhardness declined and fluctuated widely, because of the uneven distribution of phases.Previous studies stated that the influence of the second phase on precipitation strengthening mainly included its size, density, and morphology [48].Nevertheless, the precipitation hardening and grain growth-induced softening competed with each other, which comprehensively determined the final properties [49].
A tensile test was further applied to assess the mechanical properties, with stress-strain curves and corresponding tensile parameters plotted in figures 8(b) and (c).As compared to the as-built part, the ultimate tensile strength (UTS) and yield strength (YS) of the PBF-LB/T4 part were decreased to (224.8 ± 4.5) MPa and (125.6 ± 2.4) MPa, respectively, while the elongation increased to 20.1 ± 0.5%.After aging treatment, the UTS and YS were improved significantly, but the fracture elongation showed a certain loss.With respect to PBF-LB/T6-1 part, the YS and UTS were increased to (250.2 ± 3.5) MPa and (312 ± 3.7) MPa, respectively, while the elongation was maintained at a considerable level of (15.2 ± 0.3) %.As for PBF-LB/T6-2 part, despite the YS and UTS being still higher than that of the as-built part, the elongation was reduced.This phenomenon accorded with the longbelieved recognition that the deformability of Mg alloy was extremely sensitive to grain size and precipitate statement [50,51].Although aging heat treatment leads to a certain degree of grain size growth, a large number of uniformly precipitated fine grains contribute to a strong precipitation-strengthening effect, thus significantly improving mechanical properties.For PBF-LB/T6-2 part, the coarsened grains and precipitates at grain boundaries tended to deteriorate the ductility.Besides, the fine β ′ phases within the grains brought about negligible Orowan strengthening effect, since they could be sheared by basal dislocations during deformation [52].A comparison among the tensile properties of WE43 parts manufactured by various processes was summarized in figure 8(d).In comparison to the parts prepared by casting and powder metallurgy, the as-built parts exhibited considerably superior mechanical properties, which were mainly ascribed to the refined grains caused by the extremely fast cooling rate (even 10 5 K•s −1 ) during PBF-LB.Particularly, after heat treatment, a superior combination of UTS and elongation could be achieved.One reasonable explanation was that the heat treatment eliminated the defects like pores or cracks, and promoted the precipitation of desirable strengthening particles, like Mg 24 Y 5 in the present study.
To further clarify the fracture mechanism, the fracture surfaces were presented in figure 9(a).Both as-built and PBF-LB/T4 parts exhibited typical ductile fracture morphologies with numerous micron-scaled dimples.For PBF-LB/T6-1 part, a combination of dimples, tear ridges, and cleavage facets was observed, as highlighted in figure 9(b), which revealed it experienced a hybrid fracture mechanism.As for PBF-LB/T6-2 part, a relatively large-size cleavage surface presented on the fracture, indicating its poor plastic deformation capacity.These results proved that the heat treatment altered the tensile performance by tailoring the microstructure.The essence of tensile fracture was that the micro-voids produced by plastic deformation undergo nucleation, growth, aggregation, and finally lead to fracture [64].For as-built and PBF-LB/T4 parts that owned fine α-Mg grains and precipitates, numerous small deformations could be initiated via dislocation motion, twinning, and grain boundary sliding, thus achieving the formation of small dimples.For PBF-LB/T6-1 part, the α-Mg grains experienced recrystallization, resulting in grain growth and reduction of dislocation defects, which greatly limited the deformation ability.Therefore, the cracks propagated rapidly, particularly at the interface of precipitates, thus leading to the appearance of huge tear ridges and cleavage facets.As for PBF-LB/T6-2 part, the further grown precipitates acted stress concentrators and promoted the propagation of cracks, thus exhibiting high susceptibility to brittle fracture.

Degradation behavior
The biodegradation behavior was assessed using immersion tests, with the corrosion surface displayed in figure 10.The as-built part showed an uneven surface with a large number of flake-like sheets, which rendered the Mg-matrix illegible.It was believed that the Nd-rich precipitates with relatively high corrosion potential triggered serious localized corrosion.As a comparison, PBF-LB/T4 and PBF-LB/T6-1 parts exhibited relatively flat surfaces with a small amount of corrosion products.Notably, PBF-LB/T6-2 part, similar to the as-built part, also showed a rough surface that was covered with heavy corrosion product.After cleaning up the corrosion product, the as-built part revealed a mound-shaped surface.For PBF-LB/T4 and PBF-LB/T6-1 parts, a relatively flat surface was observed, as it only undertook slight corrosion.As was expected, uneven and localized corrosion pits were presented on PBF-LB/T6-2 part.
The cross-sections of the corrosion product layer were shown in figure 11(a).Clearly, the as-built part possessed an inhomogeneous degradation product film, which exhibited obvious cracks and holes.Such a corrosion product layer was more vulnerable to rupture and thereby provided more possibilities to contact between corrosive media and substrate.In contrast, dense and homogeneous films were observed on PBF-LB/T4 and PBF-LB/T6-1 parts, with an average thickness of about 28 µm and 34 µm, respectively.The EDS mapping exhibited that the corrosion layers were mainly composed of oxides.The hydrogen release rate and weight loss were used to estimate the corrosion rates, as depicted in figures 12(b) and (c).The hydrogen release rate of all samples showed a pattern of first increase and then decrease, as the corrosion film inhibited the follow-up corrosion of the substrate.The calculated degradation rates were displayed in figure 12(d).Clearly, the P H and P W of PBF-LB/T6-1 parts were (0.68 ± 0.21) mm yr −1 and (0.65 ± 0.16) mm yr −1 , respectively, which was significantly lower than that of asbuilt parts.In addition, the P H was always higher than P W for all samples, which might be related to inevitable gas escape during the hydrogen collection.
Polarization curves were further used to evaluate the corrosion susceptibility, as depicted in figure 12(a).As compared to the as-built and PBF-LB/T6-2 parts, PBF-LB/T4 and PBF-LB/T6-1 parts exhibited positive corrosion potential (E corr ) of (−1.46 ± 0.02) V and (−1.48 ± 0.03) V, respectively, manifesting a relatively low corrosion tendency.Besides, the anodic zone of as-built and PBF-LB/T6-2 parts showed an obvious inflection owing to the existence of local corrosion [65].Therefore, a cathodic polarization branch was applied to calculate corrosion current density (I corr ), with results presented in table 2. PBF-LB/T4 and PBF-LB/T6-1 parts exhibited a relatively low I corr of (24.3 ± 1.0) µA cm −2 and (32.1 ± 1.2) µA cm −2 , respectively.The collected EIS curves were plotted in figure 12(b).Obviously, the Nyquist curves exhibited two distinct types, suggesting a different electrochemical mechanism.In detail, the Nyquist curves of as-built and PBF-LB/T6-2 parts were extended to the fourth quadrant as the frequency approached zero, indicating that the corrosion product film-covered Mg matrix had undergone obvious cracking.
Based on the Nyquist plots, the equivalent circuit was fitted, as shown in figures 12(c) and (d), and the corresponding parameters were listed in table 2. The corrosion film formed on Mg alloys usually owned a bilayer structure.Thus, the equivalent electrical circuit model R(Q(R(QR))) was used to fit the EIS results.However, an inductive loop was observed at the low-frequency region of as-built and PBF-LB/T6-2 parts.Therefore, it was necessary to add an inductive component to fit the EIS results, namely, model R(Q(R(QR(RL)))).The R s , R f, and R ct referred to the solution resistance, film resistance, and charge transfer resistance of the electric double layer, respectively [66].The CPE f and CPE dl represented the capacitors of corrosion product film and electric double layer, respectively [67].Clearly, the charge transfer resistance (R ct ) and film resistance (R f ) of the as-built part was relatively small, which corresponded to its poor corrosion resistance.After heat treatment, both PBF-LB/T4 and PBF-LB/T6-1 parts showed increased R ct and R f .It was evidenced that the electron diffusion process that occurred between the electrolyte to the alloy/oxide interface was suppressed [68].Besides, there was no inductive loop (R L + L) that appeared in equivalent circuits of PBF-LB/T4 and PBF-LB/T6-1 parts, which indicated no obvious local corrosion.These results further proved that the regulation of microstructure via suitable heat treatment was able to alter the corrosion mechanism and improve the anti-corrosion ability of LPBF-processed WE43 alloy.In the present study, the unique layer-wise processing fashion of the LPBF process resulted in uneven microstructure, particularly the unevenly distributed precipitates like Nd-rich particles.Previous studies reported that the corrosion resistance of Mg alloys was closely related to the types, densities, distributions, and volume fractions of precipitates [69].Meanwhile, the forming defect like pores or cracks would further accelerate the degradation behavior of the Mg matrix.Undoubtedly, the suitable heat treatment improved the uniformity of the microstructure and

Specimens
Ecorr (V SCE ) As-built −1.eliminated the forming defect.More significantly, the uniform and fine precipitates reduced the tendency for micro galvanic corrosion.Accordingly, uniform corrosion slowly progressed, and thereby achieved a relatively slow digitation with uniform corrosion film, as shown in figure 11(a).

In-vitro biocompatibility
As presented in figure 13(a), the cells cultured in PBF-LB/T6-1 scaffold extract group exhibited relatively high cell viability as compared with those cultured in the as-built scaffold extract group.At 50% extract concentration, the cell viability for both two groups was enhanced.This indicated that the released metal ions and other products caused by scaffold degradation exerted a negative impact on cell growth.However, the cell viability gradually increased with culture time increasing.A significant difference was observed between the as-built and PBF-LB/T6-1 groups ( * p < 0.05).After culturing for 7 d, the cells on scaffolds were directly captured using a confocal microscope, as presented in figure 13(b).Clearly, more cells were observed on the PBF-LB/T6-1 scaffold.Particularly, the cells uniformly adhered to the PBF-LB/T6-1 scaffold, which connected with each other and nearly crossed the pores.As a comparison, a few of the cells firmly adhered to the strut of the as-built scaffold, while no cells stretched over the pores.These results clearly indicated that PBF-LB/T6-1 scaffold was more conducive to cell growth.As mentioned above, PBF-LB/T6-1 scaffold showed a higher anti-corrosion ability than the as-built scaffold.Thus, there was relatively small metal ions release and a mild pH in the PBF-LB/T6-1 scaffold group, which provided a friendly environment for cell growth and adhesion.

Conclusions
In the present work, WE43 parts were firstly prepared using PBF-LB, and then subjected to heat treatment.The microstructure and performance were studied.The conclusions were drawn as follows: (1) WE43 parts with a high densification rate of 99.91% were obtained using suitable parameters.It exhibited an inhomogeneous microstructure as well as a strong texture.Nd-rich precipitates were observed in the matrix and showed various morphologies at different regions of the molten pool.(2) After solution heat treatment and aging condition, fine and rectangular Mg 24 Y 5 particles with a sub-micrometer were precipitated.The grain size was slightly coarsened but showed relatively random and uniform orientation.As aging time prolonged, the precipitates turned into Mg 41 Nd 5 and formed a network structure. ( (a)).A laser diffractometer (Mastersizer 3000E) was used to examine the particle size distribution, which exhibited an approximately normal distribution with a d 10 of 17.30 µm, d 50 of 36.20 µm and d 90 of 56.88 µm (figure 1(b)).Such a size distribution guaranteed a stable powder spreading during PBF-LB.

Figure 1 .
Figure 1.Characterization of powder and PBF-LB process.(a) Original WE43 powder and (b) its particle size distribution; (c) in-suite image of PBF-LB process in the building chamber; (d) as-built WE43 parts.

Figure 2 .
Figure 2. Forming quality of as-built parts at various parameters.(a) Densification rate map; (b) top surface and (c) cross-section of as-built parts obtained at 28.6 J•mm −3 , 53.6 J•mm −3 , and 119.0 J•mm −3 ; (d) schematic of molten pool with keyhole feature.

Figure 3 .
Figure 3. Microstructure characterization of the as-built part.(a) SEM; (b) EDS of point A in figure (a); (c) bright-field image with EDS analysis; (d) IPF and pole figure maps obtained by EBSD.

Figure 4 .
Figure 4. Evolution of hardness and phase composition during heat treatment.(a) Age hardening curves; (b) XRD patterns showing the phase composition.

Figure 11 .
Figure 11.Characterization of degradation behavior in SBF.(a) Cross-section of corrosion film with EDS mapping; (b) hydrogen volume evolution; (c) weight loss; (d) degradation rates.
) The YS and UTS of the PBF-LB/T6-1 part were significantly improved to (250.2 ± 3.5) MPa and (312 ± 3.7) MPa, respectively, while the elongation was maintained at a considerable level of 15.2% ± 0.3%.(4) Homogenized microstructure obtained by post-treatment effectively inhibited the tendency of micro galvanic corrosion and promoted the development of passivation film.Therefore, the degradation rate was decreased by an order of magnitude.evolution of Al-Cu-Mg alloy produced by laser powder bed fusion: effect of heat treatment Mater.Charact.165 110364 [31] Kong D C, Dong C F, Ni X Q, Zhang L, Yao J Z, Man C, Cheng X Q, Xiao K and Li X G 2019 Mechanical properties and corrosion behavior of selective laser melted 316L stainless steel after different heat treatment processes J. Mater.Sci.Technol.35 1499-507 [32] Sun W T, Wu B, Fu H, Yang X S, Qiao X G, Zheng M Y, He Y, Lu J and Shi S Q 2022 Combining gradient structure and supersaturated solid solution to achieve superior mechanical properties in WE43 magnesium alloy J. Mater.Sci.Technol.99 223-38 [33] Yang Y W, Ling C R, Li Y G, Peng S P, Xie D Q, Shen L D, Tian Z J and Shuai C J 2023 Microstructure development and biodegradation behavior of additively manufactured Mg-Zn-Gd alloy with LPSO structure J. Mater.Sci.Technol.144 1-14 [34] Duan R X, Li S, Cai B, Tao Z, Zhu W W, Ren F Z and Attallah M M 2021 In situ alloying based laser powder bed fusion processing of β Ti-Mo alloy to fabricate functionally graded composites Composites A 222 109059 [35] Raffeis I, Adjei-Kyeremeh F, Vroomen U, Westhoff E, Bremen S, Hohoi A and Bührig-Polaczek A 2020 Qualification of a Ni-Cu alloy for the laser powder bed fusion process (LPBF): its microstructure and mechanical properties Appl.Sci. 10 3401 [36] Roehling T T, Wu S S Q, Khairallah S A, Roehling J D, Soezeri S S, Crumb M F and Matthews M J 2017 Modulating laser intensity profile ellipticity for microstructural control during metal additive manufacturing Acta Mater.128 197-206 [37] Li J F, Wu Y X, Xue L F and Wei Z Y 2023 Laser powder bed fusion in-situ alloying of refractory WTa alloy and its microstructure and mechanical properties Addit.Manuf.67 103493 [38] Chen N N, Khan H A, Wan Z X, Lippert J, Sun H, Shang S L, Liu Z K and Li J J 2020 Microstructural characteristics and crack formation in additively manufactured bimetal material of 316L stainless steel and Inconel 625 Addit.Manuf.32 101037 [39] Shi R P, Khairallah S A, Roehling T T, Heo T W, McKeown J T and Matthews M J 2020 Microstructural control in metal laser powder bed fusion additive manufacturing using laser beam shaping strategy Acta Mater.184 284-305 [40] Liu X H, Zhao C C, Zhou X, Shen Z J and Liu W 2019 Microstructure of selective laser melted AlSi 10 Mg alloy Mater.Des.168 107677 [41] Bermingham M J, Stjohn D H, Krynen J, Tedman-Jones S and Dargusch M S 2019 Promoting the columnar to equiaxed transition and grain refinement of titanium alloys during additive manufacturing Acta Mater.168 261-74 [42] Roehling T T, Shi R P, Khairallah S A, Roehling J D, Guss G M, McKeown J T and Matthews M J 2020 Controlling grain nucleation and morphology by laser beam shaping in metal additive manufacturing Mater.Des.195 109071 [43] Que Z Q, Chang L T, Saario T and Bojinov M 2022 Localised electrochemical processes on laser powder bed fused 316 stainless steel with various heat treatments in high-temperature water Addit.Manuf.60 103205 [44] Davies S J, Jeffs S P, Coleman M P and Lancaster R J 2018 Effects of heat treatment on microstructure and creep properties of a laser powder bed fused nickel superalloy Mater.Des.159 39-46 [45] Zhang Q, Li Q A, Chen X Y, Zhao J X, Bao J and Chen Z Y 2021 Dynamic precipitation and recrystallization mechanism during hot compression of Mg-Gd-Y-Zr alloy J. Mater.Res.Technol.15 37-51

Table 1 .
PBF-LB parameters are applied in the present work.

Table 2 .
Fitting results of polarization curves and Nyquist curves.