Characterization, preparation, and reuse of metallic powders for laser powder bed fusion: a review

Laser powder bed fusion (L-PBF) has attracted significant attention in both the industry and academic fields since its inception, providing unprecedented advantages to fabricate complex-shaped metallic components. The printing quality and performance of L-PBF alloys are influenced by numerous variables consisting of feedstock powders, manufacturing process, and post-treatment. As the starting materials, metallic powders play a critical role in influencing the fabrication cost, printing consistency, and properties. Given their deterministic roles, the present review aims to retrospect the recent progress on metallic powders for L-PBF including characterization, preparation, and reuse. The powder characterization mainly serves for printing consistency while powder preparation and reuse are introduced to reduce the fabrication costs. Various powder characterization and preparation methods are presented in the beginning by analyzing the measurement principles, advantages, and limitations. Subsequently, the effect of powder reuse on the powder characteristics and mechanical performance of L-PBF parts is analyzed, focusing on steels, nickel-based superalloys, titanium and titanium alloys, and aluminum alloys. The evolution trends of powders and L-PBF parts vary depending on specific alloy systems, which makes the proposal of a unified reuse protocol infeasible. Finally, perspectives are presented to cater to the increased applications of L-PBF technologies for future investigations. The present state-of-the-art work can pave the way for the broad industrial applications of L-PBF by enhancing printing consistency and reducing the total costs from the perspective of powders.

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
Additive manufacturing (AM) enables the construction of three-dimensional (3D) components in a layer-by-layer manner from the computer-aided design file [1,2], which boasts the advantages of shorter lead time and fabricating complexshaped components [3,4].Furthermore, it provides a platform to fabricate high-performance alloys with unprecedented mechanical properties due to the non-equilibrium solidification process and the corresponding metastable microstructure [5].Among the various AM technologies, laser powder bed fusion (L-PBF) draws significant interest due to a lower surface roughness and complex geometry for the printed parts [6], making it account for about 85% of the metal AM market share [7].Till now, L-PBF has been broadly implemented to fabricate a variety of materials, such as titanium alloy [8][9][10][11], nickel-based superalloy [12], aluminum alloy [13,14], magnesium alloy [15], metallic glass [16], and shape memory alloy [17].However, there are still several challenges preventing the broad industrial applications of L-PBF.One is the lack of technological standards and certification approaches to assure printing consistency since the forming quality and property depend on a variety of factors, such as the powders, process parameters, and post-treatment [18].Alternation of one variable may engender the interlocked influence of microstructure and performance, which makes the quality control (QC) of variables important.The other is the high manufacturing cost which makes the L-PBF struggle to compete with conventional manufacturing methods.In many cases, the cost of L-PBF products is higher than that of conventionallyfabricated counterparts [19].The high cost of L-PBF has restricted its applications to critical end users, such as the aerospace and medical industries.Among the various factors, metallic powders as the starting materials play a critical role in influencing printing consistency and reducing manufacturing costs, which can pave the way for broader industrial applications.
From the perspective of printing consistency, the powder characteristics, such as the particle size distribution, flowability, and sphericity, significantly influence the spreadability and recoatability of powders during the L-PBF process, which can in turn impact the densification, microstructure, and properties of as-fabricated parts.Naturally, the QC of powders is vital for achieving printing consistency, which relies on quantitative or semi-quantitative characterization of powder characteristics.For one thing, the analysis of powder characteristics can rule out the powders that are out of specification (OOS), which might cause a knock-on effect on the quality of printed parts.For another, variation of powder characteristics may result in different printability and even alter the microstructure and performance of as-printed samples [20].Thus, the quantitative characterization of powder characteristics batchto-batch is critical for achieving repeatable productions.In consideration of its critical for printing consistency, the common powder characterization methods are briefly reviewed in the beginning.
As for the total cost of L-PBF, it is dependent on the specific materials, building plate, inert gas, and other factors [21].Generally, the powder cost takes up to 18% while the costs of machine and post-processing are 63% and 8%, respectively [19].As a result, reducing the powder cost provides an efficient way to make the total fabrication cost-effective, which can be achieved by either reducing powder manufacturing costs or enhancing powder usage efficiency.Up to now, the metallic powders amenable to L-PBF mainly rely on the atomization processes.The basic principle of the atomization process is the breakup of molten liquid into fine particles.Apart from the atomization process, a cost-affordable powder preparation method named as a fluidized bed has been implemented to modify the surface of the irregular-shaped powders [22][23][24][25], making them printable.Moreover, the yield ratio of powders based on the atomization process is generally in the range of 20%-40% [26], and it is necessary to explore novel powder manufacturing methods with a higher yield ratio to reduce costs.Recently, a novel powder manufacturing method termed cold mechanically derived (CMD) powder production [26,27] has emerged and been implemented for L-PBF.Apart from the atomization process, fluidized bed, ball milling, and CMD powder production, powder mixing has become popular to provide a cost-effective and convenient route to design alloys with novel compositions [28].The working principles, pros, and cons of various powder manufacturing methods are also reviewed in the present work.
Other than exploring novel powder manufacturing methods, increasing the powder usage efficiency also aids in reducing the powder cost.During the L-PBF process, only a fraction of the powders that have been spread are used for constructing the component, while the rest of the powders either remain on the building plate or are carried into an overflow tank.Reuse of these powders that remain on the building plate or in the overflow tank can maximize the powder usage efficiency and reduce the powder cost to some extent.During powder reuse, used powders are sieved and then mixed with virgin powders, or put directly into the next printing batch.However, the powders may undergo degradation in terms of physical and mechanical properties, such as chemical composition, particle size distribution, and flowability, during the interaction between the laser and powders.The powder degradation may influence mechanical properties, which makes it unable to meet the demands of industrial applications.Furthermore, more cycles of powder reuse reduce the cost, yet the characteristics of reused powders may fall OOS [29], thus a maximum value of powder reuse iterations needs to be specified to make sure that the mechanical properties can meet the application demand.The evolution of powder characteristics and the resultant mechanical performance with powder reuse are dependent on a myriad of factors, such as the materials used, powder management protocol, powder handling condition (such as whether virgin powders are introduced between batches), and printing condition.As a result, the evolution trend of powder characteristics and part properties varies among different materials and even  [34][35][36], powder preparation [22,27,37], and powder reuse [31,38].Reprinted from [34], Copyright (2021), with permission from Elsevier.Reprinted from [35], Copyright (2022), with permission from Elsevier.Reprinted from [36], Copyright (2020), with permission from Elsevier.Reprinted from [22], Copyright (2019), with permission from Elsevier.Reproduced from [27], with permission from Springer Nature.Reproduced from [37], with permission from Springer Nature.Reprinted from [31], Copyright (2020), with permission from Elsevier.[38] (2019), reprinted by permission of the publisher (Taylor & Francis Ltd, www.tandfonline.com).
the same material under different investigations.For instance, the strength of Ti alloys generally shows an increasing trend with powder reuse [30], while that of Al alloy decreases [31].In consideration of its important role in reducing the fabrication cost, the powder reuse of various alloys is reviewed.
As above-mentioned, powder characterization, preparation, and reuse are vital for reducing manufacturing costs and ensuring printing quality and consistency.Given their importance, there are already several papers reviewing the recent progress [21,32].Sun et al [32] have systematically reviewed the powder fabrication methods of Ti alloys, mainly focusing on the gas atomization (GA) process.However, the recent progress on fluidized bed, CMD methods, and powder mixing is not mentioned.As for powder reuse, Moghimian et al [33] have reviewed the effect of powder reuse on the chemical composition (mainly oxygen content) and mechanical properties (mainly yield strength (YS) and ultimate tensile strength (UTS)) for L-PBF and electron beam melted titanium, nickel, and aluminum alloys.Yet the evolution beyond these properties is seldom discussed.Douglas et al [21] have reviewed the effect of powder reuse on the powder characteristics and part properties, however, the dynamical mechanical properties were not investigated in that work.To fill the gap, the present review mainly dedicates to investigating the three aspects of powders, including characterization, preparation, and powder reuse.The flowchart of the present review paper is concluded in figure 1.To begin with, the powder characterization methods are briefly reviewed, serving as the basis for understanding the properties of manufactured and reused powders.Subsequently, the conventional and recentlyemerged powder preparation methods for L-PBF are reviewed.The manufacturing principles, advantages, and limitations of each technology are reviewed and compared.Finally, the effect of powder reuse on the powder characteristics and part properties is comprehensively analyzed on the four most studied alloys for L-PBF, namely, steels, nickel-based superalloys, titanium and titanium alloys, and aluminum alloys.After presenting the recent progress on these three topics, we close the review by presenting the conclusions and future perspectives.

Powder characteristics
The powder characteristics play a significant role in influencing the spreadability and recoatability during L-PBF, which influences the performance of L-PBF parts, and high-quality powders are essential for fabricating high-quality parts [39].Therefore, it is critical to investigate powder characteristics to ensure the reliability and repeatability of the fabricated parts.Furthermore, the characteristics of reused powders need to be tracked to ensure they do not fall OOS to ensure reliable processing.The powder characteristics can be mainly divided into two categories: external and internal characteristics.The external characteristics influence the powder spreadability and indirectly affect the formation of defects in the final parts.In contrast, internal characteristics are more directly relevant to the microstructure and properties of the as-fabricated parts.Among the various powder characteristics, PSD, sphericity, flowability, and chemical composition are frequently used in L-PBF.Thus, these four indexes and their corresponding characterization methods are introduced.In the meantime, the relationship between powder characteristics and printability is also briefly discussed.

Particle size distribution
PSD refers to the percentage of particles with different sizes in the total amount of particles in the powders.The demand for PSD for various AM technologies is different, which is schematically shown in figure 2(a).The PSD is 20-45 µm for L-PBF, 10-45 µm for cold spraying (CS), 45-106 µm for electron beam melting (EBM) [32], around 50-150 µm for directed energy deposition (DED) [40] and 20-150 µm for binder jetting (BJ) [41].Additionally, PSD directly affects the powder packing density [42,43].A wide PSD induces a higher packing efficiency than that of a narrower PSD, yet it causes an uneven spread layer and reduces the flowability of powders [42,44].However, if the PSD is too broad, the largest particles may not be melted and the smallest particles may be over-melted, engendering the possible occurrence of defects such as spatter, pore, and balling [42].In contrast, a narrow PSD decreases inter-particle friction, favoring the bulk and rheology properties [45].In conclusion, a certain fraction of fine powders are normally required to optimize part properties, since they can be easily melted, which is conducive to improving the density and surface quality of the parts [43,44].However, fine powders have limitations in terms of agglomeration, which can adversely affect the flowability [46].Therefore, it is critical to comprehensively consider the balanced effects of flowability and agglomeration when selecting a suitable PSD.
The widely used instrument for characterizing the PSD is a laser particle size analyzer based on laser diffraction.Laser diffraction is a class of non-image-based devices that detect and analyze the diffraction pattern of a laser beam directly passing through a dispersed granular medium.Normally the particles need to be dispersed in either a liquid phase medium (suspension) or a gas phase medium (air), which are referred as the 'wet measurement' and 'dry measurement', respectively [48].The total scattered or diffracted light pattern is mathematically inverted to show the PSD of spheres [44].
Although laser diffraction can measure the PSD of a large number of powders in a relatively short period with high accuracy, the mathematical assumption is that the particles are spherical.The highly irregular particles may result in inaccurate results.Moreover, it is challenging for laser diffraction to detect PSD when the powders are agglomerated [44].

Sphericity
Sphericity refers to the extent to which the shape of particles approaches the sphere.It is commonly believed that the sphere is ideal for flow since it has the smallest surface area to volume ratio, which is beneficial for minimizing surface friction [49].Sphericity has a significant impact on flowability and spreading behaviors, which affect the quality of printed parts in turn.It is established that spherical powders have the highest packing density, which inhibits the formation of micro-defects [50,51].In contrast, irregular particles (low sphericity) tend to cause interparticle friction, deteriorating the flowability [52], and decreasing the density of the printed parts.Consequently, it is indispensable to quantitatively characterize the sphericity based on the dimensionless ratios, known as form factors.The form factor that approaches 1 indicates a higher sphericity.
Characterization of sphericity is achieved by directly observing the particles using microscope-based instruments such as optical microscopes (OMs), scanning electron microscopes (SEMs), and Malvern Morphologi G3 microscope [44,47].SEM shows the merits of a high resolution and a large magnification range, which can be combined with other analytical instruments for comprehensive analysis.Typically, the particle size and form factors are obtained by image analysis.However, the use of SEM and OM to analyze the sphericity is generally based on small areas, typically 500 µm × 500 µm [53], and the protocol on how many particles need to be analyzed to obtain a reliable sphericity has not been established.Malvern Morphologi G3 microscope enables the quantitative measurement of the shape and sphericity of each particle.For instance, the morphology of H282 powders was analyzed with the use of a Malvern Morphologi G3 microscope [47], indicating that the powders show an acceptable circularity (figure 2(b)).Besides the Malvern Morphologi G3, the dynamic image analysis module equipped with a laser particle size analyzer enables the simultaneous measurements of PSD and form factors [52].

Flowability
Flowability refers to the ability of the powders to flow, which significantly influences the spreadability and the quality of the L-PBF parts.Good flowability is essential to obtaining homogeneous deposition of the powder bed, which enables an even layer thickness, good layer quality, better packing density, and uniform laser energy absorption, thus increasing the density and mechanical strength of the part [54,55].[32], for DED is cited from [40], and for BJ is cited from [41].(b) Morphology of H282 powders measured by Malvern Morphologi G3 microscopes.Reprinted from [47], Copyright (2021), with permission from Elsevier.(c) Schematic graph of the relationship between Hausner ratio and powder flowability.The data is from [44].
The flowability of powders depends on various factors, such as PSD, particle morphology, surface texture, cohesivity, and particle interaction [56].In general, a small percentage of fine powders with high sphericity contributes to improved flowability [57].However, a high volume fraction of fine powders are detrimental to flowability since the powders might agglomerate [58].Moreover, moisture, which makes the powders sticky, also impedes flowability.To date, several technologies are attempting to quantify the flowability of powders, and the following four methods are mainly discussed.
The first method is based on the Freeman Technology FT4 powder rheometer to measure the overall flow resistance of the powders under specific conditions [58].During the rotation and axial movement when a rotating blade moves through the powders, the radial and axial resistance is measured to calculate the energy required for powder flow.FT4 powder rheometer is a useful technology to measure flowability since it is possible to measure at very low normal stress [59] and can rank powders used for AM [60].Furthermore, the Freeman Technology FT4 can measure the flow energy, which is more representative of raking behavior in L-PBF [61].However, the operation of the FT4 powder rheometer is complex and the cost of the equipment is relatively high [62].
The second method is based on a Revolution Powder Analyzer (RPA), which is more consistent with the working condition of the powders [58].RPA consists of a rotating drum to determine the flow characteristics of the powders.The drum loaded with powders is rotated, and then a digital camera with the assistance of back-light illumination takes digital images of the powders during the rotation process.The software analyzes the images to calculate various parameters, such as avalanche angle and surface fractal ratio.Avalanche angle is most commonly used as a basis for assessing powder flowability.A smaller avalanche angle indicates better flowability.Spierings et al [43] measured the dynamic flow properties (avalanche angle and surface fractal) of powders with RPA, which showed a matched flowability between the calculated and observed values within the standard error.Nevertheless, the cost of RPA is relatively high and the operation is complex [62].
The third method is based on the Hausner ratio (HR), which is defined as the ratio between the tapped density and the bulk density [58].Bulk density is the mass of powders per unit volume when the powders are naturally filled in a specified container, which is measured by the funnel method.Tapped density refers to the mass of powders per unit volume measured after being vibrated at a certain frequency in the container till the volume does not change further.Figure 2(c) presents the powder flowability corresponding to different HR.A higher HR indicates poorer flowability.In general, the HR can be used for comparative studies of fine powders, yet it cannot reflect flowability quantitatively [63].
The final method is Hall flow test with the use of a Hall Flowmeter.This method measures the time required for 50 g powders to flow through the orifice of a Hall funnel, which is then expressed as mass per unit of time [64].It shows the advantages of a low cost and that it is suitable for ranking different powder qualities used in AM, especially in EBM [65].However, it also shows some limitations.For example, it is restricted to powders with superior flowability [66] and may result in stagnation due to funnel flow design [67].Additionally, the flow time has proved to be an ineffective predictor of powder spreadability [62].

Chemical composition
Chemical composition refers to the content of each element in the powders.For different AM applications, powders must conform to the alloy composition of the specified material.Validation of the chemical composition becomes extremely important when reusing the powders or the process may change the chemical composition [44], as even a slight change in composition can engender a significant influence on the density, phase distribution, and mechanical properties of the final parts.For example, Zhou et al [68] found that the crack susceptibility during L-PBF Al-Zn-Mg alloy was closely related to the concentration of Zn and Mg.A higher laser energy density triggers the more serve evaporation of Zn and Mg, enhancing the printability.Besides, higher amounts of O and H can lead to the formation of interstitial atoms that affect the mechanical properties due to their influence on the dislocation motion [20].
Inductively coupled plasma mass spectroscopy (ICP-MS) is a spectroscopic analysis based on plasma as an emission source, which provides quantitative chemical composition.Before the ICP-MS analysis, the sample needs to be dissolved in an acid solution and fed into the plasma.Under the excitation of high temperature and high energy of plasma, various elements produce their unique emission spectra.ICP-MS measures the intensity of the emission lines of these elements in the plasma, generating quantitative composition.ICP-MS allows for quantitative measurements of multiple elements at a wide concentration range [44], yet it is expensive and destructive to the resultant parts [44].However, several elements cannot be accurately measured by ICP-MS, such as C, O, N, H, and F [69], which requires the use of inert gas fusion.The principle of inert gas fusion is similar to that of carrier gas hot extraction [70].The sample is weighed and placed in a container made of graphite where it is heated to a molten state.Oxygen, nitrogen, and hydrogen molecules released from the sample are then separated and the mass percentage of each element is analyzed.Inert gas fusion or carrier gas thermal extraction is often combined with ICP-MS to comprehensively evaluate the chemical composition of powders [20,58].
The principles, advantages, and limitations of characterization methods for PSD, sphericity, flowability, and chemical composition are summarized in table 1.It is demonstrated that the characterization methods of each powder index show their advantages and disadvantages.Thus, it is necessary to combine a variety of techniques to complement each other when characterizing the powders.

Powder preparation methods
Reducing the cost of powders is an effective way to decrease the total L-PBF fabrication cost.This can be achieved by either decreasing the powder fabrication cost or enhancing the powder usage efficiency.There are currently several technologies suitable for fabricating powders, including mechanical, chemical, and atomization methods.As atomization can control the yield ratio and physical properties of the powders compared with the other two methods, it plays a dominant role in fabricating metallic powders [33].Among them, the widely used methods are GA [72], electrode induction melting GA (EIGA) [73], vacuum induction melting GA (VIGA) [74], plasma atomization (PA) [75], water atomization (WA) [76], centrifugal atomization (CA) [77], and plasma rotating electrode process (PREP) [78].Although the working principles of different atomization technologies vary in detail, they all share three similar procedures, including using a high-energy source to melt raw materials into a solution stream in an adjusted environment, followed by solidifying in a controlled manner.These three main steps can be defined as melting, atomization, and solidification [32].Apart from the atomization method, powder modification and mechanical methods have also been implemented to fabricate metallic powders for L-PBF [79].Fluidized bed, a cost-affordable powder preparation method, is mainly used to modify the morphology of irregular-shaped powders to improve the flowability [22][23][24][25].Furthermore, a cost and energy-effective powder fabrication process, named as CMD, has been used in L-PBF to produce Al alloy powders with a high yield and efficiency recently [27].Finally, powder mixing, a powder preparation method for mixing existing powders to obtain powders with novel chemical compositions, has been increasingly studied in recent years for its advantages of a low cost and a high freedom of composition design [80].In this section, the preparation principles, advantages, and limitations of the above-mentioned powder preparation methods are described in turn.In addition, the mechanical properties of L-PBF parts with powders fabricated via different preparation methods are also compared, including the Ti6Al4V powders prepared by GA, EIGA, PA, and PREP [81][82][83][84][85][86][87][88][89], AlSi10Mg powders prepared by GA and PA [20,[90][91][92][93], Inconel 718 powders produced by GA and PREP [94,95], 316L stainless steel powders fabricated by GA, VIGA, and WA [96][97][98][99], and 304L stainless steel powders yielded by GA and WA [100].

GA
The inception of GA can be traced back to 1872 when the first patent was filed by Marriott of Huddersfield [32].GA usually uses elemental raw material or pre-alloyed ingots as the raw material and can be used to fabricate various alloys [101].As shown in figure 3(a), the raw material is skull-melted in a water-cooled copper crucible with an electric arc or induction coils.Then the molten alloy is homogenized and enters the atomization chamber in free fall under gravity or through a confined refractory nozzle with a high pressure difference, where it is pulverized into droplets by the inert high-pressure gas jet, such as argon or helium [102].Finally, the metal droplets cool and solidify to form powders.The selection of atomizing gas depends on cost, thermal conductivity, and chemical reactivity with the alloy [103].The GA-fabricated powders possess relatively high sphericity, though satellites can be observed (figures 3(a-i) and (a-ii)).However, it shows the disadvantages in terms of a wide particle size distribution (50-300 µm for median particle size) [104] and a low yield ratio [101].In addition, the cost of GA-fabricated powders is • Destructive [44] high, for instance, the Ti-6Al-4V ((−150/+45) µm) powder produced by ATI Powder Metals is US$130 per pound [105].
The powder characteristics, such as the particle size distribution and yield ratio, produced by GA are influenced by several parameters, such as the atomizing gas pressure, melt superheat, and the gas-to-metal ratio (G/M) [112].It is found that the increase in air pressure, melt superheat and G/M decrease the median particle size, and increase the sphericity for the Al-17Si alloy powders [113].Martin et al [114] showed that increasing atomizing gas pressure substantially improves the yield ratio of particles with sizes smaller than 150 µm.EIGA was developed and patented by ALD Vacuum Technologies for producing active and highly spherical metal powders (especially Ti alloy powders) in 1991 [115].Pre-fabricated electrodes that rotate slowly and move downward electrically are melted directly on the surface without the need for ceramic lining (figure 3(b)).The tip of the electrode is melted in a conical induction coil and the liquid metal falls into the center of the gas nozzle, where it is atomized by a high-speed gas jet and pulverized into droplets.EIGA has been widely used due to the high yield ratio of fine powders and good powder sphericity [101].It is found that the gas-atomized Ti6Al4V powders (figure 3(a-i)) and Inconel 718 powders (figure 3(a-ii)) show inferior sphericity to that of EIGA powders (figures 3(b-i) and (b-ii)).However, the EIGA also shows some challenges.On one hand, the raw material must be in the form of pre-alloyed rods.On the other hand, element segregation of the rod may lead to the formation of powders with heterogeneous chemical composition [106].
The properties of EIGA-fabricated powders are influenced by several parameters, including the diameter of the atomizer and electrode.Wu et al [116] found that the diameter of the atomizer greatly affected the yield ratio of fine powders.When the diameter of the atomizer was 30 mm, the highest yield ratio of fine powders was obtained which was because the velocity of the recirculation zone and the argon flow were the largest.When the diameter exceeded 30 mm, the velocity, position, and turbulent kinetic energy of the recirculation zone decreased, decreasing the efficiency of primary atomization and thus the yield ratio of fine powders.Wei et al [117] demonstrated that the yield ratio of fine powders gradually increased and the particle size became more uniform as the pressure of atomizing gas increased from 2 MPa to 6 MPa.This is due to the fact that the back-flow region continues to expand and the gas velocity increases, improving the atomization efficiency.

VIGA.
VIGA is a technology widely used in producing industrial non-reactive metal powders, such as Fe-based, Ni-based, Co-based, and Cu-based alloys [96].The pressure needs to be reduced to a vacuum condition before melting, the alloy is subsequently heated to a molten state in the crucible, which then flows through the diversion tube to form a stable liquid column and enters the atomization chamber.The melt is broken into tiny metal droplets under a high-pressure gas jet, which is then cooled down quickly and solidified into powders of different sizes, entering the powder collection tank eventually.The whole process mainly includes the following steps: liquid shearing, primary atomization, second atomization, cooling, and solidification (figure 3(c)).The atomizing gas is generally argon with high purity to avoid powder oxidation and the introduction of impurity elements during the atomization process.VIGA has a high yield ratio of fine powders which is similar to EIGA.Powders prepared by VIGA possess certain advantages, including high purity and good sphericity (figures 3(c-i)-(c-iii)) [118].Taking Inconel 718 powders (figures 3(a-ii), (b-ii), and (c-ii)) as an example, VIGA powders exhibit a higher sphericity relative to that of GA and EIGA powders.Besides, the YS (∼470 MPa) and UTS (∼580 MPa) of the as-printed 316L stainless steel parts using VIGA powders are relatively lower than those using gas-atomized powders (∼520 MPa for YS, over 620 MPa for UTS), but the elongation of samples printed with VIGA powders (∼35%) is higher than that based on gas-atomized powders (∼28%) [96][97][98].Although VIGA is a mature technology, there are still a few issues that deserve deeper investigations, such as the undesired wide particle size distribution (40-70 µm for median particle size) [104] and the formation of satellites.
The properties of VIGA-fabricated powders are influenced by melt viscosity and atomization pressure.When the viscosity is 5.24 MPa•s and 5.45 MPa•s, the median particle size is close to 27.4 µm and 32.5 µm, respectively [119].Under these two viscosities, the continuous atomization process results in a good powder sphericity and a high yield ratio of fine powders.However, when the viscosity continues to decrease, the powders become coarser.This is because the flow velocity of the alloy liquid is too high and most of the particles solidify in a lower atomization atmosphere without complete crushing.The disorder of the two-phase flow field increases with increasing gas pressure [118], which is conducive to sufficient breakage of the melt, leading to a narrower particle size distribution and a decrease in the median and average particle size.Atomization pressure also influences the powder characteristics.When the atomization pressure increases from 2 MPa to 5 MPa, the median particle size decreases from 87.15 µm to 66.27 µm and the average particle size decreases by 23.4 µm.

PA
In 1996, Entezarian et al [120] developed PA technology for producing high-purity spherical powders with an average particle size of 40 µm.The pre-alloyed wire is normally used as the raw material.The electric energy provided to the plasma torch is converted into high thermal energy during PA, which generates an extremely hot ionized inert gas jet with high speed.The angle between the plasma torch and the vertical plane is about 20 • -40 • [121].The wire fed into the hot zone is melted and broken into droplets at high temperatures (figure 4(a)), which then solidify into powders in the atomization chamber.The wire in PA is melted and atomized by extremely high-temperature plasma simultaneously.In contrast, the metal is melted by induction coils or other heat sources and then atomized by high-pressure gas in GA.Since the whole process is carried out in an inert atmosphere, the liquid metal has no access to any contaminations before solidification, engendering a very low level of impurities.Furthermore, the yield ratio of fine powders using PA is relatively high [101].As for the powder morphology, the gas-atomized Ti6Al4V powders (figure 3(a-i)) and Inconel 718 powders (figure 3(a-ii)) exhibit a poorer sphericity than that of plasma-atomized powders (figures 4(a-i)-(a-iii)).However, the tensile properties of the Ti6Al4V parts printed with the plasma-atomized powders are comparable with those based on gas-atomized and EIGA powders, where the YS is ∼1000 MPa, UTS is ∼1100 MPa, and elongation is ∼10%, respectively [81][82][83][84][85][86][87][88].The hardness of Ti6Al4V parts printed with gas-atomized (∼360 HV) and EIGA powders (∼354 HV) is also similar [82,84].For AlSi10Mg powders, the samples printed with plasma-atomized powders exhibit slightly higher UTS and elongation yet a lower YS relative to those printed with gas-atomized powders [20].However, the main drawback of PA is that the feedstock needs to be in the form of wire [101] and the cost of plasma-atomized powders is expensive.For example, the price of Ti-6Al-4V ((−30/+250) µm) powders produced by Raymor is US$118 per pound [105].
The properties of PA-fabricated powders are influenced by several parameters, such as the diameter and the feed rate of the wire, the velocity and temperature of the plasma jet, the gas-to-metal ratio (G/M), and the angle of attack between the wire and the plasma jet [33,128].Dion et al [128] showed that preheating the wire before being fed into the hot zone significantly improves productivity, but the preheating temperature should not exceed the melting point of the wire to avoid melting.Larouche et al [129] reported that the yield of fine Ti-6Al-4V powders increased from 39.9% to 59.6% by increasing G/M from 8.7 to 12.9 and shortening the distance between the wire and plasma outlet.Yurtkuran and Ünal [130] found that the suitable geometry of the plasma torch and high-speed nozzle would improve the atomization efficiency.

WA
The high-pressure WA was established to fabricate copper and copper alloy powders during the decades of the 1930s and 1940s for the first time [73].At present, WA is mainly used to prepare non-reactive metal powders, especially stainless steel [101].WA is similar to GA and the main difference is the atomization medium.WA atomizes liquid metal by a high-pressure water stream while GA uses inert gas.As shown in figure 5(a), raw material is melted in an induction furnace and poured into the tundish.The molten stream is then broken into droplets by the water stream from the high-pressure pump.Finally, the resulting powders are dried and classified according to the desired particle size.Although WA benefits from a relatively low fabrication cost, the asfabricated powders need to be dried to remove the moisture and exhibit high oxygen content as well as low sphericity [101].The low sphericity of water-atomized powders is reflected from 316L stainless steel powders (figure 5(a-i)), and Inconel 625 powders (figure 5(a-ii)).As for mechanical performance, the YS and UTS of the 316L stainless steel parts printed by the water-atomized powders are slightly lower than that of the gas-atomized powders, but the difference in the elongation and hardness is marginal [96,97].Similarly, the YS (470 MPa), UTS (674 MPa), and elongation (29%) of the 304L stainless steel samples based on the water-atomized powders are lower than those of the gas-atomized counterparts (YS of 507 MPa, UTS of 688 MPa, and elongation of 69%) [100].In contrast, Cacace and Semeraro [99] found that the mechanical properties were comparable for L-PBF samples printed with gas-atomized and water-atomized 316L stainless steel powders.
The properties of powders manufactured by WA are affected by several process parameters, including water pressure, apex angle, melt flow rate, and melt superheat.The pressure of the water stream has a great impact on the quality of the powders.A high pressure engenders a reduced sphericity.In contrast, a pressure lower than a certain value may make the atomization process incomplete.As a result, it is necessary to set an optimum pressure level.Dhokey et al [135] reported that the particle size was inversely proportional to the water pressure and the apex angle under the condition of controlled variables based on the established mathematical model.Persson et al [136] confirmed the significant role of water pressure on the median size while that of the melt flow rate was inconspicuous.The particle size decreased by 45%-50% when the water pressure increased from 8 MPa to 17.5 MPa while the particle size decreased by 5%-15% when the diameter of the tundish reduced from 7 mm to 4 mm (melt flow rate decreased).Yenwiset and Yenwiset [137] found that water pressure, melt velocity, and melt superheat directly affected the average particle size.Moreover, a flow rate of molten metal higher than that of water pressure slows down the cooling rate of metal particles, generating a smaller quantity of irregularshaped powders.Finally, the increase of water pressure and melt superheat can significantly increase the yield of powders.

CA
CA is a technology to produce Sn, Zn, Ti, and steel powders [101,133,134] with high efficiency, using a rapidly rotating disc or cup of different wall angles [133] to break up the molten stream into droplets.During the CA process, the molten metal flows from the nozzle to the center of the high-speed rotating disc and spreads to the edge of the disc under the action of centrifugal force, followed by the pulverization into metal droplets (figure 5(b)).Finally, the droplets rapidly solidify into fine powders.The CA process is similar to the rapid solidification rate (RSR) process, which was first developed in the 1970s by Pratt and Whitney for making superalloy powders [138].The main advantages of CA are low production cost and the ability to produce powders of a wide range of particle sizes with a narrow particle size distribution.However, it is challenging to manufacture high-quality fine powders unless a very high speed can be reached, limiting the broad applications of CA [101].In addition, the CA-fabricated Sn and Zn powders are highly irregular (figures 5(b-i) and (b-ii)).
The properties of CA-fabricated powders are affected by several process parameters, such as atomizer geometry and rotation speed.It was shown that the particle size decreased with increasing atomizer speed and decreasing melt flow rate.Moreover, particle size is reduced by 25% using a cup with steep walls in comparison to a flat disc [139].Additionally, it was found that the powder size decreased with increasing disc rotation speed based on the mathematical model and experiment [140].Ho and Zhao [141] found that a solid skull would form on the disc in CA, which deteriorated the quality of as-fabricated powders.Furthermore, it was found that the volume of the skull decreased with increasing liquid metal flow rate, initial disc temperature, and initial liquid temperature based on the computational fluid dynamics.Zhang and Zhao [133] found that a wall angle of 67.5 • engendered the finer powders under the same atomizer rotation speed and melt flow rate because of the enhanced dynamic wetting between the melt and the atomizer.Finally, the particle size distribution gets narrowed when the atomizer rotation speed and melt flow rate are low due to the reduced film thickness of the melt before disintegration.

PREP
The PREP is a CA method [124], which was improved based on the rotating electrode process in the early 1970s, using plasma to replace the electric arc as the heat source [121].In recent years, PREP has been highly recommended for fabricating spherical Ti alloy powders with extremely low porosity [32,142].The metal electrode rod is melted by the plasma arc while rotating at a high speed.The molten metal is ejected from the electrode rod under the action of centrifugal force to form droplets, which solidify into spherical powders in the chamber (figure 4(b)).PREP powders are generally characterized by high purity, extremely good powder sphericity, and fewer satellites (figures 4(b-i)-(b-vi)) [101], which can be supported by the Ti6Al4V powders (figures 3(a-i), (b-i) and 4(a-i), (b-i)) [72,107,123], and Inconel 718 powders (figures 3(a-ii), (b-ii), (c-ii) and 4(a-ii), (b-v), and (b-vi)) [75,94,108,110].As for mechanical properties, the Ti6Al4V parts built by PREP powders achieve the highest YS of 1234 MPa and UTS of 1286 MPa relative to those based on GA, EIGA, and PA, but the lowest elongation of 5.22% [89].For Inconel 718 powders, the samples printed with PREP powders exhibit similar YS and UTS relative to those fabricated via gasatomized powders [94,95].Nevertheless, PREP also has its challenges in terms of a low yield ratio of fine powders [32].Furthermore, the cost of PREP powders is higher relative to that of GA and PA powders.It is reported that the price of Ti-6Al-4V ((−150/+45) µm) powders produced by Advanced Specialty Metals is US$189.00per pound [105].
The properties of PREP-fabricated powders are influenced by several parameters, such as electrode rotation speed, electrode diameter, and plasma current.Zhao et al [142] observed that the average diameter of powders decreased with increasing electrode rotation speed and diameter.In the meanwhile, the melting rate of 316 stainless steel powders improves with the increase of plasma current, generating a wider particle size distribution and an increased particle size.Nie et al [125] demonstrated that the volume fraction of fine powder increased with increasing the rotation speed, reducing the flowability and porosity of powders.

Fluidized bed
Spherical powders are desired in AM due to the better flowability and packing performance compared with irregular-shaped powders, which is beneficial for mechanical properties [52,54,55,143].However, powders with high sphericity are often expensive and difficult to fabricate.For instance, commercially pure titanium (CP-Ti) or Ti alloy powders prepared by GA, PA, and PREP (US$250-300 per kg) for L-PBF [144] are far more expensive than the hydrogenation-dehydrogenation (HDH) Ti powders [145] (around US$30 per kg [79]), which cannot be directly used for L-PBF due to highly irregular morphology and poor flowability [79].To this end, a fluidized bed has been implemented to modify the low-cost irregular-shaped metallic powders [22][23][24][25].Fluidized bed improves the flowability of the irregular-shaped powders and makes them printable for L-PBF.As shown in figure 6(a), the raw powders are treated through a fluidized bed reactor under high-purity argon gas in conjunction with heating the powders.The argon gas makes the solid powders in a state of suspended motion with a stable and high flow rate.During the high-temperature fluidization under flowing argon gas, collision, friction, and drag forces among individual particles occur, thus modifying the surface of powders [22].The irregular-shaped Ti powders after treating at 450 • C for 10 min are nearly spherical with a better smoothness and exhibit an enhanced flowability as presented in figure 6(a).The original sharp edges are ground due to collision and friction between individual particles and the impact of high-speed gas flowing on the particles.
The fluidization temperature plays a critical role in influencing the powder characteristics, such as PSD and flowability.The PSD becomes narrow with increasing temperature (figure 6(b)).The flowability decreases with increasing fluidization temperature (figure 6(c)).The flowability of (35.2 ± 0.3) s•50 g −1 treated at 450 • C decreases to (32.6 ± 0.2) s•50 g −1 after treatment at 500 • C, which is comparable with the Ti powders widely used in AM [146].However, the flowability becomes worse as the fluidization temperature increases to 600 • C, which is due to the fact that the Ti powders are slightly sintered and the fluidization state becomes worse when the temperature exceeds 500 • C [22].
Nevertheless, a critical problem during the fluidization of Ti powders at high temperatures is thermal oxidation [147][148][149], which significantly influences the final performance of the as-built parts [23].The oxygen content of Ti powders rises sharply with increasing fluidization temperature (figure 6(c)), Ding et al [23] established a model to elucidate the relationship between the oxygen content and oxide thickness of HDH Ti powders as a function of fluidizing temperature.This work helps to optimize the fluidized bed processing and control the oxygen pick-up during fluidization.
The printing parameters play a significant role in the mechanical performance of fluidized bed-treated Ti powders.When the scanning speed is 1200 mm•s −1 , the as-built pure Ti exhibits a good synergy of strength and ductility (figure 6(d)), which are superior to those prepared by GA and unmodified HDH powders as shown in figure 6(e) [24].Furthermore, it is demonstrated that the fatigue properties of the L-PBF pure Ti parts fabricated with fluidized bed-treated HDH Ti powders are comparable to those of conventionally fabricated pure Ti parts and L-PBF Ti-6Al-4V parts [24].
Apart from HDH Ti powders [22][23][24], the fluidized bed also enables the modification of pre-alloyed WMoTaTi refractory high-entropy alloy (RHEA) powders [25].The as-built WMoTaTi RHEA based on the fluidized bed-treated powders shows a comparable microhardness ((617.2± 4.1) HV) relative to that of other RHEA systems [25].However, the fluidized bed approach may not work effectively for metal powders with high density, such as Ta powders.This is because the collision speed of Ta powder particles descends due to high density, and the impact pressure between particles is much lower than the YS of Ta at the fluidizing temperature (400 • C) [25].Therefore, the flowability of the as-treated Ta powders is not improved.
To sum up, the fluidized bed process provides a costeffective method to treat irregularly shaped powders by modifying morphology, PSD, and surface roughness, enhancing the flowability.The parts printed with as-treated powders exhibit excellent mechanical properties, which shows a promising prospect in L-PBF.However, its applicability beyond HDH Ti and WMoTaTi RHEA powders still requires further investigation.[22], Copyright (2019), with permission from Elsevier.(d) Representative tensile engineering stress-strain curves of the L-PBF pure Ti as a function of scanning speed at room temperature, (e) summary of ultimate tensile engineering stress versus engineering strain for the L-PBF pure Ti including the as-treated HDH powders with the optimal tensile properties, and other results using the GA powders with the oxygen levels of 0.12 wt% [150], 0.13 wt% [151], 0.17 wt% [152], and the HDH powders prepared by jet milling [153] and ball milling [79] with the oxygen levels of 0.17 wt% and 0.27 wt%, respectively.(d), (e) Reprinted from [24], Copyright (2023), with permission from Elsevier.

CMD
The powder feedstock for AM atomization processes discussed above is typically produced on large, non-local atomization facilities with long lead times [27].Thus, the AM supply chain is constrained to some extent due to the energy required, non-locality of the current manufacturing strategy, and limited available materials [154].For these reasons, a direct low-energy powder preparation method, named as CMD, has been invented to fabricate non-spherical metallic powder for AM locally with high efficiency.The first patent on metallic powders for L-PBF fabricated via CMD was filed by John E Barnes and Christopher B Aldridge from Metal Powder Works in 2019 [26].At present, this method has been utilized to produce Al alloy powders, while its applicability beyond Al alloys is unclear and deserves further investigation [27].
During CMD, the bar is rotated at a specified speed and it is struck using a specially designed tool that has multiple teeth, producing particles of a specific size range, as determined by the input parameters when impacting.The impact is controlled to produce millions of particles per second with a highly repeatable size [26].Though the CMD powders are more irregular relative to GA powders, it has the advantage of high yield (over 95%) compared to that of GA powders (<20%) (figure 7(a)).Moreover, there is no significant difference in PSD for CMD and GA powders (figure 7(b)).In comparison to GA powders, CMD powders exhibit a slightly larger average particle size and tighter distribution with a lower sphericity (figures 7(c) and (d)).The CMD and GA powders show similar flowability at low rotation speeds (figure 7(e).When increasing speed, cohesion and the angle of repose increase for CMD powders, confirming their poor flowability.Interestingly, the samples printed with CMD powders with a low sphericity surprisingly exhibit higher strength and ductility (figure 7(f)) than those manufactured based on GA powders and wrought 7075 Al alloy.This indicates that the non-spherical powders fabricated by CMD can be successfully used in AM, enabling excellent mechanical properties on par with those of wrought counterparts [27].

Powder mixing
For the abovementioned powder preparation methods, one batch of atomization/fluidized treatment/CMD process engenders powders with a defined composition, which increases the cost when multiple alloy compositions need to be investigated.To reduce the cost of powders, powder mixing, viz.in-situ alloying, provides a more cost-effective way to explore powders with novel compositions to achieve high performance, either enhancing the printability [155] or mechanical properties [156].This rules out the necessity of an atomization process for each chemical composition [80,157].The basic principle of powder mixing is shown in figure 8(a).To achieve a homogeneous mixture of powders, different equipment or methods have been adopted, such as mechanical mixing [158][159][160], low-energy [161][162][163] or highenergy [164] ball milling machine and electrodeposition [28].The mixed powders are composed of matrix powders and added powders for tailoring the microstructure and mechanical properties of printed parts.The added powders can take the form of metallic powders [159] or ceramic powders [67].When the size of added powders is referred to, it can be in the form of nano-sized powders (figure 8(b)) [28], submicronsized (with the particle size ranging from hundreds of nanometers to several micrometers) [165] powders (figure 8(c)) [166], and micron-sized powders (figure 8(d)) [167].The nano-sized powders tend to attach to the surface of matrix powders, while the micron-sized powders tend to distribute uniformly in the matrix powders (figures 8(d-i) and (d-ii)) though some agglomeration of powders can be occasionally observed (as circled in figures 8(d-i) and (d-ii)).The powder mixing method has provided a tremendous space for designing novel alloys and has been used extensively for fabricating various alloys, including steels [168], nickel-based superalloys [169], and titanium alloys [170].
Though powder mixing can reduce the cost and ensure composition design freedom [171], there are still some problems that need to be solved before the technology can be widely used.Firstly, uniform distribution of matrix powders and added powders is critical for ensuring the consistency of powders, which depends on the factors involved during the powder mixing process, including the size and shape of the powders, as well as the equipment and parameters used for powder mixing.If they are not controlled properly, the nonuniform chemical composition [172] and agglomeration of fine powders under the action of van der Waals force [173] might occur, generating the formation of printing defects.Secondly, the investigation of the rheological properties of mixed powders is quite limited yet important since the rheological properties of the mixed powders may be different from those of the constituting powders [165].Finally, it is still challenging to reuse the powder mixture, since the composition might vary after sieving.
Based on the above discussion, table 2 gives a comprehensive summary of various powder preparation methods, including the adapted materials, as well as the pros and cons of each powder preparation method.It is observed that each technology exhibits both advantages and limitations in terms of production rate, yield ratio of fine powders, sphericity of powders, and cost.These factors need to be taken into consideration before selecting an appropriate manufacturing method by balancing the powder characteristics and total cost for manufacturing powders.

Powder reusability for different alloy systems
Apart from exploring novel powder preparation methods, another most commonly used strategy for reducing the cost is to sieve the used powders for the subsequent manufacturing cycles.The powder utilization rate can be increased up to 95% with powder reuse [174], which is beneficial for reducing the total cost of L-PBF.However, the powder characteristics undergo variations in terms of shape, size, and flowability, due to the interaction between the laser and powders and contamination from the L-PBF chamber and powder handling [175,176].
To date, the changes in the powder characteristics and properties of L-PBF parts due to the powder reuse are complex depending on the specific alloy system and powder reuse methods [177], thus, no accepted protocols for powder reuse have been reached [178].With the increased research on powder reuse of different alloy systems in recent years, it is necessary to understand the mechanisms leading to these variations to guide the adoption of powder reuse.It is noted that the scope of powder reuse in the present review is limited to powders with a defined chemical composition, and does not involve the reuse of mixed powders.This is mainly due to the challenges encountered during the reusing of mixed powders.On one hand, evolution trends of powder characteristics might vary for the matrix powders and added powders, which makes the prediction of the evolution trend of properties of the mixed powders difficult.For instance, the composition of mixed powders might vary after sieving since it is difficult to guarantee that the powders passing through the sieves follow the initially defined chemical composition.On the other hand, the compositional inhomogeneity of the mixed powders may get worse over multiple reusing cycles.A combination of extensive simulations or trial and error experiments is required to obtain a reasonable collocation of different powders before they are adopted in the industry [179].
The following part summarizes the changes in powder characteristics and properties of L-PBF parts based on the reused powders, which are schematically summarized in figure 9.The variation of powder characteristics is in the form of powder morphology, PSD, and rheological properties, while evaluation of the parts is characterized based on the densification, microstructure, hardness, and tensile properties.The dynamic mechanical properties, such as fatigue and creep properties of parts, are also investigated in some studies.Research work based on steels, nickel alloys, titanium and titanium alloys, aluminum alloys, and other less-focused alloy systems is selected to analyze the powder reusability.Through the systematic analyses of the evolution of powder characteristics and part properties, the possibility of whether the powders have the value of powder reuse is evaluated.

Powder degradation during powder reuse
Due to the focused laser heating on the powder bed, the L-PBF process bears risks of various defects that are brought about by its instability [184,185].The formation of spatters, which is an inevitable by-product of the L-PBF process, is the main reason for powder degradation during reuse [186].The spatters tend to trigger the formation of defects for L-PBF parts, which is not desired for high performance [187].Given their critical role, it is crucial to classify the spatters and evaluate their influence on powder properties separately.Although the alloy systems used are different, similar classifications of spatters based on the spatter state, location of generation, morphology, and chemical composition have been established.The generation of liquid-based laser spatters can arise from surface tension, vapor recoil pressure, explosion, laser energy uneven deposition, and movement process of melt and powder [188].Besides, solid-state spatters also exist due to metal vapor-induced entrainment [189] or metal vapor recoil pressure [190].One example of spatter classification for 316L stainless steel powders is given in figure 10, showing the various by-products due to the interaction between laser and powders, including the formation of oxide nodules, satellite powders, particle fusion, fume, condensate, particle collision, and ejecta (or spatters) in terms of full oxide ejecta, partially oxidized metallic ejecta, solid-state ejecta, and laser affected ejecta.Even between ejecta, different cooling rates lead to variations of phase formation and microstructure.
Due to the formation of spatters, the powder characteristics in terms of PSD, sphericity, and flowability also undergo variation, as listed in table 3.In terms of PSD, most of the studies have shown that it tends to shift towards coarse particles after reuse [183], due to the agglomeration of spatters [191] and reduction of fine particles [183].However, different trends, such as a left shift in PSD [192] or no significant change [181], are also found for different alloy systems.Due to the irregular shape of the spatters, it is uncertain whether they can pass through the sieve during sieving [180], which can explain why the D10, D50, and D90 of the sieved waste are all lower than the mesh size [31].Such uncertainty may explain the unclear trend of PSD variation of reused powders.In addition, the existence of irregular-shaped spatters, such as satellites or collision particles, may deteriorate the sphericity of the reused powders [191].However, different trends, such as the maintenance [193] or even increase [194] of the sphericity of the reused powders, are reported due to the removal of irregular satellite powders during sieving.In summary, the impact of powder reuse on sphericity is diverse depending on the specified alloy systems.Similar to the PSD and sphericity, the flowability of reused powders measured based on the HR [195], Hall/Carney funnel [191], RPA [195], FT4 [196], or Granudrum [194] does not reach a unified conclusion, which depends on the specific alloy systems.• 316L [191,197] • Hastelloy [198], • Inconel 718 [182,194,199], • Ti6Al4V [183], • CoCrMo [200] • Scalmalloy [31] • Agglomeration of spatter particles [191] • Melting of fine particles [183] Left shift • 17-4 PH [192], • Inconel 738 [201], • Inconel 625 [202] • Ti6Al4V [203,204], • AlSi10Mg [193,205], • Removal of large particles during the sieving process [193,205] • Increase of fine particles due to spattering [203,204] Widen • 316L [206] • 904L [207] • Simultaneous splattering, agglomeration, and large particle shattering [206]

Powder reusability of steel
Steels are the most common structural materials and take the largest portion for L-PBF [1,222].Among the various steels amenable for L-PBF, stainless steels, such as 316L stainless steel and 304L stainless steel, are the most frequently studied [223].In addition, precipitation hardening (PH) steels [224], martensitic steel [225], and 904L super austenitic steel [207] also attract interest for L-PBF.The wide applications of steels in L-PBF raise the necessity of studying their powder reusability [180,191,192,195,212,226,227].
As for the phase formation of reused steel powders [115,191,195], it is interesting to note that the reused 316L stainless steel powders exhibit a higher fraction of δ-ferrites due to the massive solidification through supercooling [180,191].EBSD results reveal that the reused powder contains highly spherical particles with ferritic structure (figures 11(a-i)-(a-iv)).With the formation of δ-ferrites with high magnetic susceptibility, the flowability of reused powders tends to decrease, generating the formation of defects [115].
For 304L stainless steel, 316L stainless steel, 904L stainless steel, and 17-4 PH steel, the major form of chemical composition variation of reused powders is the increase of the oxygen content [180,191,195,207,227], which is by the formation of oxides on the surface of reused powders for the reused 904L stainless steel powders [207] (figure 11(b-ii)).However, it is found that the oxygen levels do not exceed the typical limit for gas-atomized powder (∼1000 ppm) [191].It is indicated that the diameter of the circular oxidation spots was 1.80-10.21µm, attaching to the surface of spattered or heat-affected 316L stainless steel powders [197].Apart from the oxygen, the content of Cr, Ni, Mo, Si, and Mn decreased slightly, which would not modify the solidification path of 316L stainless steel [197].
Due to the minor variations of powder composition, there is no significant difference in the microstructure of L-PBF parts with reused powders.However, It is found that [191] the grains of the parts made of reused 316L stainless steel powders are refined to a certain extent, and the average grain size (d ave ) decreases from 108.5 µm for the samples printed with virgin powders (figure 11(c-i)) to 80.5 µm for samples printed with reused powders for 14 times (figure 11(c-ii)).It is possibly because the oxides in the powders are uniformly distributed in the matrix, acting as nucleating agents to refine the grains.
Regarding the effect of powder reuse on the densification of L-PBF steels, there are no consistent results [180,191,226,227].The density of L-PBF 316L stainless steel [191] based on Archimedes ′ principle is quite stable within the initial 5-6 printing cycles and then shows a decreasing trend (figure 11(d)).The reduction of density may be related to the increase of the recoil pressure of the molten pool and the change of surface tension caused the oxidation.In another study, an increased fraction of voids are found in the 316L stainless steel samples with powder reuse, coinciding with an increase in gas content in reused powders [206].Sutton et al [227] found that the powder reuse did not affect the density of 304L stainless steel parts, which fluctuated between 99.03% and 99.13%.However, relative density (98.98%) of 17-4 PH steel printed with reused powders for 20 times is higher than that of samples printed with virgin powders (98.82%) [226], ascribing to the irregular size of reused powders, which reduces the surface roughness of the parts.
Owing to the formation of oxides and variation of densification, the mechanical performance, including hardness, tensile properties, and fatigue properties, of steel with reused powders shows different evolution trends with those printed with virgin powders depending on the specific alloy systems.It is found that the hardness remains stable at 212 HV with a standard deviation of 8 HV [191] for the 316L stainless steel samples printed with virgin and reused powders.In contrast, the L-PBF 17-4 PH steel based on reused powders shows a decreased hardness by 16 HV even though the parts exhibit a higher densification relative to that printed using virgin powders [226].The detailed mechanism leading to the hardness reduction is not given.
In terms of static tensile properties, it is observed [191] that the YS, UTS, and elongation to fracture of the 316L stainless steel parts remain stable, even though the grain size decreases for samples printed with reused powders.A slight decrease in the UTS and YS concurrent with an improved ductility for the 316L stainless steel using reused powders is also observed (figure 11(f)).Lanzutti et al [206] found a slight increase in YS, UTS, and a decrease in elongation of 316L stainless steel, due to the formation of sub-micrometric oxide inclusions introduced by the reused powders.For 17-4 PH steels [226], the YS (859 MPa) decreases printed using virgin powders to 770 MPa printed with reused powders, while the UTS (1038 MPa) based on reused powders is higher than that (984 MPa) printed using virgin powders.It is worth noting that the 17-4 PH parts prepared using virgin and reused powders show different work hardening behaviors (figure 11(e)), where the samples printed using reused powders show a deformation plateau from the strain of 4%-12%.This lacks further explanations and may be associated with the variation of chemical composition and change of stacking fault energy [228].
Apart from the static mechanical properties, there are also some studies investigating the dynamic properties of L-PBF steels with powder reuse.The impact toughness of the 304L stainless steel parts printed using reused powders decreases by about 50 J with the increase of reusing times (figure 11(g)) [227], which may be associated with the increased fraction of oxide inclusions in the reused powders, which serve as the sites for micropore formation.Soltani-Tehrani et al [192] observed that the 17-4 PH steel specimens fabricated from reused powders (14 times) showed a higher cycle fatigue strength than those prepared from virgin powders (figure 11(h)).This is primarily related to larger defects in the part fabricated by virgin powders since defects are often the origin of cracks.

Powder reusability of nickel alloy
Nickel-based alloys, which retain most of their strength even after prolonged exposure to extremely high temperatures, show extensive applications in the turbine section of jet engines [229].Due to its poor machinability, nickel-based superalloys, such as Inconel 718 [230], Inconel 738 [231], Inconel 625 [232], or Hastelloy X alloy [233], are frequently investigated for L-PBF due to the advantage of near-net shaping.To reduce the fabrication cost and improve the powder usage efficiency, there are already some investigations on powder reuse for L-PBF nickel alloys [35,42,174,182,194,198,199,201,234,235].
Similar to the steel, the change of composition in the reused Inconel 718 alloy powders seems to be limited to the increase of oxygen content [35,199,201,234], while the variation of other elements, such as C, N, and H, is marginal [182,194,199,234].The increased oxygen content is in the form of Cr, Al, and Ti-rich oxides with a size of several micrometers [35,182,201] attaching to the surface layer of reused Inconel 718 powders.As for the microstructure of Hastelloy X powders [198], both the virgin and reused powders show no obvious texture (figures 12(a-i) and (a-iii)).However, the stress is concentrated on the surface of virgin powders while it is concentrated at the grain boundaries (figures 12(a-ii) and (a-iv)).
As for the printability, it is indicated that the densification of reused Inconel 718 powders can be maintained [194] or slightly increased [182].The slight increase in density of L-PBF Inconel 718 with powder reuse, from 98.3% to 98.7%, may be related to the enhanced flowability of the powders [182].However, Yi et al [199] observed that the increase of powder reuse times increases the sphericity of the pores and the porosity level in the part based on the x-ray CT results.This is explained by the trapped gas inside the large-sized particles in the reused Inconel 718 powders [236].As to Hastelloy, a decreasing trend of densification was found [198] due to more defective particles in the reused powders.
There is no obvious microstructure variation, including grain size, dendrite size, and texture, when reusing the Inconel 718 or Inconel 738 powders [182,199,201].For instance, it is found that the L-PBF Inconel 718 samples printed with both virgin and reused powders exhibit continuous γ matrix and flower-shaped γ ′ precipitates, as well as similar texture and grain size [234] (figures 12(b-i) and (b-iii)).The average kernel average misorientation (KAM) of Inconel 718 samples fabricated using reused powders is higher than that with virgin powders (figures 12(b-ii) and (b-iv)), which may imply an increase in the dislocation density.However, the mechanism of variations on average KAM is not presented.
As for hardness, the studies hitherto indicate that the hardness undergoes a marginal decrease when printed by using the reused Inconel 718 powders [42,194] since the physical and chemical properties of the powders did not change with the reuse.Gruber et al [194] observed that the hardness of samples prepared using reused Inconel 718 powders decreased marginally from 334 HV to 329 HV.When the room temperature tensile properties are referred to, different evolution trends have been obtained for L-PBF parts printed with reused Inconel 718 or Inconel 738 powders [182,199,201,234].Chen et al [201] observed that samples prepared using virgin and reused Inconel 718 powders exhibited nearly identical UTS, YS, and elongation.However, for hightemperature tensile properties, it is observed that samples prepared with reused Inconel 718 powders exhibited higher YS and UTS, but slightly lower ductility at most of the tested temperatures, which may be related to more substitutional and fewer interstitial solute atoms in the solid solution [234].The high-temperature tensile properties of heat-treated Inconel 718 samples (figure 12(c)) indicate that the first three powder reuse cycles can increase the YS and UTS, and the elongation is maintained.However, the retardation effect of dislocation motion by alumina decreased at 650 • C owing to the increased diffusion rate, resulting in a nearly constant YS and UTS.The slightly lower UTS ((753 ± 8) MPa) and elongation (11.1%) of L-PBF Hastelloy fabricated from reused powders than those of virgin samples ((780 ± 7) MPa and 14.2%) result from the significant defects (>40 µm) in the samples printed with recycled powder [198].
Up to now, there is only one study investigating the creep behavior of L-PBF Inconel 718 prepared from reused powders [234], which indicates that samples prepared with virgin powders exhibit better creep properties at 550 • C and 600 • C.However, the opposite trend is observed when tested at 700 • C (figure 12(d)).The mechanism may be due to the fact that the alloy fabricated with reused powders contains more available dislocations, which can be easily activated to accommodate the creep strain.However, the higher dislocation content at 700 • C hinders the diffusion process, thereby increasing its creep lifetime.
In terms of fatigue properties, the fatigue crack propagation (FCP) rate of Inconel 718 specimens at both 25 • C and 650 • C does not appear to be related to the powder reuse cycles [235] (figures 12(e-i) and (e-ii)).Paccou et al [182] observed that L-PBF Inconel 718 printed with reused powders showed slightly reduced fatigue life (figure 12(f)) when loaded with small plastic amplitudes (∆ε p /2 = 0.000 04), However, the difference was even smaller than the difference of different parts prepared under the same parameters.Furthermore, it is found [201] that powder reuse has a weak negative effect on fatigue life and scatter at lower stress amplitudes for the Inconel 738 powders (figure 12(g)).

Powder reusability of titanium and titanium alloy
Titanium or titanium alloys have extensive industrial applications due to their high specific strength, low density, high corrosion resistance, and good biocompatibility [30,237].To date, various titanium and titanium alloys have been extensively investigated via AM technologies, including the workhorse Ti6Al4V, β-type titanium alloy [238,239], and CP-Ti alloy [240].In consideration of high cost of the titanium and titanium alloys, it is meaningful to understand the variation of powder characteristics and part properties of titanium and titanium alloys with powder reuse [181,183,196,203,204,[241][242][243][244][245].
Since titanium is quite sensitive to oxygen, the reused powders tend to exhibit a higher oxygen content compared to that of virgin powders, as supported by various studies [181,183,196,203,242,245].Although no change in powder composition is observed by EDS [203], the content of O and N elements in the reused powders increases [183,204,244].It is observed that the O content of the powders increases from 0.128 wt% to 0.140 wt%, and the N content increases from 0.006 wt% to 0.019 wt% after reuse for 8 times [183].An oxide layer with a thickness of about 20 nm is on the surface of powders which have been reused for 18 times [245] (figures 13(a-i) and (a-ii)).The increase in oxide thickness was also reported for Ti5553 [246], in which the thickness of hydroxylated mixed oxide composed of TiO 2 and Al 2 O 3 increased from (5.6 ± 0.7) nm to (8.3 ± 1.1) nm after repeated use.The changes in the thickness or composition of the oxide layer on the surface of the powders may lead to changes in the moisture absorption capacity of the powders.Although there is an increase in content of other elements, such as nitrogen and hydrogen, they do not exceed the limits of grade 23 based on ASTM F13-136 [181].Additionally, it was found that the oxygen content of CP-Ti powders was stable at 0.12% after reuse for ten cycles [209].
In terms of the microstructure of the powders, it is found that both the virgin and reused Ti6Al4V powders consist of a dark brown with an irregular lighter region after etching with ammonium bifluoride (figures 13(b-i) and (b-ii)) [247].However, the lamellar structure can be observed in the lighter area for reused powders (figure 13(b-ii)) Yet, the mechanisms behind the variation in the microstructure or phase composition were not given in the study.
The pick-up of oxygen and nitrogen of Ti and Ti alloy powders causes various evolution trends in the densification of printed parts among different researches [203,204,242,245,248].The inconspicuous difference in densification for samples printed with virgin and reused powders is commonly observed.It is shown [242] that the maximum difference in the densification of L-PBF Ti6Al4V parts based on virgin and reused powders is only 0.4%.In another study [203], the powder reuse has no significant influence on the part density, which fluctuates between 99.6% and 99.9%.Finally, it is reported that the density of the samples fabricated using virgin and reused CP-Ti powders for 10 times [209] is the same.However, Soltani-Tehrani et al [183] observed that the porosity of the parts tended to decrease first and then increase as the  [247], with permission from Springer Nature.(c) EBSD map and phase distribution of virgin (i and ii) and reused (iii and iv) samples.Reproduced from [248], with permission from Springer Nature.(d) Tensile properties, including yield strength (i) and elongation (ii).Reprinted from [242], Copyright (2020), with permission from Elsevier.(e) Strain-life comparison [196].Reproduced from [196], with permission from Springer Nature.(f) Log-normal distribution of fatigue lives.Reprinted from [183], Copyright (2023), with permission from Elsevier.(g) Charpy results.Reprinted from [245], Copyright (2023), with permission from Elsevier.number of reuse times increased.The improved packing state of the powder bed with reuse might account for the reduction of defects in the first 4 cycles, yet the increased O and N content in the reused powders may be the main reason for a higher fraction of spatters, increasing defects in parts fabricated with powders reused for 8 times.
As for microstructure, an obvious difference in microstructure for the Ti and Ti alloy samples prepared using the virgin and reused powders is not observed [183,203].Both of the Ti6Al4V samples consist of a large amount of martensite, while repeated heating between layers results in the formation of α + β phase.However, the increased O and N content in the reused powders might increase the rate of martensite transformation, leading to the formation of a higher fraction of α ′ phase [243], which can explain the decrease of β-phase from 4.6% using virgin powder to 2.4% using reused powder (figures 13(c-i)-(c-iv)).For CP-Ti, there is no microstructure difference between samples prepared using virgin and reused powders [209].
The mechanical properties of L-PBF Ti and Ti alloys experience variations due to composition change.An increasing trend is observed for the hardness of samples fabricated with reused Ti6Al4V powders [241,243,248], which is linked to the elevated oxygen content in an interstitial solid solution, since it may modify the lattice parameters or c/a ratio, influencing the dislocation glide and causing an increase in the strength but a reduction in ductility [249].In terms of tensile properties, most of the studies [183,242,243] show that the YS and UTS increase for samples printed with reused Ti6Al4V powders.For example, the YS increases from 1106 MPa to 1158 MPa with increasing reuse cycles, while the UTS is enhanced from 1134 MPa and 1180 MPa, respectively [183].Alamos et al [242] reported that the UTS did not change significantly due to powder reuse, yet the YS increased from 930 MPa printed with virgin powders to about 950 MPa for samples printed with reused powders for 8 times (figure 13(di)).This may be due to the effect of O and N, causing the interstitial solid solution strengthening.Meanwhile, the increase of defects reduces the ductility (figure 13(d-ii)).As for CP-Ti, the YS of the sample increases from 412 MPa to 426 MPa while the UTS increases from 553 MPa to 565 MPa after powder reuse.Meanwhile, the elongation remains stable at around 25%.However, a detailed mechanism is not presented for property variation.
In terms of fatigue performance, different evolution trends are reported among different studies.Carrion et al [196] have shown that the fatigue life of machined specimens fabricated from reused powders is significantly longer than that of specimens fabricated from virgin powders (figure 13(e)).The improvement in fatigue performance is due to a combination of increased flowability resulting in a more uniform layer distribution within the powder bed and smaller pores within the as-built parts.In contrast, opposite trends are also observed in which the fatigue performance first increases and then decreases, showing the same trend as that of defects [183] (figure 13(f)).The improvements in fatigue performance are explained by the powder rheology, which reduces defect formation during the first few cycles of powder reuse.However, as the powders are further reused, the accumulation of O and N in the powders causes a higher fraction of spatters, leading to an increased fraction of defects.As for the Charpy test, Meier et al [245] reported a trend of embrittlement especially after 10 reuse cycles, as shown in figure 13(g).The decrease in impact toughness may be related to the increase of TiO 2 content in the sample, which originates from the oxide layer on the surface of the reused powders (figure 13(a-ii)).

Powder reusability of aluminum alloy
As the second most used alloy after steel, aluminum alloys show extensive applications in aerospace, the initial adopter of AM technology.Since its inception, the Al alloys suitable for AM mainly rely on the near-eutectic Al-Si alloy [250], since L-PBF high-strength alloys, such as Al2000, Al6000, and Al7000 alloy, usually engender cracks, deteriorating the mechanical performance [94,251].With the development of inoculation treatment to refine grains, crack-free high-strength Al alloys can be fabricated via L-PBF [252,253], which has triggered the rapid development of L-PBF Al alloys.However, compared with the steels, nickel-based superalloys, and Ti alloys, the investigations on the powder reuse of Al alloys are limited, which are restricted to Al-Si-Mg alloy [31,193,205,208,211,[254][255][256][257] and AlMgScZr alloy, e.g.Scalmalloy [252].The evolution of properties of L-PBF aluminum alloys with powder reuse is listed in turn.
Similar to Ti6Al4V, aluminum is considered to be an extremely reactive material with oxygen.The oxygen content of powders increases rapidly during the initial period of reuse, while surface passivation of the AlSi10Mg powders slows down the rate of oxidation with further reuse [193,208].As a major source of increased oxygen content, the number of spatters increases stably after each cycle, and the rough oxide layers riched in Al and Mg can be observed on the surface of spatters (figure 14(a-ii)), which are not detected in the virgin powders (figure 14(a-i)).Apart from oxygen, Fiegl et al [208] observed that the hydrogen content of the AlSi10Mg powders increased from 80 ppm to 130 ppm after 30 months of usage due to the absorption of water and oxygen by the oxide layer on the surface of powders.
As for the densification of printed parts, the porosity level of L-PBF aluminum alloys printed with reused powders increases in most of the studies as a consequence of the increased oxygen and hydrogen content [31,208,256].Oxide inclusions due to high oxygen content act as a source of crack formation, and increased hydrogen content leads to the formation of larger porosities [31,208].In contrast, several other studies [211,257] show that powder reuse has no significant effect on densification.The enhanced flowability of the powders may compensate for the porosity caused by the increased oxygen content.
When the microstructure is referred to, no significant changes are observed for AlSi10Mg [255] and AlSi7Mg powders [257] with powder reuse.For the Scalmalloy [31], it is observed that the melt pools of samples prepared with reused powders appear to be larger and exhibit a more pronounced periodicity compared with those fabricated with virgin powders.Moreover, samples printed with virgin powders exhibit a larger area of fine grains along the melt pool boundaries (figures 14(b-i) and (b-ii)).However, this may be associated with sample preparation and cannot be attributed to the influence of powder reuse.
Given the increased fraction of pores, the powder reuse essentially brings about a decrease in tensile properties for  [193], Copyright (2021), with permission from Elsevier.(b) Microstructure of virgin (i) and reused (ii) Scalmalloy samples.Reprinted from [31], Copyright (2020), with permission from Elsevier.(c) Analysis of the surface roughness of virgin and reused samples with respect to exposure time interval to salt spray test: (i) Ra (roughness average value) and (ii) Rt (roughness total height value).Reproduced from [258], with permission from Springer Nature.
L-PBF Al alloys [31,205,208,257].For AlSi10Mg, it is shown [205] that a systematic decrease in the YS and UTS is observed as the number of reuse times increases.The YS decreases from 213 MPa to 204 MPa while the UTS decreases from 313 MPa to 302 MPa.In the meanwhile, the elongation fluctuates at 18%.Fiegl et al [208] observed that the hardness and tensile strength of samples fabricated with reused AlSi10Mg powders decreased relative to those printed with virgin powders.The YS decreased from 222 MPa to 191 MPa while the UTS decreased from 335 MPa to 285 MPa.It is proposed that the main reason for the performance degradation was the existence of numerous larger gas pores inside samples printed with reused powders.Cordova et al [31] observed that the YS, UTS, and elongation of the L-PBF Scalmalloy printed with reused powders were reduced.
Apart from the tensile property, the corrosion property of AlSi10Mg samples [258] in terms of the surface roughness of samples printed with reused powders is slightly higher than that printed via virgin powders at all exposure times in the salt spray test.This is related to the initial surface condition of the part.The irregular powders presented in the reused powders make the surface rougher.Furthermore, the wear rate is higher for the samples printed with reused powders [259].The possible mechanism is that the detached particles from the surface of samples printed with reused powders adhere between the contact pair due to the formation of the oxide layer, which remains on the wear track throughout the entire duration of the sliding test, increasing the friction force at the surface and raising the wear value.In contrast, the detached particles fill up the vacant junction and continuously provide a wear-resistant film between the contacts for samples printed with virgin powders, therefore reducing the wear rate.
It is worth noting that there is a lack of relevant research on the dynamic mechanical properties of parts prepared from reused Al alloy powders.The only current study performed by Del Re et al [205] showed that the powder reuse led to a decrease in the high cycle fatigue performance of the part prepared by AlSi10Mg, which was associated with the presence of internal defects.

Powder reusability of other alloy systems
Apart from the several common alloys mentioned above, several studies on powder reusability of other alloy systems, including Co-Cr-based alloys [200,260,261], Cu alloys [262], and high entropy alloys [263], have also been conducted, which are described in turn in the following.
Co-Cr-based alloys have been extensively implemented in biomedical implant engineering due to their good biocompatibility and corrosion resistance in chloride environments [264].L-PBF is suitable for fabricating medical or dental parts, due to their complex geometry, low volume, and strong individualization [265].The reusability of Co-Cr alloy powders has been reported in several studies, including Co-Cr-Mo [200,266] and Co-Cr-W [260,261].Overall, Co-Cr alloys are unsuitable for multiple cycles of reuse from two aspects.Firstly, the content of Co and Cr ions significantly increases with increasing reuse times, which is more cytotoxic to human gingival cells, and induces more cell apoptosis than the alloys fabricated from virgin powder [266].Secondly, the porosity of the samples printed using reused powders increases due to the generation of more spatters.Due to a higher fraction of porosities, the YS decreases from 855 MPa to 689 MPa while the elongation decreases from 11.15% to 3.82%, respectively [261].
Cu alloys, especially bronze, are widely used in various industries for their excellent electrical and thermal conductivity, corrosion resistance, tensile strength, and ductility [267].
Though several studies have been conducted for L-PBF Cu alloy [162], there is limited research on powder reusability of Cu alloy powders.To the best of our knowledge, only one study [262] focuses on the reusability of chromium bronze.It was found that the chromium bronze printed with reused powders exhibited a YS of (136.8 ± 8.7) MPa, a UTS of (187.4 ± 10.1) MPa, and an elongation of (15.5 ± 2.3)%.However, the mechanical performance of samples fabricated with virgin powders was not investigated.
Fabricating high-entropy alloys via L-PBF has attracted increased research interest in recent years due to its potential to achieve high mechanical performance.However, the investigation of the reusability of high-entropy alloy powders is limited.Guo et al [263] investigated the effect of powder reuse on L-PBF AlCoCrFeNi 2.1 .Unlike the formation of an ultra-fine layered structure consisting of alternating FCC and B2 phases in the samples fabricated with virgin powders, samples fabricated using reused powders exhibit a cellular structure composed of nearly cubic FCC phases below the melt pool boundary, surrounded by B2 phases.The mechanism leading to the different microstructures is that the oxide impurities adhering to the surface of the reused powders interfere with the stable growth of thin films, leading to the formation of cellular structures.As for mechanical properties, the YS of samples printed using reused powders ((1042 ± 25) MPa) is lower than that using virgin powders ((1210 ± 23) MPa), yet exhibits a higher elongation ((26 ± 2)% vs. (16 ± 1)%).The relevant studies indicate that the oxides in the reused powders may lead to the microstructure evolution and affect the mechanical performance.

Conclusion
In the present review, the state-of-the-art knowledge on powder characterization, preparation, and reuse of the L-PBF process is thoroughly reviewed.The typically studied alloy systems, namely, the steels, nickel-based superalloys, titanium and titanium alloys, aluminum alloys, as well as other lessstudied alloys are selected to review the influence of powder reuse on the evolution of powder characteristics and properties of as-printed parts.Based on the previous work, the conclusions can be drawn as below: (1) Powder characterizations, including particle size distribution, flowability, sphericity, and composition, are critical for ensuring printing consistency.By characterizing the powders, the printing feasibility of powders and printing consistency can be evaluated.(2) Various technologies have been implemented to fabricate the powders for L-PBF, and the GA in terms of the EIGA, VIGA, and PREP takes the largest portion.The fluidized bed method effectively improves the flowability of irregularly shaped powders, enabling the L-PBF parts with good mechanical properties.The CMD metallic powders with low sphericity may potentially enjoy popularity due to their low fabrication cost.Finally, powder mixing engenders a tremendous compositional space to design novel alloys in a cost-effective manner.(3) Powder reuse provides an effective route to reduce manufacturing costs, yet the powder degradation during reuse may influence the printability, microstructure, and the resultant mechanical performance depending on the alloy systems.Due to the multiple variables, including the material systems, L-PBF process, characterization methods of powders and L-PBF parts, the results of powder reusability are not yet unified.Generally, the chemical and mechanical properties of Ti6Al4V and Al alloys are more susceptible to powder degradation relative to that of steels and nickel-based superalloys.The oxygen pick-up of reused Ti6Al4V powders tends to enhance the strength, while the oxygen pick-up of aluminum powders after reuse leads to a significant decline in tensile properties.

Outlook and perspective
Table 4 summarizes the information given by the recent research work related to powder reuse, including processing methods, preparation methods, and the change of properties.
It is seen that the evolution trends of powder characteristics and part properties vary among different alloy systems and even within the same alloy system, indicating its complexity.This makes the proposal of an accepted powder reuse protocol infeasible.To keep the L-PBF affordable with its expansion, the following aspects of powder characterization, preparation, and reuse need to be taken into consideration for future investigations.
(1) At present, apart from the experimental methods [62] to characterize the powder property, the simulation method has also been adopted to understand the effect of powder morphology and PSD on flowability, such as through the discrete element method (DEM) [143].Compared with the experimental methods, DEM shows high flexibility, and a variety of models can be established to further analyze the powder-spreading mechanism.Future work can be conducted by exploring more simulation methods to understand the powder characteristics.(2) There are already some pioneering works investigating the feasibility of fabricating metallic powders for L-PBF based on mechanical methods, such as fluidized bed and CMD.However, the current study on the fluidized bed treated powders is mainly on HDH Ti and WMoTaTi RHEA powders, especially HDH Ti powders.Its usability extending to other metallic powders needs to be further explored.Besides, the relevant work on CMD is limited to Al powders, and its applicability beyond Al powders, such as stainless-steel powders and Ti alloy powders, also deserves deeper investigation.(3) As above-mentioned, the printing quality of L-PBF parts is a complex function of various parameters, including the feedstock powders, the printing parameters, and the building environment.Thus, when the effect of powder reuse on the part property of L-PBF fabricated alloys is Various studies on the reusability of different alloy powders.NA denotes that no relevant studies were performed.↑ investigated, it is necessary to keep the other parameters consistent.However, this is generally not mentioned in the published work.A typical variable is the moisture content, which influences the flowability and subsequent printability [268].As a result, it is necessary to set up a standard to rule out the possible influences from other variables except the variation of powder quality during reuse.For instance, powder drying is an effective method to remove the moisture.(4) Specific to the powder reuse methods, it is essential to distinguish the two powder reuse strategies and investigate the reuse times that ensure the powder quality is within the specifications.One is based on the single batch of powders that are continuously reused build after build till the powders are OOS, while the other strategy is to refresh the powders with virgin powders.Although the latter strategy poses challenges for traceability and process consistency, for practical applications, adding virgin powder can slow down powder degradation and ensure sufficient powder volume for the next use.It is necessary to find the limit of powder reusing times by mixing the virgin powder with reused powder for renewal.What's more, further processing of recovered powders to enhance their printability is also worth investigating [79].(5) The powder degradation in terms of variation in powder characteristics and part properties is dependent on various factors during the L-PBF process, such as the building chamber conditions, part size, and printing parameters [269].For instance, the building chamber with a higher oxygen content inevitably results in higher oxygen content in the reused powders.The dependence of printing environment, printing parameters, and part size on powder degradation needs to be explored to better understand the reusability of powders.(6) Powder reuse is mainly focused on pure alloy powders, while it is challenging to reuse mixed powders consisting of two or more kinds of powders.Differences in the initial state of powders may be magnified after powder reuse, e.g. the powder composition, PSD, and other properties may undergo significant variations after recycling.By optimizing the sieving method, such as selecting the appropriate sieve size or sieving times, and then exploring the powder characteristics changing trend of different components of the mixed powder, the powder efficiency will be enhanced.(7) Up to now, the effect of powder reuse on the powder characteristics, printability, and mechanical performance mainly focuses on steels, Inconel alloys, and Ti alloys.
With the rapid development of alloy systems that can be processed via L-PBF, such as high entropy alloys [270], it is necessary to evaluate the powder reusability of these newly-emerged alloys.(8) Last but not least, the investigation on the evolution of chemical composition for reused powders is mainly focused on oxygen, nitrogen, and hydrogen, while the change of other elements is relatively undervalued.However, elements with high evaporation pressure, such as Zn and Mg, may also suffer from sublimation during powder reuse, which can affect printability and mechanical properties [94].

Figure 2 .
Figure 2. PSD required for different AM technologies, powder morphology of Haynes 282 (H282) superalloy, and flowability corresponding to different Hausner ratios.(a) Typical requirement of PSD for different AM technologies.Data of powder size for L-PBF, CS, and EBM is cited from[32], for DED is cited from[40], and for BJ is cited from[41].(b) Morphology of H282 powders measured by Malvern Morphologi G3 microscopes.Reprinted from[47], Copyright (2021), with permission from Elsevier.(c) Schematic graph of the relationship between Hausner ratio and powder flowability.The data is from[44].

Figure 6 .
Figure 6.Principle diagram of fluidized bed and comprehensive properties of the powders.(a) Schematic graph of fluidized bed and comparison of the raw Ti powders with the as-treated Ti powders under 450 • C for 10 min, properties of the raw and as-treated Ti powders as a function of temperature (b) PSD, and (c) flowability and oxygen content.(a)-(c) Reprinted from[22], Copyright (2019), with permission from Elsevier.(d) Representative tensile engineering stress-strain curves of the L-PBF pure Ti as a function of scanning speed at room temperature, (e) summary of ultimate tensile engineering stress versus engineering strain for the L-PBF pure Ti including the as-treated HDH powders with the optimal tensile properties, and other results using the GA powders with the oxygen levels of 0.12 wt%[150], 0.13 wt%[151], 0.17 wt%[152], and the HDH powders prepared by jet milling[153] and ball milling[79] with the oxygen levels of 0.17 wt% and 0.27 wt%, respectively.(d), (e) Reprinted from[24], Copyright (2023), with permission from Elsevier.

Figure 7 .
Figure 7. Preparation and properties of CMD powders.(a) CMD powders processing approaches and morphology, (b) PSD of CMD and GA powders, (c) aspect ratio of CMD powders, (d) aspect ratio of GA powders, (e) corresponding relationship between cohesive index and rotation speed of drum for CMD and GA powders, (f) tensile properties of L-PBF Al7075 alloy printed with CMD and GA powders.The mechanical performance of plate 7075 T6, L-PBF Scalmalloy, and AlSi10Mg are also shown for benchmarking.Reproduced from [27], with permission from Springer Nature.

Figure 8 .
Figure 8. Schematic diagram of powder mixing, and examples of mixing powders of different sizes of added powder.(a) Schematic diagram.Reprinted from [168], Copyright (2020), with permission from Elsevier.(b) Micron-sized Al7075 particles with nano-sized WC particles.Reproduced from [28], with permission from Springer Nature.(c) Elemental powder feedstock of micron-sized Al and submicron-sized Cu.Reprinted from [166], Copyright (2022), with permission from Elsevier.(d) Distribution of mixed micron-sized powders, (d-i) Ti-64 and Ti-5553 and (ii) CP-Ti, Ti-64 and Ti-5553.The circles in the (d-i) and (d-ii) represent the clusters of ingredient alloy.Reprinted from [167], Copyright (2020), with permission from Elsevier.

Figure 9 .
Figure 9.Influence of powder reusability on the powder characteristics and part properties.EBSD of powder microstructure is reprinted from[180], Copyright (2019), with permission from Elsevier.SEM of powder morphology is reproduced from[181].CC BY 4.0.EBSD of printed parts is reprinted from[182], Copyright (2021), with permission from Elsevier.SEM of the fatigue fracture surface is reprinted from[183], Copyright (2023), with permission from Elsevier.

Figure 13 .
Figure 13.Evolution trend of titanium alloy powder or part properties with powder reusing.(a) SEM of the edge of virgin (i) and reused (ii) Ti6Al4V powders.Reprinted from[245], Copyright (2023), with permission from Elsevier.(b) Cross-section of virgin (i) and reused (ii) powder after etching.Reproduced from[247], with permission from Springer Nature.(c) EBSD map and phase distribution of virgin (i and ii) and reused (iii and iv) samples.Reproduced from[248], with permission from Springer Nature.(d) Tensile properties, including yield strength (i) and elongation (ii).Reprinted from[242], Copyright (2020), with permission from Elsevier.(e) Strain-life comparison[196].Reproduced from[196], with permission from Springer Nature.(f) Log-normal distribution of fatigue lives.Reprinted from[183], Copyright (2023), with permission from Elsevier.(g) Charpy results.Reprinted from[245], Copyright (2023), with permission from Elsevier.

Figure 14 .
Figure 14.Evolution trend of aluminum alloy powder or part properties with powder reusing.(a) Surface morphology of virgin (i) and reused (ii) AlSi10Mg powders.Reprinted from[193], Copyright (2021), with permission from Elsevier.(b) Microstructure of virgin (i) and reused (ii) Scalmalloy samples.Reprinted from[31], Copyright (2020), with permission from Elsevier.(c) Analysis of the surface roughness of virgin and reused samples with respect to exposure time interval to salt spray test: (i) Ra (roughness average value) and (ii) Rt (roughness total height value).Reproduced from[258], with permission from Springer Nature.
denotes the increase of performance while ↓ denotes the decrease of performance.-denotesno significant change in performance.→ indicates that the PSD shifts to the right while ← denotes that the PSD shifts to the left.Process parameters include layer thickness (µm), laser power (W), scanning speed (mm•s −1 ), and hatch spacing (µm) in turn.Changes of chemical composition exclude the oxygen content, since it shows a consistent upward trend.

Table 1 .
Advantages and limitations on the method of powder characterization commonly used in AM.

Table 2 .
Materials, advantages, and limitations of the methods on powder manufacturing and preparation.

Table 3 .
Variation trend and mechanisms analysis of different powder properties.NA denotes that the mechanisms were not presented.