In-Situ Laser Polishing Additive Manufactured AlSi10Mg: Effect of Laser Polishing Strategy on Surface Morphology, Roughness and Microhardness

Laser polishing is a widely used technology to improve the surface quality of the products. However, the investigation on the physical mechanism is still lacking. In this paper, the established numerical transient model reveals the rough surface evolution mechanism during laser polishing. Mass transfer driven by Marangoni force, surface tension and gravity appears in the laser-induced molten pool so that the polished surface topography tends to be smoother. The AlSi10Mg samples fabricated by laser-based powder bed fusion were polished at different laser hatching spaces, passes and directions to gain insight into the variation of the surface morphologies, roughness and microhardness in this paper. The experimental results show that after laser polishing, the surface roughness of Ra and Sa of the upper surface can be reduced from 12.5 μm to 3.7 μm and from to 29.3 μm to 8.4 μm, respectively, due to sufficient wetting in the molten pool. The microhardness of the upper surface can be elevated from 112.3 HV to 176.9 HV under the combined influence of the grain refinement, elements distribution change and surface defects elimination. Better surface quality can be gained by decreasing the hatching space, increasing polishing pass or choosing apposite laser direction.


Introduction
Additive manufacturing technology (AM), also known as 3D printing technology, is an advanced material forming technology. This technology is completely different from traditional manufacturing technologies. Parts with high density and complicated structure can be quickly obtained using AM technology [1,2]. Laser-based powder bed fusion (L-PBF) is a widely used metal additive manufacturing technology using a high-energy-density laser to scan through the metal powder bed [3]. AlSi10Mg alloy is a typical AM material. Many numerical and experimental studies regarding L-PBF AlSi10Mg parts have been reported [4,5]. Although the L-PBF process offers many advantages compared to traditional techniques, many defects inevitably occur in the L-PBF parts, such as unexpected rough surfaces and unavoidable pore defects and crack defects [6,7]. The mechanical properties of AM parts are significantly influenced by the surface roughness and defects. For example, the shallow surface cracks can lead to low fatigue resistance of AM stainless steel parts [8]. The fatigue strength is extremely affected by the surface roughness of AM Ti-6Al-4V

Physical Model and Assumptions
Nanosecond-laser polishing is based on the surface remelting to improve the surface quality of the part. A schematic description of laser polishing is shown in Figure 1. When the laser beam is irradiated on the upper surface, the temperature of the upper surface soon reaches its melting point, and then the molten pool and heated affected zone (HAZ) occur. The molten pool moves along the laser scanning direction.
To account for the rough surface evolution mechanism during laser polishing, the finite element method was used in the theoretical study. In this paper, the commercial software (COMSOL multi-physics 5.4, COMSOL Inc., Stockholm, Sweden) was employed considering heat transfer, laminar flow, gravity, recoil pressure, surface tension and Marangoni effect. The established time-dependent multi-physics coupled two-dimensional model was displayed in Figure 2. The level-set method was adopted to track the geometric properties of the interface between the rough surface and the protective atmospheric gas. As shown in Figure 2, the sizes of the laser polishing layer and protective argon gas zone were 2.5 mm × 1 mm and 2.5 × 0.5 mm, respectively. The upper surface was set as a random rough surface. The laser beam scanned along the positive X-axis. Both the heating and cooling time were 1 ms. The material physical properties and process parameters in this model are shown in Table 1. To account for the rough surface evolution mechanism during laser polishing, the finite element method was used in the theoretical study. In this paper, the commercial software (COMSOL multi-physics 5.4, COMSOL Inc., Stockholm, Sweden) was employed considering heat transfer, laminar flow, gravity, recoil pressure, surface tension and Marangoni effect. The established time-dependent multi-physics coupled two-dimensional model was displayed in Figure 2. The level-set method was adopted to track the geometric properties of the interface between the rough surface and the protective atmospheric gas. As shown in Figure 2, the sizes of the laser polishing layer and protective argon gas zone were 2.5 mm × 1 mm and 2.5 × 0.5 mm, respectively. The upper surface was set as a random rough surface. The laser beam scanned along the positive X-axis. Both the heating and cooling time were 1 ms. The material physical properties and process parameters in this model are shown in Table 1.   To account for the rough surface evolution mechanism during laser polishing, the finite element method was used in the theoretical study. In this paper, the commercial software (COMSOL multi-physics 5.4, COMSOL Inc., Stockholm, Sweden) was employed considering heat transfer, laminar flow, gravity, recoil pressure, surface tension and Marangoni effect. The established time-dependent multi-physics coupled two-dimensional model was displayed in Figure 2. The level-set method was adopted to track the geometric properties of the interface between the rough surface and the protective atmospheric gas. As shown in Figure 2, the sizes of the laser polishing layer and protective argon gas zone were 2.5 mm × 1 mm and 2.5 × 0.5 mm, respectively. The upper surface was set as a random rough surface. The laser beam scanned along the positive X-axis. Both the heating and cooling time were 1 ms. The material physical properties and process parameters in this model are shown in Table 1.   Figure 2. Representation of the model used in the numerical simulation. To simplify the numerical calculations, several assumptions were proposed as follows: (1) The laser beam is regarded as a continuous Gaussian heat source due to the short pulse separation pass of the laser (pulsed Nd: YAG laser) [27]. (2) The deformation of mechanical behavior has negligible effect on fluid field [27].
(3) The laser absorptivity of the material is assumed to be constant [27]. (4) The flow field in the molten pool is assumed as an incompressible Newtonian laminar flow [27]. (5) The material is isotropic and homogenous [27]. (6) The metal loss caused by evaporation during L-PBF is ignored [28].

Governing Equations and Boundary Conditions
As shown in Figure 2, the upper surface of laser polishing domain is irradiated by the laser heat flux. The convection and radiation to the environment were also considered. The heat transfer equation is governed by Fourier s Law as follows: where ρ is the density, C m is the material heat capacity, λ is the thermal conductivity, Q is the source term of heat transfer and u is the flow velocity. The flow field in the molten pool can be calculated from the Navier-Stokes equation as follows: where p is the flow pressure, I is the identity matrix, µ is the flow dynamic viscosity and F V is a source term corresponding to surface tension, gravity and Marangoni force. The laser beam can be descripted as a Gaussian moving heat and the heat flux Q laser can be defined as: where A is the laser beam absorptivity, P laser is the laser power, r 0 is the laser beam radius, x 0 is the initial laser beam position and V laser is the laser beam speed. During laser polishing, when the temperature reaches the melting point of the material, the upper layer material starts to melt and transforms from the solid phase to the liquid phase. The heat capacity, C m , is used to describe the latent heat of phase change.
where C s and C l are specific heat of solid and liquid, respectively. L m refers to the latent heat of fusion. θ is phase indicator, which can be described as: where T l is the liquid temperature and T s is the solid temperature. α m is the latent heat distribution of phase change, which can be expressed as follows: where ρ s is the solid density and ρ l is the liquid density. The Marangoni effect and the surface tension caused by the thermal gradient on the molten area can be given as follows: where κ is the curvature, γ is the surface tension coefficient, n is the unit normal to the surface and ∇ s is the surface gradient operator. ∇ s γ represents the Marangoni effect due to the temperature gradient. The escaping vapor leads to a recoil pressure as the temperature of the upper surface rises to the boiling point. The recoil pressure, P recoil , can be summarized as follows: where P 0 is the ambient pressure, T is the surface temperature, H V is the latent heat of evaporation, R is the universal gas constant and T v is the evaporation temperature.

Numerical Simulation Results and Discussions
The temperature distribution and the rough surface evolution process can be shown in Figures 3 and 4. The temperature of the upper surface starts to rise under a stationary laser source. When the temperature rises to the melting point, the molten pool appears. The molten pool is gradually expanded upon continuous heating. It can be observed that there is a distribution of temperature gradient along the molten pool. The molten pool moves along the positive X direction as the laser beam moves. In the polished area, the material of the convex starts to flow to the concave at t = 0.25 ms. That is because the temperature coefficient is negative, so that the material tends to flow from high-temperature regions to low-temperature regions. At the same time, the height difference between the crest and trough is decreased gradually. At t = 1 ms, the upper surface starts to cool down.
The high-temperature regions gradually disappear. Little change happens in the surface morphology due to the high cooling rate. After cooling (t = 2 ms), the free surface is obviously smoothed. However, due to the high heating and cooling rate, wetting and flow of the molten pool is not enough, so that a completely flat surface cannot be easily obtained.  The velocity distribution can be shown in Figure 5. It can be observed that the maximum velocity can reach 1.8 m/s in the molten pool during laser polishing. Li et al. [29] found that the surface tension and Marangoni flow played key roles in smoothing the surface during laser polishing Ti-6Al-4V. In this paper, due to Marangoni flow and surface tension, the rough surface moves into the tail region of the molten pool. The vortex is generated in the molten pool, which had a great impact on the mass transfer and surface topography. Then, the molten material flows down under the effect of gravity. Up to 1 ms, the velocity of the flow in the molten pool approaches around zero. This can explain why the morphology does not change much in the cooling process.  The velocity distribution can be shown in Figure 5. It can be observed that the maximum velocity can reach 1.8 m/s in the molten pool during laser polishing. Li et al. [29] found that the surface tension and Marangoni flow played key roles in smoothing the surface during laser polishing Ti-6Al-4V. In this paper, due to Marangoni flow and surface tension, the rough surface moves into the tail region of the molten pool. The vortex is generated in the molten pool, which had a great impact on the mass transfer and surface topography. Then, the molten material flows down under the effect of gravity. Up to 1 ms, the velocity of the flow in the molten pool approaches around zero. This can explain why the morphology does not change much in the cooling process. The velocity distribution can be shown in Figure 5. It can be observed that the maximum velocity can reach 1.8 m/s in the molten pool during laser polishing. Li et al. [29] found that the surface tension and Marangoni flow played key roles in smoothing the surface during laser polishing Ti-6Al-4V. In this paper, due to Marangoni flow and surface tension, the rough surface moves into the tail region of the molten pool. The vortex is generated in the molten pool, which had a great impact on the mass transfer and surface topography. Then, the molten material flows down under the effect of gravity. Up to 1 ms, the velocity of the flow in the molten pool approaches around zero. This can explain why the morphology does not change much in the cooling process. generated in the molten pool, which had a great impact on the mass transfer and topography. Then, the molten material flows down under the effect of gravity. ms, the velocity of the flow in the molten pool approaches around zero. This can why the morphology does not change much in the cooling process.

Sample Fabrication
The material of the powder particles was selected to be AlSi10Mg. Figure 6a shows scanning electron microscope (SEM, Tescan Mira3 SEM, Tescan Ltd., Brno, Czech Republic) images of the gas atomized AlSi10Mg particles. Most of the powder particles were spherical in shape. The particle diameter of the AlSi10Mg metal powders used in this paper ranged from 20 µm to 60 µm. The average diameter of the powders was 30 µm. The nine samples, each measuring 10 mm × 10 mm × 2.5 mm, were fabricated using Dimetal-280 L-PBF equipment (South China University of Technology, Guangzhou, China). A continuous wave fiber laser (IPG YLR-400-WC, IPG Photonics Corporation, Oxford, MA, USA) with a wavelength of 1060 nm, an output power of 70 W and a spot size of 70 µm was used during the printing process. The laser scanning speed, hatching space and layer thickness were 1 m/s, 70 µm and 30 µm, respectively, in the manufacturing process. The cubic samples were printed at the center of the substrate plate.

Sample Fabrication
The material of the powder particles was selected to be AlSi10Mg. Figure 6a shows scanning electron microscope (SEM, Tescan Mira3 SEM, Tescan Ltd., Brno, Czech Republic) images of the gas atomized AlSi10Mg particles. Most of the powder particles were spherical in shape. The particle diameter of the AlSi10Mg metal powders used in this paper ranged from 20 μm to 60 μm. The average diameter of the powders was 30 μm. The nine samples, each measuring 10 mm × 10 mm × 2.5 mm, were fabricated using Dimetal-280 L-PBF equipment (South China University of Technology, Guangzhou, China). A continuous wave fiber laser (IPG YLR-400-WC, IPG Photonics Corporation, Oxford, MA, USA) with a wavelength of 1060 nm, an output power of 70 W and a spot size of 70 μm was used during the printing process. The laser scanning speed, hatching space and layer thickness were 1 m/s, 70 μm and 30 μm, respectively, in the manufacturing process. The cubic samples were printed at the center of the substrate plate. After printing, there was no need to take the samples out of the printer. The laser of the printer was also employed to polish the samples. This enables in-situ polishing to avoid oxidation in the polishing process. The power and speed of the polishing laser were set to 400 W and 0.5 m/s, respectively, in this paper. The laser polishing strategies were listed in Table 2. The different polishing directions can be seen in Figure 7. The direction D1 means that the polishing laser scans along as the same as the printing direction. The direction D2 means that the polishing laser scans along negative Y direction. The directions D3 and D4 refer to the angles between the polishing laser and the printing laser as 45° and 90°, respectively. After printing, there was no need to take the samples out of the printer. The laser of the printer was also employed to polish the samples. This enables in-situ polishing to avoid oxidation in the polishing process. The power and speed of the polishing laser were set to 400 W and 0.5 m/s, respectively, in this paper. The laser polishing strategies were listed in Table 2. The different polishing directions can be seen in Figure 7. The direction D1 means that the polishing laser scans along as the same as the printing direction. The direction D2 means that the polishing laser scans along negative Y direction. The directions D3 and D4 refer to the angles between the polishing laser and the printing laser as 45 • and 90 • , respectively.  After polishing, the samples were separated from the substrate b discharge machine (FH-020C, Suzhou Xingjie CNC Technology Co., L for follow-up measurement.

Morphology Observation, Roughness Tests, Cross-Section Observation Tests
The samples were ultrasonically cleaned in the alcohol for 20 mi powders and dirt. Scanning electron microscopy (SEM; Tescan Mira Brno, Czech Republic) was employed to observe the microstructures sive spectrometer (EDS) was applied to take the elemental analysi measurement system for surface profile (Form Talysurf PGI  After polishing, the samples were separated from the substrate by a wire cut, electric discharge machine (FH-020C, Suzhou Xingjie CNC Technology Co., Ltd., Suzhou, China) for follow-up measurement.

Morphology Observation, Roughness Tests, Cross-Section Observation and Microhardness Tests
The samples were ultrasonically cleaned in the alcohol for 20 min to remove residual powders and dirt. Scanning electron microscopy (SEM; Tescan Mira3 SEM, Tescan Ltd., Brno, Czech Republic) was employed to observe the microstructures. The energy dispersive spectrometer (EDS) was applied to take the elemental analysis. A comprehensive measurement system for surface profile (Form Talysurf PGI 830, Taylor Hobson Ltd., Leicester, UK) and 3D Optical Profiler (NewView 8000, ZYGO Ltd., Middlefield, OH, USA) were respectively taken to measure the surface roughness of Ra and Sa. Ra and Sa can be defined as follows. On the upper face of each sample, the six measurement results were averaged to obtain the final Ra value. The center zones of the samples were selected as the Sa measurement areas. The sizes of the upper surface and Ra tested area were 10 mm × 10 mm and 7 mm × 7 mm, respectively. The description of the Ra measurement is shown in Figure 8a.
where l is the surface profile length, y(x) is the deviation of the surface profile at the point x from the mean surface profile height, A is the measured area and Z(x, y) represents the height of the surface, relative to the best fitting surface.
were averaged to obtain the final Ra value. The center zones of the samples were selected as the Sa measurement areas. The sizes of the upper surface and Ra tested area were 10 mm × 10 mm and 7 mm × 7 mm, respectively. The description of the Ra measurement is shown in Figure 8a.
where l is the surface profile length, y(x) is the deviation of the surface profile at the point x from the mean surface profile height, A is the measured area and Z(x, y) represents the height of the surface, relative to the best fitting surface.  The cross-sections of the samples with/without laser polishing were observed with an optical microscope (OM, MR-5000, Nanjing, China). Before observation, the crosssections of the samples were etched by Keller s solution (1%HF + 1.5% HCL + 2.5%HNO3 + 95%H 2 O).
The Vickers microhardness measurement of the upper surfaces of the samples was carried out on a Vickers microhardness testing machine (430SVD, Wilson Ltd., Fort Wort, TX, USA). A load of 9.8 N and an indentation pass of 10 s were used in the Vickers microhardness tests. As can be seen in Figure 8b, the nine tested points around the center of the upper surface were chosen to measure. The final microhardness value was obtained by averaging the nine values.

Experimental Results and Discussions
Figure 9 depicts that the surface pore defects can be eliminated by the post laser. Under the input of the polishing laser energy, molten material was transported along the temperature gradient [30]. Due to surface tension and Marangoni force, the surface pore defects were dragged to the tail in the molten pool, connected with the previous solidified layer, and finally removed. The optical morphologies of the surfaces with/without polishing can be shown in Figure 10. After laser polishing, there were more ordered and obvious tracks and less large peaks and valleys on the upper surface. The height of the peak and the depth of the valley were obviously reduced. The surface roughness of the nine samples are shown in Table 3.
temperature gradient [30]. Due to surface tension and Marangoni force, the surface pore defects were dragged to the tail in the molten pool, connected with the previous solidified layer, and finally removed. The optical morphologies of the surfaces with/without polishing can be shown in Figure 10. After laser polishing, there were more ordered and obvious tracks and less large peaks and valleys on the upper surface. The height of the peak and the depth of the valley were obviously reduced. The surface roughness of the nine samples are shown in Table 3.   Under the input of the polishing laser energy, molten material was transported along the temperature gradient [30]. Due to surface tension and Marangoni force, the surface pore defects were dragged to the tail in the molten pool, connected with the previous solidified layer, and finally removed. The optical morphologies of the surfaces with/without polishing can be shown in Figure 10. After laser polishing, there were more ordered and obvious tracks and less large peaks and valleys on the upper surface. The height of the peak and the depth of the valley were obviously reduced. The surface roughness of the nine samples are shown in Table 3.    As shown in Figure 11, after polishing, the microstructure changed from a coarse columnar grain structure to fine equiaxed grain structure. The thermal undercooling caused by high cooling rates during laser polishing promoted the grain-refined effect [31]. Due to the Hall-Patch relationship, the finer grains caused by overlapping remelting can lead to the mechanical change.

Effect of Hatching Space
The SEM images of the laser polished samples with different hatching spaces are shown in Figure 12. The EDS spectrums of the white area of the SEM images were also depicted. It can be observed that after laser polishing, the upper surface was significantly smoothed. The powder particles on the upper surfaces of the samples were remelted during laser polishing. The orderly melted tracks, a few unmelted particles adhered to the upper surface and several scratches could be clearly observed. At the hatching space of 40 µm, there were less unmelted powders on the upper surface. That is because that the hatching space of 40 µm is smaller than the laser spot diameter, which leads to remelting in some overlapping areas on the upper surface. Therefore, the generation of unmelted particles on the surface can be largely reduced. As shown in Figure 11, after polishing, the microstructure changed from a coarse columnar grain structure to fine equiaxed grain structure. The thermal undercooling caused by high cooling rates during laser polishing promoted the grain-refined effect [31]. Due to the Hall-Patch relationship, the finer grains caused by overlapping remelting can lead to the mechanical change.

Effect of Hatching Space
The SEM images of the laser polished samples with different hatching spaces are shown in Figure 12. The EDS spectrums of the white area of the SEM images were also depicted. It can be observed that after laser polishing, the upper surface was significantly smoothed. The powder particles on the upper surfaces of the samples were remelted during laser polishing. The orderly melted tracks, a few unmelted particles adhered to the upper surface and several scratches could be clearly observed. At the hatching space of 40 μm, there were less unmelted powders on the upper surface. That is because that the hatching space of 40 μm is smaller than the laser spot diameter, which leads to remelting in some overlapping areas on the upper surface. Therefore, the generation of unmelted particles on the surface can be largely reduced. The surface roughness of the laser polished surfaces of the samples were evidently decreased compared with the as-built sample in Figure 13. The surface roughness of the Ra and Sa of the unpolished surface were 12.5 µm and 29.3 µm, respectively. Ra and Sa of the laser polished surfaces with different hatching spaces ranged from 7.2 µm to 9.4 µm and from 8.9 µm to 14.4 µm, respectively. At the hatching space of 100 µm, the unpolished area occurred due to the polishing tracks with gaps. Thus, the surface reduction was obviously less. It is the lower hatching space that can be chosen to reduce more surface roughness. Obeidi et al. [32] also found that increasing the overlapping area could improve the surface quality of AM 316L.
As shown in Figure 14, the use of laser polishing resulted in significantly higher microhardness of the polished surfaces. The microhardness of the unpolished surface was 112.3 HV. The microhardness of the polished surfaces with different hatching spaces of 40 µm, 70 µm and 100 µm were 147.2 HV, 134.9 HV and 130.4 HV, respectively. The increments were 31.1%, 20.1% and 16.2%, respectively. Ipbal et al. [33] found that the existence of the defects had a great negative effect on the microhardness of AM 316L. As discussed above, the effect of surface defects elimination due to laser polishing is an important factor for improving microhardness. In addition, the solid-solution strengthening was also a factor that could not be ignored. In Figure 12, elements distribution was presented by EDS analysis. The Si and Mg elements play great roles in AlSi10Mg alloy mechanical properties [34]. The presence of Si and Mg elements in the samples can lead to solid-solution strengthening, which has a great effect on the mechanical properties [35]. EDS spectrums showed there were more Si and Mg elements on the polished surface with the hatching space of 40 µm, which may enhance the solid-solution strengthening effect.

Effect of Polishing Pass
SEM images and EDS spectrums of laser polished samples with different polishing passes are shown in Figure 15. After laser polishing according to the three passes strategy, there were more ordered molten tracks on the upper surface. Fewer unmelted particles were seen on the upper surface. As can be seen in Figure 16, Ra and Sa of the polished surfaces with different polishing passes of 1, 2 and 3 were ranged from 8.1 µm to 3.7 µm and from 13. The surface roughness of the laser polished surfaces of the samples were evidently decreased compared with the as-built sample in Figure 13. The surface roughness of the Ra and Sa of the unpolished surface were 12.5 μm and 29.3 μm, respectively. Ra and Sa of the laser polished surfaces with different hatching spaces ranged from 7.2 μm to 9.4 μm and from 8.9 μm to 14.4 μm, respectively. At the hatching space of 100 μm, the unpolished area occurred due to the polishing tracks with gaps. Thus, the surface reduction was obviously less. It is the lower hatching space that can be chosen to reduce more surface Materials 2021, 14, x FOR PEER REVIEW roughness. Obeidi et al. [32] also found that increasing the overlapping area c prove the surface quality of AM 316L.  Figure 14, the use of laser polishing resulted in significantly hi crohardness of the polished surfaces. The microhardness of the unpolished sur 112.3 HV. The microhardness of the polished surfaces with different hatching spa μm, 70 μm and 100 μm were 147.2 HV, 134.9 HV and 130.4 HV, respectively. T ments were 31.1%, 20.1% and 16.2%, respectively. Ipbal et al. [33] found that the e of the defects had a great negative effect on the microhardness of AM 316L. As d above, the effect of surface defects elimination due to laser polishing is an importa for improving microhardness. In addition, the solid-solution strengthening was a tor that could not be ignored. In Figure 12, elements distribution was presented analysis. The Si and Mg elements play great roles in AlSi10Mg alloy mechanical p [34]. The presence of Si and Mg elements in the samples can lead to solid strengthening, which has a great effect on the mechanical properties [35]. EDS sp showed there were more Si and Mg elements on the polished surface with the space of 40 μm, which may enhance the solid-solution strengthening effect.

Effect of Polishing Pass
SEM images and EDS spectrums of laser polished samples with different p passes are shown in Figure 15. After laser polishing according to the three passes there were more ordered molten tracks on the upper surface. Fewer unmelted were seen on the upper surface. As can be seen in Figure 16, Ra and Sa of the surfaces with different polishing passes of 1, 2 and 3 were ranged from 8.1 μm t and from 13.2 μm to 8.4 μm, respectively.

Effect of Polishing Pass
SEM images and EDS spectrums of laser polished samples with different polishing passes are shown in Figure 15. After laser polishing according to the three passes strategy, there were more ordered molten tracks on the upper surface. Fewer unmelted particles were seen on the upper surface. As can be seen in Figure 16, Ra and Sa of the polished surfaces with different polishing passes of 1, 2 and 3 were ranged from 8.1 μm to 3.7 μm and from 13.2 μm to 8.4 μm, respectively.   As shown in Figure 17, the microhardness of the laser polished surfaces with differ ent polishing passes of 1, 2 and 3 were 134.9 HV, 164.3 HV and 176.9 HV, respectively The increments were 20.1%, 46.4% and 57.6%, respectively. The repeated grain-refine effect and solid-solution strengthening enhance the microhardness increment. The nex polishing laser can melt the remaining unmelted particles and rough surface, so tha As shown in Figure 17, the microhardness of the laser polished surfaces with different polishing passes of 1, 2 and 3 were 134.9 HV, 164.3 HV and 176.9 HV, respectively. The increments were 20.1%, 46.4% and 57.6%, respectively. The repeated grain-refined effect and solid-solution strengthening enhance the microhardness increment. The next polishing laser can melt the remaining unmelted particles and rough surface, so that smaller roughness can be guaranteed. The surface can be smoothed by sufficient wetting during repeated polishing. The polishing effect was significantly improved with the increase of polishing passes.  The SEM images and EDS spectrums of the laser polished samples with differen polishing directions are shown in Figure 18. It can be observed that the polishing directio is of great effect on the morphology of polished surfaces. There were fewer unmelted pa ticles on the polished surface using the D4 polishing direction. However, severe distorte tracks of the upper surface occurred because the wetting and diffusion process of th

Effect of Polishing Direction
The SEM images and EDS spectrums of the laser polished samples with different polishing directions are shown in Figure 18. It can be observed that the polishing direction is of great effect on the morphology of polished surfaces. There were fewer unmelted particles on the polished surface using the D4 polishing direction. However, severe distorted tracks of the upper surface occurred because the wetting and diffusion process of the remolten area were affected by the remaining manufacturing laser tracks.

Effect of Polishing Direction
The SEM images and EDS spectrums of the laser polished samples with different polishing directions are shown in Figure 18. It can be observed that the polishing direction is of great effect on the morphology of polished surfaces. There were fewer unmelted particles on the polished surface using the D4 polishing direction. However, severe distorted tracks of the upper surface occurred because the wetting and diffusion process of the remolten area were affected by the remaining manufacturing laser tracks. As can be observed in Figure 19, Ra and Sa of the laser polished surfaces with different polishing directions of D1, D2, D3 and D4 ranged from 8.8 μm to 4.1 μm and from 13.2 μm to 9.1 μm, respectively. In Figure 20, the microhardness of the laser polished surfaces with different polishing directions of D1, D2, D3 and D4 were 134.9 HV, 125.5 HV, 151.9 HV and 138.3 HV, respectively. The microhardness increments were 20.1%, 11.8%, 35.3% and 23.2%, respectively. At the laser polishing direction of D3, the polishing time is a little longer so that the more sufficient wetting and diffusion could be ensured. Thus, lower As can be observed in Figure 19, Ra and Sa of the laser polished surfaces with different polishing directions of D1, D2, D3 and D4 ranged from 8.8 µm to 4.1 µm and from 13.2 µm to 9.1 µm, respectively. In Figure 20, the microhardness of the laser polished surfaces with different polishing directions of D1, D2, D3 and D4 were 134.9 HV, 125.5 HV, 151.9 HV and 138.3 HV, respectively. The microhardness increments were 20.1%, 11.8%, 35.3% and 23.2%, respectively. At the laser polishing direction of D3, the polishing time is a little longer so that the more sufficient wetting and diffusion could be ensured. Thus, lower surface roughness can be obtained. As can be observed in Figure 19, Ra and Sa of the laser polished surfaces with differ ent polishing directions of D1, D2, D3 and D4 ranged from 8.8 μm to 4.1 μm and from 13. μm to 9.1 μm, respectively. In Figure 20, the microhardness of the laser polished surface with different polishing directions of D1, D2, D3 and D4 were 134.9 HV, 125.5 HV, 151. HV and 138.3 HV, respectively. The microhardness increments were 20.1%, 11.8%, 35.3% and 23.2%, respectively. At the laser polishing direction of D3, the polishing time is a littl longer so that the more sufficient wetting and diffusion could be ensured. Thus, lowe surface roughness can be obtained.

Conclusions
In this paper, the effects of in-situ laser polishing on L-PBF AlSi10Mg alloy usin different polishing strategies were investigated. Further, the effects of laser polishin strategies on the other mechanical properties of the polished samples, including tensil property, creep performance and fatigue resistance, will be studied in our followin

Conclusions
In this paper, the effects of in-situ laser polishing on L-PBF AlSi10Mg alloy using different polishing strategies were investigated. Further, the effects of laser polishing strategies on the other mechanical properties of the polished samples, including tensile property, creep performance and fatigue resistance, will be studied in our following study. The main conclusions can be summarized as follows: (1) The laser polishing relies on the mass transfer caused by melting and re-solidification in the laser polishing process. The mass at the crests will move into the troughs due to sufficient wetting and diffusion of the molten pool. Thus, the rough surface can be significantly smoothed. (2) The results of the experiment found clear support for the obvious effect of laser polishing. The surface roughness of the polished upper surface was greatly reduced and the microhardness was significantly increased after laser polishing. The surface roughness of Ra and Sa could be decreased by 70.4% and 71.3%, respectively. The microhardness could be increased by 57.6% after polishing. (3) The hatching space, polishing pass and direction play great roles in the laser polishing effect. An appropriate polishing strategy can help to attain lower surface roughness, more ordered tracks and greater mechanical properties.