A novel approach of jet polishing for interior surface of small-grooved components using three developed setups

It is a challenge to polish the interior surface of an additively manufactured component with complex structures and groove sizes less than 1 mm. Traditional polishing methods are disabled to polish the component, meanwhile keeping the structure intact. To overcome this challenge, small-grooved components made of aluminum alloy with sizes less than 1 mm were fabricated by a custom-made printer. A novel approach to multi-phase jet (MPJ) polishing is proposed, utilizing a self-developed polisher that incorporates solid, liquid, and gas phases. In contrast, abrasive air jet (AAJ) polishing is recommended, employing a customized polisher that combines solid and gas phases. After jet polishing, surface roughness (Sa) on the interior surface of grooves decreases from pristine 8.596 μm to 0.701 μm and 0.336 μm via AAJ polishing and MPJ polishing, respectively, and Sa reduces 92% and 96%, correspondingly. Furthermore, a formula defining the relationship between linear energy density and unit defect volume has been developed. The optimized parameters in additive manufacturing are that linear energy density varies from 0.135 J mm−1 to 0.22 J mm−1. The unit area defect volume achieved via the optimized parameters decreases to 1/12 of that achieved via non-optimized ones. Computational fluid dynamics simulation results reveal that material is removed by shear stress, and the alumina abrasives experience multiple collisions with the defects on the heat pipe groove, resulting in uniform material removal. This is in good agreement with the experimental results. The novel proposed setups, approach, and findings provide new insights into manufacturing complex-structured components, polishing the small-grooved structure, and keeping it unbroken.

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
High-performance heat exchangers with new spatial designs are now possible due to the rapid development of additive manufacturing (AM) [1][2][3].Among these, axially grooved heat pipes (AGHPs) are commonly used as heat-exchange devices in iso-thermalized radiator panels, which efficiently conduct heat from electronics to panels in satellite platforms and can be fabricated using laser powder bed fusion (L-PBF) [4,5].This innovative approach allows for the simultaneous design and manufacture of heat pipes along with support seats, eliminating the occurrence of wrinkles or thin-walling phenomena in bending areas, which are common in conventional drawn heat pipes [6][7][8].As a result, AM offers noteworthy advantages in the production of AGHP, thereby enhancing thermal conductivity and reliability.
However, L-PBF products present several inherent issues, including an inadequate surface roughness (Sa), which constrain their potential for widespread application in space environments [9].Meanwhile, satellite, balling, and stalactite defects considerably increase the Sa of components [10][11][12].When the build angle is below 90 • , the morphology of the overhanging surface is dramatically degraded due to the 'stair-step' effect [13][14][15].Especially for the down-skin, the Sa increases and the shape accuracy decreases as the build angle decreases.Certain defects are close to the width of the AHGP micro-groove, resulting in working medium blockage and increasing the reflux resistance, thus further reducing the heat-transfer coefficients [8,16].Furthermore, satellites and balling could potentially fall off in space, impairing the formation of boundary layers and leading to blockage, threatening the stability and durability of the structure.Therefore, some post-processing is required to improve the surface quality for L-PBF fabricated AGHPs.
Additionally, the optimization of L-PBF parameters plays a pivotal role in Sa, extending beyond a mere dependence on post-processing.The ultimate surface quality after postprocessing relies not only on the initial AM Sa but also on the defect types.The as-built composite side surface exhibits a lower surface quality than the top surface for defects resulting from (i) instability of the molten pool forming the curved edges, (ii) loosely adhered/partially melted particles, and (iii) gap-lapping layers [17][18][19][20].The side-surface quality is controlled via contour laser power, scanning speed, and layer thickness.Under optimal process factors, the linear surface roughness (Ra) is typically in the range of 5-15 µm [10,11,[21][22][23].To date, most studies have focused on the singletrack or regularly shaped samples.However, few studies have examined the Sa of complex structures such as spiral channels and micro-grooves.Meanwhile, a comprehensive quantitative analysis of the relationship between the L-PBF parameters and surface defects (i.e.size, volume, and distribution) is remarkably required.
L-PBF parts with high Sa are usually smoothed via postprocessing, including milling, shape-adaptive grinding, or polishing [24][25][26].Milling and shape-adaptive grinding are frequently chosen for processing the external surface; however, their application is constrained by the intricate shape of the part.Electropolishing, electrochemical-mechanical polishing (ECMP) [27][28][29][30], magnetic abrasive finishing (MAF) [31][32][33], multi-jet hydrodynamic finishing (MJHF) [34], and abrasive flow finishing (AFF) [35,36] are usually applied to polish and smooth interior channels.Zhao et al [29] and An et al [30] proposed to polish the L-PBF fabricated channel with different diameters via an ECMP process.As a result, a Sa of 5.06 µm was achieved for a 30 mm diameter 316L stainless steel channel, while that of a 3 mm diameter 304 stainless steel channel reached 3.88 µm.However, it is extremely difficult for the cathode fabrication used in electropolishing and ECMP to match complex, spatially curved, and microgrooved narrow channels [27].In MAF, the surface material is removed by the combined action of magnetic abrasive particles (provided by magnetic pole feeding) and workpiece rotation [31,33].Guo et al [31] proposed a vibrating MAF method to improve the Sa of a double-layer channel, achieving a final Sa of less than 1 µm.The removal mechanism of MAF makes it applicable to rotary bodies but not to micro-grooved channels.MJHF has been used to improve the surface quality of 1-5 mm diameter Inconel 625 channels, obtaining an improvement of 60%-90% (final Ra, Sa ⩽ 1 µm) [34].Basha et al [35] used AFF to polish copper tube sections, achieving a maximum Sa reduction as high as 90%.However, the MJHF and AFF devices are relatively complex and not suitable for thin-walled and complex-shaped tubes.In summary, no effective and cost-effective approach has been developed to date for polishing thin-walled axial grooves.
Recent studies have employed abrasive air jet (AAJ) and multi-phase jet (MPJ) polishing to improve the Sa of rigid and brittle materials [36][37][38][39].Wang et al [36,37] optimized AAJ polishing parameters by utilizing a low jet pressure, small Q Gu et al jet angle, and small-sized abrasives.This optimization led to a minimum Sa of 0.244 µm while polishing quartz glass.Similarly, in MPJ polishing experiments on Si surfaces, a minimum Sa of 0.2 µm was achieved using SiC abrasives at a jet angle of 70 • [38].Aluminum alloy components produced via laser powder bed fusion (L-PBF) exhibit a hardness of approximately 127 HV, which is lower than that of quartz and Si.Consequently, these two jet polishing methods exhibit substantial potential in enhancing the surface quality of AM parts.Notably, most previous studies have focused on flat surfaces.However, the material removal behavior of AAJ and MPJ polishing in curved channels remains unexplored.In contrast to flat surfaces, where abrasive particles deflect upon impact, an intricate inner geometry might result in repeated acceleration and multiple bounces.Consequently, investigating the material removal mechanism of abrasives on the channel surface is essential.Moreover, the equipment used for AAJ and MPJ polishing is simple and cost-effective, as opposed to that used in the aforementioned polishing methods.The required pressure is typically below 0.8 MPa, avoiding the need for a sizable compressor to supply power.These two processes can also reuse the abrasive particles, thus promoting an environmentally sustainable post-processing methodology.
In this study, the AGHPs were fabricated via L-PBF under different parameters and build angles.Then, the interior channel defects of the fabricated AGHPs were quantitatively analyzed and evaluated to reveal the effect of different laser parameters and build angles on L-PBF quality.Additionally, the specially designed and engineered 2D and 3D AGHPs were printed to study the effectiveness of AAJ and MPJ polishing in terms of defect removal in AGHPs.Finally, computational fluid dynamics (CFD) simulations of the AAJ polishing process were conducted to reveal the material removal mechanisms and assess the feasibility and uniformity of the 2D and 3D AGHPs.

Sample preparation
Figure 1 shows the L-PBF machine, printing material, crosssectional view of the AGHP, scanning strategy, and printing scheme for the AGHP sample.The L-PBF system used to fabricate the AlSi10Mg AGHPs was an independently developed 3D printing system.A Gaussian profile continuous wave fiber laser (YLR-500 fiber laser, IPG, Germany) with a 1070 nm wavelength and 500 W maximum laser power was selected.A scanning electron microscopy (SEM; JSM-IT800SHL, JEOL, Japan) image of the gas-atomized AlSi10Mg powders (Vilory Advanced Materials, China) used for the L-PBF process is shown in figure 1(b).The powder particle size distribution was measured using a laser particle size analyzer (Mastersizer 2000, UK).The powder diameter size range was 15-53 µm (the median diameter D 50 = 36.2µm).The AGHP sample heights for L-PBF parameter optimization and polishing were 15 mm and 75 mm, respectively.The cross-sectional shape is shown in figure 1(d).The groove height and thickness were 0.81 mm and 0.45 mm, respectively, and the distance between adjacent grooves was 0.22 mm.The grooves were arranged at a centrosymmetric mode in a plane, with the normal axis being a sinusoid or helix, forming an astronautic 2D or 3D AlSi10Mg heat pipe.The L-PBF processing parameters and conditions are listed in table 1.A meander scanning strategy was adopted to improve the mechanical properties and manufacturing efficiency of the AGHPs (figure 1(e)).After L-PBF, all samples were observed via SEM.ImageJ, an image processing software, was used to measure the defect sizes and quantities.

AAJ and MPJ polishing methods
For both the AAJ and MPJ polishing processes, #400 Al 2 O 3 abrasive was selected; its SEM image and particle size distribution are shown in figures 2(c) and (f), respectively.In both methods, compressed air was provided by a small air compressor, with the pressure being set at 0.4 MPa.During AAJ polishing, the abrasive was mixed with compressed air inside the mixing chamber.Then, it was ejected through the nozzle to form an abrasive jet for polishing the AGHPs (figure 2(b)).For MPJ polishing, Al 2 O 3 abrasive was mixed with water with a concentration of 10 wt%.Then, the slurry was sucked into the mixing chamber by the compressed air.Inside the mixing chamber, the slurry was atomized via compressed air, forming a MPJ.Subsequently, the MPJ was further accelerated and dispersed through the connected nozzle before entering the AGHP (figure 2(e)).During MPJ polishing, a high-speed camera (v2512, Phantom, USA) was used to capture the slurry jet at the outlet.The AGHPs were polished via AAJ or MPJ for 50 min.At 10 min intervals, the AGHPs were removed from the nozzle and cleaned with ethanol by an ultrasonic cleaner before weighting.The removed material weight was recorded until the AGHP weight became stable.
The as-printed and post-processed AGHPs were cut by a precision diamond-wire cutting machine before measurement.The interior micro-groove surface defects and Sa were measured via SEM and 3D optical surface profilometry (Zygo NewView 9000, USA).

CFD simulation setup
The CFD models were established in Ansys-Fluent to investigate the interactions between the abrasives and intricate AGHPs, as well as to compare the material removal uniformity in the 2D and 3D AGHPs after AAJ polishing.Only the fluid flow area was modeled, while a 20 mm entry component was set as the nozzle for the fluid flow to stabilize.The inlet and outlet surfaces were set as static pressure boundaries, with the pressures being 201 kPa and 101 kPa, respectively.Two-thousand abrasives with size of 15-120 µm were used as the solid phase, which was simulated by a discrete phase model.The boundary conditions and governing equations can be found in [36][37][38][39][40][41][42].A MATLAB script was employed to analyze the simulation results for the impact location, frequency, velocity, and velocity loss of the abrasives.

Influence of contour parameters on as-fabricated AGHP surface defects
The contour process parameters have a much greater influence on the Sa than the infill ones [13,43].Therefore, only the former is examined in this section.Figure 3 depicts the morphologies and Ra of the outer and inner groove surfaces.Defects such as satellites and balling are more frequently observed on the inner surface; hence, the inner Ra (8.251 µm) is larger than the outer Ra (5.968 µm).For the narrow-groove structure, the heat from the opposite surface of the groove could potentially result in a higher printing temperature, leading to more powders sticking to the inner surface.Therefore, this study mainly focused on the formation and elimination of inner surface defects.
The morphologies of the inner grooves under a laser power of 95 W, 135 W and 175 W with a scanning speed of 900 mm s −1 are shown in figure 4.There are two types of defects on the AGHP inner groove surface.Balling, larger than 50 µm, is formed by the fusion of several particles or an unstable melting pool.Most defects smaller than 50 µm are satellites, which are partially melted powders or spatters.A distinction between satellites and balling is established using 50 µm as a threshold.This criterion is chosen considering the particle size range of the L-PBF printing powder, which falls within 15-53 µm.As a result, satellites primarily consist of individual powder particles.As the laser power increases, the number of balling decreases rapidly, while the satellites continue to occur for all samples.
Based on our prior linear AGHP jet polishing experiments, all satellites were entirely removed, while balling could only be partially eliminated.Consequently, the correlation between defect size distribution and process parameters during L-PBF was quantitatively analyzed.Defect measurements were performed on the SEM images, and the defect size distributions according to laser power and scanning speed are presented in figures 5 and 6, respectively.As the laser power increases, the total number of defects decreases; at a laser power of 95 W, the defect count is nearly twice as high as that at 155 W, as shown in figures 5 and S1.The quantity of defects exceeding 50 µm in size also decreases with an increase in laser power, and such defects mainly consists of balling, which occurs as its size surpasses most AlSi10Mg powders.Notably, a significant decrease in defects larger than 100 µm occurs when the laser power is equal to or exceeds 135 W. Therefore, for a sufficient surface quality improvement, the minimum laser power should be at least 135 W when printing AGHPs.
The relationship between the number of defects and the laser scanning speed at the 175 W laser power is shown in figures 6 and S2.The total number of defects initially increases and then decreases as the scanning speed increases.However, the number of defects with sizes ⩾ 50 µm first decreases and then increases as the scanning speed increases.Notably, for laser scanning speeds between 900 and 1300 mm s −1 , both the total number of defects count and defects with sizes ⩾ 50 µm are at their minimum, with no defects exceeding 100 µm.Consequently, this scanning speed range is considered optimal for AGHP AM.
Figure 7 displays the morphologies of all AGHP internal groove surfaces, which are classified into three categories according to defects.The quantities and sizes of balling and satellites vary according to both laser power and scanning speed.In figure 7, the panels in blue boxes represent groups with numerous balling and satellite defects.It is evident that the number of defects decreases initially as the scanning speed decreases or the laser power increases.Additionally, the panels in green boxes represent the groups with the best surface quality, showing almost no large balling and only small satellites.However, as the laser power increases and the scanning speed decreases, the Sa becomes high again, and balling appears, as shown in the panels in red color boxes.
In previous studies, laser energy density (LED) was used to illustrate the interaction between laser power and scanning speed, which can be calculated as follows [44]: where P c is the laser power, and s c is the laser scanning speed.Accordingly, except in figures 7(a)-(i) and (b)-(i), the above mentioned three categories can also be classified according to LED.The blue panels represent E L < 0.135 J mm −1 and the green panels represent 0.135 J mm −1 < E L < 0.22 J mm −1 , while the remaining two red panels represent E L > 0.22 J mm −1 .Most spattered satellites and balling defects tend to spherically agglomerate at the smallest Gibbs surface free energy.Therefore, the defect volume can easily be calculated through the volume formula for spheres when the defect size distribution is obtained.It can easily be deduced that a 100 µm  defect is eight times larger than a 50 µm defect and 37 times larger than a 30 µm one.Therefore, the volume of the larger defects that need to be removed is much higher.Moreover, small defects of ∼10-50 µm in size are primarily satellites that are moderately sintered to the surface, making their removal much easier.
Large balling is much harder to be removed in postprocessing.The defect volume of samples fabricated with different laser powers is shown in figure 8(a), as derived from the defect size results.Owing to the increased occurrence of balling and satellites, the total defect volume requiring removal under a 95 W laser power is 12 times larger than that under a 135 W laser power.An empirical relationship between the unit defect volume and LED can be obtained as follows: where V D is the unit defect volume and E L is the LED.Equation (2) reveals that when the LED exceeds 0.135 J mm −1 , V D decreases, which is conducive to post-processing across a larger range.To the best of our  knowledge, this is the first quantitative investigation into defect size distribution, as well as the first attempt to examine defect volume from a post-processing perspective.Identical criteria will be employed in the following section to assess the impact of different build angles.

Influence of build angle on the interior surface morphology
The build angle affects the ultimate forming capacity of complex spatial AGHPs.In this section, the LED, laser power,   and scanning speed are selected as 0.135 J mm −1 , 175 W, and 900 mm s −1 , respectively, which are the optimal parameters obtained as described in the previous section.As shown in figures 9 and 10, there is a notable increase in defects as the build angle decreases, which increases the difficulty for post-processing.When the build angle reaches 70 • , not only does the number of defects increase, but the defect size also increases.New 'stalactite' defects are formed, whose sizes are much larger than those of normal balling hanging on the down skin.Such defects are found to occasionally connect to the surface of groove walls, as shown in figures 9(d)-(f), which completely block the ∼130 µm narrow wick groove.Figure 10 reveals an increase in the total number of defects when the build angle changes from 85 • to 75 • .Then, despite the number of defects remaining constant or decreasing as the build angle continues to decrease, the number of large defects, especially those larger than 100 µm, increases from only a few defects to dozens.For AGHPs, such quantity of large defects is not appreciated as dozens of large defects block the groove, hindering the fluid flow.Consequently, it is recommended that the build angle for AGHPs should exceed 75 • and not be less than 60 • .
The stalactite formation primarily occurs because a substantial amount of powder melts without solid support, as depicted in figure 11.A portion of the molten pool comes into contact with the previously solidified section, while the rest contacts the loose powder.The portion in contact with the loose powder is defined by the overhang length (L), which is contingent upon the build angle (θ).The relationship between L and θ is expressed as follows: where t is the layer thickness.During solidification, the surrounding powders are constantly absorbed by the molten pool, leading to the formation of large stalactites on the groove down-skin (figure 11).These stalactites, which adhere to the overhanging surface, impair the down-skin Sa, and compromise the groove shape accuracy.

Performance of MPJ and AAJ polishing for 2D AGHPs
In numerous engineering applications, high-performance thermal systems are designed as an 'S' shape in the plane and as helixes in space.Owing to their complex shapes, AGHPs can increase the heat transfer rate via (i) an increase in the heattransfer area and (ii) generating swirling or secondary flows or both [45,46].To assess the efficiency of jet polishing for defect removal from the interior channels of 2D and 3D AGHPs, several AGHPs were fabricated via L-PBF (figure 12(a)) and subsequently polished.Two types of centerline tracks for the 2D AGHPs were defined according to the following functions: where x is the centerline track abscissa and z is the AGHP printing height.
To briefly discuss these AGHPs, we labeled them as 2.5-S and 5-S, respectively.Both exhibited axisymmetric structures (relative to their centerline tracks) about z = 37.5 mm.To facilitate the nozzle installation and compare the jet polishing results, an additional 5 mm of material was printed at the top and bottom of each AGHP for a total height of 75 mm, as shown in figure 13(a).
The 2D 2.5-S and 5-S AGHPs exhibit minimum build angles of 76.4 • and 64.2 • , respectively.As discussed in section 3.2, these build angles are considered acceptable for L-PBF surface quality.The laser parameters employed for the contour printing of these AGHPs are optimized.The positions near z = 5 mm, 21.25 mm, 37.5 mm, 53.75 mm, and 70 mm of the as-printed 2.5-S AGHP were analyzed, as shown in figures 12(b)-(f).Most of the defects are satellites.Defects are not only observed on the top surface of the groove but also randomly distributed on the sidewall and bottom.However, in the inclined section, large balling and stalactite defects are observed on the down-skin of the groove.Some of these defects become connected with those on the opposite side, creating a small bridge that partially obstructs the groove, as shown in figure 12(c).The defects, particularly large-sized balling and satellites, are even more severe in the 5-S AGHP, owing to the lower minimum build angle, as shown in figure S3.
The AAJ polishing results test is shown in figures 13 and S4, which indicates the efficient removal of almost all satellite, balling, and stalactite defects from the surface.The shape accuracy of the interior groove is relatively high.However, in the inclined area, stalactite remnants are still observed, as shown in figure 13(c).A comparison between the surface morphologies in positions III, and V (nearly perpendicular) suggests an increase in Sa as the polishing distance increases, resulting in a more noticeable surface waviness.Moreover, the quantity of residual large-sized balling components is found to increase as the polishing distance increases.The polishing performance degradation with respect to distance is attributable to the interior fluid resistance, which reduces the material removal rate of AAJ polishing at the outlet.Consequently, surface ripples formed during printing could not be effectively removed.
The defective material removal process in AAJ polishing is depicted in figure 14.The compressed air disperses the abrasive particles in the container.The constant airflow created within the AGHP by the compressed air accelerates the dispersed abrasive particles through the action of drag forces.Subsequently, high-velocity abrasive particles impact the defect, leading to the shearing and removal of material.The windward side of the defect experiences a continuous impact from abrasive particles, progressively removing the material.However, material on the leeward side of the defect remains on the groove surface, as illustrated in figure 13, where the defect is oriented towards the AAJ outlet.Although both methods effectively eliminated the major defects at the top and bottom of the groove, MPJ polishing resulted in a smoother surface and improved shape accuracy, as shown in figures 15 and S5.Moreover, the outlet surface of the MPJ-polished sample does not exhibit a prominent waviness.When comparing the groove wall thickness measurements at positions I-V, the groove wall thickness achieved via MPJ polishing is smaller than that after AAJ polishing, as shown in figures 16(b) and (c); both are smaller than the as-printed surface.The reduced groove wall thickness via MPJ polishing is attributed to the higher material removal rate, as shown in figure 16(a).the groove wall thickness at position I is notably minimized for both polishing methods due to the turbulent jet at the inlet, despite our efforts to improve the jet stability nozzle adjustments.However, the surface quality of the outlet remained inferior compared to that of the inlet, indicating a constant polishing performance degradation under MPJ polishing.Under both polishing methods, the weight decreases rapidly in the first 20 min and then the rate of change decreases.This phenomenon can be attributed to the presence of satellites and some balling adhering to the groove surface, making them easily removable.However, the larger-sized balling and stalactites are partially melted and integrated into the substructure.These types of defects can only be gradually removed by accelerated abrasives.The jet polishing results also contribute to the quantitative defect size distribution analysis presented in sections 3.1 and 3.2.Notably, the removal rate for 5-S exceeded that for 2.5-S, as more defects are presented on the internal groove surface of the former.However, the width of the protrusions after both polishing methods is nearly identical.
The magnified SEM images and Sa of the 2.5-S sample at the inlet and outlet positions are shown in figure 17.The surface damage produced by MPJ polishing is less than that produced by AAJ polishing at the inlet position.For both polished surfaces, the small-angle scratching marks were identified as denticles.However, the sizes and depths of these scratch marks are much larger under AAJ polishing.Consequently, the Sa at the inlet position decreased from 8.596 µm to 0.701 µm and 0.336 µm under AAJ and MPJ polishing, respectively.At the outlet position, the Sa under AAJ polishing (1.161 µm) was still larger than that under MPJ polishing (0.845 µm) due to more residual defects.
It is noteworthy that under MPJ polishing, abrasive particles are first dispersed in water and subsequently atomized using compressed air.A mist jet containing thousands of slurry droplets (size: 0.01-1 mm) is used for polishing [38,42].Because the abrasive particles are dispersed within the slurry and only 15-120 µm in size, a single droplet may contain a few or even dozens of abrasive particles, as illustrated in figure 18.This is analogous to enlarging the effective diameter of the polishing particles.Furthermore, the water flow within MPJ prevents the dispersion of abrasive particles, facilitating an effective jet beam convergence.Consequently, the droplet forms a larger polishing median, exhibiting higher kinetic energy than the individual abrasive particles presented in AAJ polishing.This clustering effect enhances the material removal rate under MPJ polishing, producing a much cleaner polished interior.
Similar to AAJ polishing, in the MPJ polishing process, the slurry droplets impact the windward side of the defect, exerting a shear force that removes the defects.However, the material on the leeward side persists on the groove surface, as illustrated in figure 18(b).The distinguishing factor is that MPJ polishing employs water within the jet as a buffer layer, thereby decreasing the impact depth and the subsequent volume of material removed by each abrasive.

Performance of MPJ and AAJ polishing for 3D AGHPs
2D AGHPs are frequently used for heat dissipation in electronic devices.In contrast, 3D AGHPs are more prevalent in high-performance space components, and they exhibit more intricate architectures and more defects after L-PBF [44,45].The two key types of main centerline track for 3D AGHPs are helical and expressed as follows: where x is the centerline track abscissa, y is the centerline track ordinate, z is AGHP printing height, and α is an equation parameter.
In this case, the AGHPs are denoted as 2.5-H and 5-H for brevity.Notably, the build angles of the 3D AGHPs lower than those of 2D counterparts (71.13 • and 55.65 • , respectively).Owing to the lower build angles, the size and numbers of defects within the 3D channel are more severe than those presented in the 2D AGHPs, as shown in figures 19(a-ii)-(aiv) and S6(a-i)-(a-iii).The morphologies of the 3D AGHPs deteriorate due to the presence of overhung surfaces (in both the x-and y-directions) as z increases.Therefore, the tops, bottoms, and walls of the micro-groove surfaces are regarded as overlap surfaces in 3D AGHPs (in contrast, only the groove walls constitute an overlap surface in 2D AGHPs).
The outlet morphologies of the AAJ-and MPJ-polished 2.5-H and 5-H AGHPs are shown in figures 19 and S6.The residual defects are relatively small under both methods for the 2.5-H samples, with only a few stalactites remaining, as marked by the white circles.However, the Sa of the 5-H AGHPs is considerably higher, as considerable stalactite material is left after both AAJ and MPJ polishing, despite the fact that the weight of the removed material for the 5-H AGHPs is larger.The rough surface quality of the 5-H AGHPs is attributed to the lower build angle, which is lower than the limit of build angle (60 • ) discussed in section 3.2.
However, as shown in figure 20(a), the total material weight removed via AAJ polishing is inferior to that removed via MPJ polishing.The higher removal rate of MPJ polishing compared to that of AAJ polishing has been analyzed, as discussed in section 3.3.Moreover, a comparison between the results presented in figures 16(a) and (b) reveals that the removed material weight of the 3D AGHPs exceeded that of the 2D AGHPs under both AAJ and MPJ polishing.This can be attributed to the fact that 3D AGHPs have more defects due to their poorer surface quality and longer centerline lengths.The groove wall thickness remains relatively consistent along the polishing direction.Pressure loss significantly impacts the Sa while exerting a comparatively lesser effect on structural size.Additionally, the groove wall thickness is slightly reduced after MPJ polishing compared to AAJ polishing.In summary, MPJ polishing is deemed favorable for both 2D and 3D heat pipes, owing to its superior removal rate and smooth surface quality.Nevertheless, these two jet polishing methods can be extensively employed in other AM components featuring complex spatial channels.

CFD simulations of the AAJ polishing process for the 2D and 3D AGHPs
Although the experimental results demonstrated the effectiveness of both AAJ and MPJ polishing for L-PBF defect removal in AGHPs, the removal mechanisms within the complex structure of 2D and 3D AGHPs, as well as the material removal uniformity along the axial direction of the heat pipes are uncertain.Specifically, there are three key aspects to consider: (i) air resistance causes a gradual decline in pressure along the polishing direction; (ii) the impact velocity of abrasives is affected by the loss of kinetic energy; (iii) the impact frequency of jet polishing is generally believed to be higher at small curvature turns.It is unclear whether the bending rate of 2D and 3D AGHPs influences the impact frequency.To address these issues, CFD simulations of the AAJ polishing process were conducted.Only the AAJ polishing process in AGHPs was simulated because MPJ polishing involves three phases: gas, liquid, and solid.The interaction between the high-speed slurry and the solid surface creates a complex liquid-solid interface.Compression of the liquid results in the formation of a pressure zone at the interface.Moreover, the compression generates shock waves in both the slurry droplets and solid material, possibly causing the abrasive in the slurry to rebound off the liquid droplet.Currently, simulations are confined to predicting flow fields and particle trajectories upon jet impact and replicating the intricate processing of MPJ polishing within a specific model remains challenging [38,42].
The CFD simulation results for the AAJ polishing process reveal a gradual and uniform decrease of internal pressure in both the 2D and 3D AGHPs along the polishing direction, as shown in figures 21 and S7.Consequently, as shown in figure 17, following AAJ polishing, the Sa of the 2.5-S AGHP at the outlet is larger than that at the inlet, with values of 1.161 µm and 0.701 µm, respectively.It is worth noting that a sudden pressure shift in pressure occurs at the junction of the nozzle and the AGHP inlet, suggesting an efficient abrasive particle dispersion and abrasive jet stabilization by the nozzle.Surprisingly, the velocity of the abrasive particles remains relatively consistent along the polishing direction, despite varithe abrasive particle velocity at different locations, as shown in figures 21(b-i) and (b-ii).The reduction in abrasive particle velocity is more pronounced after sharp turns within the AGHP.Given that the bending curvature of the 3D AGHP is uniform across its structure, the variation in abrasive particle velocity at the inner bend of the 2D AGHP is more substantial than that of the 3D AGHP.An examination of the trajectory and velocity of individual abrasive particles revealed that these particles undergo multiple impacts within the AGHP.Upon impact with a defect, a solitary abrasive particle experiences a rapid decrease in velocity, resulting in the removal of a portion of the defects.Nevertheless, the abrasive particles undergo further acceleration due to the compressed air within the AGHP, subsequently restoring their kinetic energy and persistently eradicating defects from the groove surfaces.Consequently, this phenomenon elucidates that defects located at the AGHP outlet also experience impacts of abrasive particles, resulting in material removal.Furthermore, our previous research findings propose a preferential removal of defects from the groove surface [47].
As opposed to the 2D AGHP, the 3D AGHP subjects the same quantity of abrasive particles to a higher impaction frequency, as shown in figure 22.More specifically, the total impact count for the 3D 5-H AGHP was almost twice that of the 2D 5-S AGHP, resulting in a higher removal rate.Furthermore, the impact location distribution in the 3D AGHPs is more uniform compared to that of 2D AGHPs, as more abrasives impact upon the turning area of 2.5-S and 5-S.Therefore, despite the higher defect occurrence on the asprinted 3D AGHPs, the AAJ polishing process yields a relatively smooth grooved surface.
The impact velocity and impact velocity losses were assessed and tabulated using MATLAB, as shown in figures 23 and S8.The impact energies are associated with the abrasive mass via the kinetic energy equation, E = 1 2 mv 2 .A higher kinetic energy for the abrasive implies more efficient defects removal.The mean impact velocity of 2.5-S and 2.5-H is almost identical (24.2 m s −1 and 23.9 m s −1 , respectively).Conversely, the mean impact velocity of 5-S is slightly higher than that of 5-H (22.4 m s −1 and 19.8 m s −1 , respectively).Therefore, considering both the impact frequency and impact velocity, AAJ polishing method is found to be more effective for defect removal from 3D AGHPs.

Conclusion
In this study, defects in AGHPs fabricated via L-PBF were quantitatively studied and their formation mechanism was revealed.Two jet polishing methods, namely, AAJ and MPJ polishing, were proposed for post-processing AGHPs with 2D and 3D spatial complex structures.Based on the results, the main conclusions of this study can be summarized as follows: (1) The contour LED plays an important role in defect morphology and quality, with the optimal LED range being 0.135-0.22J mm −1 .A build angle greater than 75 • is recommended.Additionally, the non-linear relationship between the defect total volume and the L-PBF LED was revealed and can be fitted by a quartic function.
(2) AAJ and MPJ polishing are found to effectively remove all satellites and most balling/stalactite defects in 2D AGHPs.MPJ polishing technique is observed to achieve a higher material removal rate, as multiple abrasives cluster inside one droplet.Furthermore, the inlet area Sa after AAJ and MPJ polishing was 0.701 µm and 0.336 µm, respectively.The low Sa after MPJ polishing is attributed to the water buffering effect.(3) The material removal rate for 3D AGHPs is found to exceed that for 2D AGHPs under both AAJ and MPJ polishing.However, more residual defects are observed in the former due to their poorer raw surfaces.The higher and more uniform impact times observed in the AAJ polishing CFD simulation results for 3D AGHPs indicated their potential use for polishing complex spatial channels.
The methods and findings of this study provide some insights into the defects found in AGHP interior channels while proposing efficient jet methods to remove such defects.Our findings are expected to aid in the optimization of L-PBF parameters for AGHP manufacturing, as well as in the development of advanced surface finishing methods.Moreover, this study on L-PBF and jet polishing provides a comprehensive approach toward the fabrication of complex spatial AGHPs, thereby expanding their potential application.

Figure 1 .
Figure 1.L-PBF sample preparation.(a) L-PBF machine.(b) SEM image and (c) particle size distribution of the AlSi10Mg powders.(d) Cross-sectional view and dimensions of the AGHP.(e) AGHP meander scanning strategy.(f) Schematic of the L-PBF process.

Figure 2 .
Figure 2. Jet polishing process.(a) Device picture and (b) schematic diagram of the AAJ polishing.(c) SEM image and (f) particle size distribution of the Al 2 O 3 abrasive particles.(d) Device picture and (e) schematic diagram of the MPJ polishing.

Figure 3 .
Figure 3. SEM images and Ra of the outer and inner AGHP surfaces.(a) SEM image and (b) Ra of the outer surface.(c) SEM image and (d) Ra of the inner surface.

Figure 7 .
Figure 7. SEM images of the AGHP inner grooves fabricated at various laser powers and scanning speeds.(a)-(e) AGHPs fabricated at different laser powers: (a) 95 W, (b) 115 W, (c) 135 W, (d) 155 W, and (e) 175 W. The panels are color-coded according to the three different stages: numerous balling and satellites (blue), zero balling (green), and a little balling (red).

Figure 8 .Figure 9 .
Figure 8. Statistical results and fit function of defect volume.(a) Statistical results of defect volume per unit area at different laser powers and (b) fit function relationship between unit defect volume and linear energy density.

Figure 11 .
Figure 11.Schematic of defect formation on the overhanging surface.

Figure 12 .
Figure 12.Optical picture and SEM images of as-printed 2.5-S AGHP samples.(a) Optical picture of as-printed 2.5-S AGHP.(b)-(f) Groove surface and cross-sectional SEM images at five different positions: (b) I, (c) II, (d) III, (e) IV, and (f) V.For details regarding the five positions, refer to figure 13(a).

Figure 13 .
Figure 13.Schematic and SEM images of the 2.5-S AGHP internal grooves.(a) Section of the 2.5-S AGHP and the five positions recorded via SEM: I (z = 5 mm), II (z = 21.25 mm), III (z = 37.5 mm), IV (z = 53.75mm), and V (z = 70 mm).(b)-(f) Groove surface and cross-sectional SEM images after AAJ polishing at five positions I-V, respectively.Yellow dashed lines and white dashed circles denote the shape accuracy and residual defects, respectively.

Figure 14 .
Figure 14.Schematic and removal mechanism of AAJ polishing.(a) Schematic and (b) removal mechanism.

Figure 15 .
Figure 15.Groove surface and cross-sectional SEM images of the 2.5-S AGHP after MPJ polishing.(a)-(e) SEM images for positions I-V, respectively, and (f) expanded view of (e).Yellow dashed lines and white dashed circles denote shape accuracy and residual defects, respectively.

Figure 16 .
Figure 16.Removed material weight and groove wall thickness the 2.5-S and 5-S AGHPs.(a) Removed material weight for the 2.5-S and 5-S AGHPs under AAJ and MPJ polishing.As-printed, AAJ-, and MPJ-polished groove wall thickness for the (b) 2.5-S and (c) 5-S AGHPs.

Figure 17 .
Figure 17.SEM and Sa at the inlet and outlet positions for the 2.5-S AGHP.(a) SEM images at the inlet position after AAJ polishing.(b) SEM images at the inlet position after MPJ polishing.(c) Sa of the as-printed, AAJ-, and MPJ-polished AGHPs at the inlet position, respectively.(d) SEM images at exit position after AAJ polishing.(e) SEM images at the outlet position after MPJ polishing.(f) Sa of the as-printed, AAJ-, and MPJ-polished AGHPs at the outlet position.

Figure 18 .
Figure 18.MPJ polishing process.(a) Schematic and (b) removal mechanism of MPJ polishing.(c) Slurry droplets captured via a high-speed camera.

Figure 20 .
Figure 20.Removed material weight and groove wall thickness for the 2.5-H and 5-H AGHPs.(a) Removed material weight for the 2.5-H and 5-H AGHPs under AAJ and MPJ polishing.As-printed, AAJ-, and MPJ-polished groove wall thickness for the (b) 2.5-H and (c) 5-H AGHPs.

Figure 21 .
Figure 21.CFD simulation results for the 2.5-S and 2.5-H AGHPs under AAJ polishing.(a) CFD simulation results for 2.5-S AGHP.(a-i) Static pressure changes in 2.5-S.(a-ii) and (a-iii) Tracks and velocities for all abrasive particles and a single abrasive particle in 2.5-S.(b) CFD simulation results for 2.5-H AGHP.(b-i) Static pressure changes in 2.5-H.(b-ii) and Tracks and velocities of all abrasive particles and a single abrasive particle in 2.5-H.

Table 1 .
L-PBF processing conditions and processing parameters used in the AGHP AM.