3D Printing of Highly Pure Copper

Copper has been widely used in many applications due to its outstanding properties such as malleability, high corrosion resistance, and excellent electrical and thermal conductivities. While 3D printing can offer many advantages from layer-by-layer fabrication, the 3D printing of highly pure copper is still challenging due to the thermal issues caused by copper’s high conductivity. This paper presents a comprehensive review of recent work on 3D printing technology of highly pure copper over the past few years. The advantages and current issues of 3D printing methods are compared while different properties of copper parts printed by these methods are summarized. Finally, we provide several potential applications of the 3D printed copper parts and an overview of current developments that could lead to new improvements in this advanced manufacturing field.


Introduction
During the last decades, 3D printing of metallic parts directly from 3D computer-aided design (CAD) models has drawn significant attention from both academia and industry [1][2][3][4][5][6]. One of the important advantages of 3D printing is the higher design freedom compared to conventional fabrication technologies [7][8][9][10]. In fact, complex 3D parts such as components with many internal structures or cellular parts can be fabricated by 3D printing without considering specific design rules. Moreover, the topological optimization of 3D CAD models with no additional cost in 3D printing provides many advantages such as complex shapes, short lead times, weight saving, and multifunctional integration [11]. Nowadays, many high-performance metallic alloys and pure metals can be processed successfully with achievable properties comparable to those fabricated by conventional methods such as forming or casting [1][2][3][4]. Additionally, parts with epitaxial growth and fine microstructures can be fabricated by process-inherent rapid and direct solidification in 3D printing. This offers a great potential in new design possibilities to achieve locally-desired material properties. Besides, the application of high cooling rates during 3D printing processes provides new possibilities in alloy and metal design [1][2][3][4].
3D printing of highly pure copper with superior electrical and thermal properties has been studied extensively due to its broad potential in many applications including electronic devices [1,[12][13][14][15][16][17][18][19][20], thermal management systems [4,12], and the aerospace industry [4,12,[21][22][23][24][25]. Compared to conventional  The build plate is usually connected to support structures which are necessary for the part's fixation in the powder bed and heat dissipation (Figure 1a, Detail A). The distortion of the printed parts can be avoided by pre-heating the build plate to lower thermal gradients and reduce residual stresses during the SLM process. Inert gas such as nitrogen or argon is continuously supplied to the building chamber to provide an inert atmosphere for protecting the metal powder and the heated metal parts from oxidation. Once the printing process is completed, the substrate plate can be removed from the printed part. Important process parameters such as layer thickness, hatch spacing, laser power, and scanning speed need to be optimized to fabricate high quality printed parts [1,3].

Indirect Selective Laser Melting
Badrinarayan et al. [49] first reported the indirect manufacturing of pure copper parts using the SLM method. A powder mixture of copper and poly(methyl methacrylate) (PMMA) was used as a feedstock for the printing process. Since PMMA acted as an intermediate binder, it was melted during the laser scanning process to bind the copper powders and form copper green parts. The green parts were debinded under reducing conditions afterwards and the resulting copper parts were sintered for several hours. However, the mixing ratio of PMMA in the fabrication process was too high (8 wt.%), leading to the low green density (under 27%) [49,50]. Consequently, the sintered copper parts had a density much lower than the theoretical density (about 77-88%) with the introduction of significant porosity. To address the issue, polymer-coated granules of small copper particles and bimodal mixed powders could be used to improve the powder packing density. Another drawback of the indirect SLM was that the impurities from the debinding process could be introduced to the final copper parts and might lower their properties [49,50].

Direct Selective Laser Melting
Most commercial SLM systems use continuous wave and long-pulse lasers working in the near infrared radiation (IR) with a wavelength of about 1 µm [51,52]. Although SLM has been demonstrated for many materials ranging from metals to polymers and ceramics [53][54][55][56][57][58], the processing of materials with high thermal conductivities and high melting points such as pure copper is still challenging [59,60]. In addition to the rapid heat dissipation problems, reflectivity of copper to conventional laser light near IR is very high [61][62][63], resulting in low deposition of laser energy in the materials for the melting process [51]. Therefore, to achieve high laser energy density for fabricating dense copper parts, greater laser output power and lower scanning speed, layer thickness, and hatch spacing are required [52,59,64].
Generally, the minimal laser power for a successful SLM of pure copper is approximately 300 W, although some studies have reported the fabrication of copper parts at lower laser powers [65,66]. Lykov et al. [65] studied the SLM process of pure copper powder using a system equipped with a 200 W CO 2 laser source. The process used a scanning speed ranging between 100 and 150 mm/s whereas the layer thickness and hatch spacing were fixed at 50 µm and 0.12 mm, respectively. Although the absorptance of pure copper powder for CO 2 laser was very low (only 0.26), the copper powder could be selectively melted since the laser beam diameter was adjusted to as small as 35 µm to obtain sufficient power density for the melting process. Consequently, the printed copper parts had dense structures with good surface finish quality and no dimensional distortions were observed. Specifically, the relative density of the printed parts could be as high as 88.1% while their tensile strength could reach 149 MPa [65].
In another approach, Kaden et al. [52] reported a successful SLM process of pure copper by using ultrashort laser pulses with 500 fs pulse duration at a center wavelength of 1030 nm. At a 20 MHz repetition rate, the lasers could provide extremely high pulse energy (1-1.5 µJ) and line energy (100 J/m), which were sufficient for copper melting. The powder layer thickness and laser spot size were controlled at 30 µm and of 35 µm, respectively, to fabricate both bulky and thin-walled copper structures with thicknesses below 100 µm (Figure 2). The scanning electron microscope (SEM) images in Figure 2b suggested that the printed parts had porous structures with melting beads formed during the powder melting. This might stem from the large powder size (35 µm) compared to the laser spot diameter (35 µm) employed in the process. Therefore, the use of copper powder with small grain sizes was recommended to reduce porous structures of the printed copper parts and improve their relative density [52]. laser spot diameter (35 µm) employed in the process. Therefore, the use of copper powder with small grain sizes was recommended to reduce porous structures of the printed copper parts and improve their relative density [52]. By using a laser source with much higher power, Ikeshoji et al. [64] could fabricate 3D copper parts with a relative density of up to 96.6% by SLM. In the fabrication process, the laser power was set at 800 W and the used hatch pitch was in a range of 0.025-0.12 mm. Additionally, the scanning speed and powder bed thickness were fixed at 300 mm/s and 0.05 mm, respectively. As can be seen in Figure 3a, the surfaces of all the printed cubes had rough texture and ballshaped asperities with no severe oxidation observed. These ball-shaped asperities might stem from the nonmolten residual powder particles while large asperities might be formed by splatter. Also, the surface texture suggested the presence of voids within the copper structures [64]. Due to the large asperities, smooth powder bed surfaces might not be achieved during the squeezing process. Internal voids could be formed at the shadow of the laser irradiation under the ball-like asperities and, therefore, lowered the relative density of the printed parts. As shown in Figure 3b, the relative density of the printed parts decreased at narrower hatch pitches, although the asperities could be exposed to the laser beam several times and melted down to fill the voids during the laser scanning process. Besides, transient heat transfer including melting and solidification was By using a laser source with much higher power, Ikeshoji et al. [64] could fabricate 3D copper parts with a relative density of up to 96.6% by SLM. In the fabrication process, the laser power was set at 800 W and the used hatch pitch was in a range of 0.025-0.12 mm. Additionally, the scanning speed and powder bed thickness were fixed at 300 mm/s and 0.05 mm, respectively. As can be seen in Figure 3a, the surfaces of all the printed cubes had rough texture and ballshaped asperities with no severe oxidation observed. These ball-shaped asperities might stem from the nonmolten residual powder particles while large asperities might be formed by splatter. Also, the surface texture suggested the presence of voids within the copper structures [64]. laser spot diameter (35 µm) employed in the process. Therefore, the use of copper powder with small grain sizes was recommended to reduce porous structures of the printed copper parts and improve their relative density [52]. By using a laser source with much higher power, Ikeshoji et al. [64] could fabricate 3D copper parts with a relative density of up to 96.6% by SLM. In the fabrication process, the laser power was set at 800 W and the used hatch pitch was in a range of 0.025-0.12 mm. Additionally, the scanning speed and powder bed thickness were fixed at 300 mm/s and 0.05 mm, respectively. As can be seen in Figure 3a, the surfaces of all the printed cubes had rough texture and ballshaped asperities with no severe oxidation observed. These ball-shaped asperities might stem from the nonmolten residual powder particles while large asperities might be formed by splatter. Also, the surface texture suggested the presence of voids within the copper structures [64]. Due to the large asperities, smooth powder bed surfaces might not be achieved during the squeezing process. Internal voids could be formed at the shadow of the laser irradiation under the ball-like asperities and, therefore, lowered the relative density of the printed parts. As shown in Figure 3b, the relative density of the printed parts decreased at narrower hatch pitches, although the asperities could be exposed to the laser beam several times and melted down to fill the voids during the laser scanning process. Besides, transient heat transfer including melting and solidification was Due to the large asperities, smooth powder bed surfaces might not be achieved during the squeezing process. Internal voids could be formed at the shadow of the laser irradiation under the ball-like asperities and, therefore, lowered the relative density of the printed parts. As shown in Figure 3b, the relative density of the printed parts decreased at narrower hatch pitches, although the asperities could be exposed to the laser beam several times and melted down to fill the voids during the laser scanning process. Besides, transient heat transfer including melting and solidification was analyzed by the finite element method (FEM) to study the effects of hatch pitches on the SLM process of pure copper. The various predicted dimensions of the melt pool with time indicated a slight overlap for wider hatch pitches and instability for narrower hatch pitches. This result suggested that the relative density of the printed copper parts was strongly related to the hatch pitch [64].
Later, a feasibility window of the SLM process of pure Cu powder was investigated by Colopi et al. [59] using a 1kW laser source. A total of 70 experimental conditions were conducted by varying the laser power, layer thickness, and scan speed in ranges of 200-1000 W, 50-100 µm, and 1000-4000 mm/s, respectively. The process outcome was classified into three categories: not fused (insufficient solidification), delamination (insufficient interlayer bonding), and acceptable (no evident defect). As shown in Figure 4a, the process using a layer thickness of 50 µm had the widest stability region, which might stem from its higher process resolution. analyzed by the finite element method (FEM) to study the effects of hatch pitches on the SLM process of pure copper. The various predicted dimensions of the melt pool with time indicated a slight overlap for wider hatch pitches and instability for narrower hatch pitches. This result suggested that the relative density of the printed copper parts was strongly related to the hatch pitch [64]. Later, a feasibility window of the SLM process of pure Cu powder was investigated by Colopi et al. [59] using a 1kW laser source. A total of 70 experimental conditions were conducted by varying the laser power, layer thickness, and scan speed in ranges of 200-1000 W, 50-100 µm, and 1000-4000 mm/s, respectively. The process outcome was classified into three categories: not fused (insufficient solidification), delamination (insufficient interlayer bonding), and acceptable (no evident defect). As shown in Figure 4a, the process using a layer thickness of 50 µm had the widest stability region, which might stem from its higher process resolution. The "delamination" defect occurred in both layer thicknesses, suggesting the crucial role of interlayer bonding on obtaining a stable powder bed densification. The occurrence of this defect at low laser power indicated the instability of the process melt pool (Figure 4a). Since the previously deposited copper layer had high thermal conductivity, thermal cooling could be elevated while melt pool stability might be hindered. Besides, the formation of the inter-layer defect might lead to thermally insulating gases in between the layers and, hence, limit the penetration of the melt pool to the previous interlayer defect. Consequently, the delamination issue might be further worsened due to the formation of thermally-induced stresses, resulting in more defect formation and part deformations [59].
As shown in Figure 4b,c, the relative density of the printed copper parts yielded up to 97.8% of the separate measurement between the core volume of sections and the border. The porosity in the border regions was always higher, which might be attributed to the temperature differences between the border and the central region. Due to thermal accumulation, most internal regions could reach higher temperatures compared to the border. Furthermore, the border was usually in contact with cooler powder in the powder bed during the scanning process. Since copper has very high thermal conductivity, strong thermal dissipation could be achieved in the border areas, resulting in their partial melting and higher porosity. The process parameters optimized the feasibility window which could be used to produce complex 3D copper parts, as shown in Figure 4c   The "delamination" defect occurred in both layer thicknesses, suggesting the crucial role of inter-layer bonding on obtaining a stable powder bed densification. The occurrence of this defect at low laser power indicated the instability of the process melt pool (Figure 4a). Since the previously deposited copper layer had high thermal conductivity, thermal cooling could be elevated while melt pool stability might be hindered. Besides, the formation of the inter-layer defect might lead to thermally insulating gases in between the layers and, hence, limit the penetration of the melt pool to the previous interlayer defect. Consequently, the delamination issue might be further worsened due to the formation of thermally-induced stresses, resulting in more defect formation and part deformations [59].
As shown in Figure 4b,c, the relative density of the printed copper parts yielded up to 97.8% of the separate measurement between the core volume of sections and the border. The porosity in the border regions was always higher, which might be attributed to the temperature differences between the border and the central region. Due to thermal accumulation, most internal regions could reach higher temperatures compared to the border. Furthermore, the border was usually in contact with cooler powder in the powder bed during the scanning process. Since copper has very high thermal conductivity, strong thermal dissipation could be achieved in the border areas, resulting in their partial melting and higher porosity. The process parameters optimized the feasibility window which could be used to produce complex 3D copper parts, as shown in Figure 4c [59]. Jadhav et al. [51] studied the influences of different laser scan parameters on the textural development of SLM-printed copper parts. The fabrication process used a 1 kW laser source, with a beam diameter of 40 µm, and a layer thickness of 30 µm. At optimum processing conditions, printed parts with a relative density of more than 98%, electrical conductivity of 88% IACS (International Annealed Copper Standard definition of 58 MS/m as 100% IACS for electrical conductivity), and thermal conductivity of up to 336 W/m.K could be produced (Figure 5a,b). Jadhav et al. [51] studied the influences of different laser scan parameters on the textural development of SLM-printed copper parts. The fabrication process used a 1 kW laser source, with a beam diameter of 40 µm, and a layer thickness of 30 µm. At optimum processing conditions, printed parts with a relative density of more than 98%, electrical conductivity of 88% IACS (International Annealed Copper Standard definition of 58 MS/m as 100% IACS for electrical conductivity), and thermal conductivity of up to 336 W/m.K could be produced (Figure 5a,b). The analysis of the crystallographic texture of the SLM copper parts suggested that the crystal orientation of the YZ and middle (XY-MID) plane was relatively random. This result could be attributed to the alteration in the heat gradient directions and subsequent re-melting which were formed by applying a 90° scan rotation strategy in between the subsequent layers. However, the top surface (XY-TOP) had strong crystallographic texture owing to the solidification morphology controlled by the temperature gradients within the melt pools. Due to the extremely high laser energy density and laser power used in the process, damages of the optical coating could be formed on the laser mirror, as shown in Figure 5d [51].

Electron Beam Melting
Similar to SLM, electron beam melting (EBM) is also a common powder bed-based process used to fabricate highly-dense metallic parts [67,68]. In the EBM process, metal powder from a hopper is supplied and spread uniformly on a build plate by a rake ( Figure 6) [1,69]. The powder layer thickness is usually in the range of 50-200 µm. According to the CAD data, the powder is melted selectively The analysis of the crystallographic texture of the SLM copper parts suggested that the crystal orientation of the YZ and middle (XY-MID) plane was relatively random. This result could be attributed to the alteration in the heat gradient directions and subsequent re-melting which were formed by applying a 90 • scan rotation strategy in between the subsequent layers. However, the top surface (XY-TOP) had strong crystallographic texture owing to the solidification morphology controlled by the temperature gradients within the melt pools. Due to the extremely high laser energy density and laser power used in the process, damages of the optical coating could be formed on the laser mirror, as shown in Figure 5d [51].

Electron Beam Melting
Similar to SLM, electron beam melting (EBM) is also a common powder bed-based process used to fabricate highly-dense metallic parts [67,68]. In the EBM process, metal powder from a hopper is supplied and spread uniformly on a build plate by a rake ( Figure 6) [1,69]. The powder layer thickness is usually in the range of 50-200 µm. According to the CAD data, the powder is melted selectively using an electron beam as a heat source. The electron beam is produced by an electron gun and accelerated with a high acceleration voltage. Then, electromagnetic lenses are used to focus the beam and a magnetic scan coil is employed to direct it to selected areas of the metal layers in the XY plane. Important EBM process parameters such as electron beam power, scan speed, and focus of the electron beam usually depend on the speed function, beam current, and focus offset, respectively [70].
At first, a defocused beam pre-heats the powder bed by scanning the powder bed surface several times. Then, a high beam current and proper scan speed are employed to heat and sinter the powder. Similar to the SLM process, the build plate in EBM is then lowered and metal powder is fed and spread uniformly across the work area. The process of powder deposition, powder pre-heating, electron beam scanning, and build plate lowering is repeated until the part printing is completed. The EBM process operates under a vacuum and helium is fed to the work area to prevent the metal powder from electrical charging while improving the heat conduction and cooling of the melt [1,71]. using an electron beam as a heat source. The electron beam is produced by an electron gun and accelerated with a high acceleration voltage. Then, electromagnetic lenses are used to focus the beam and a magnetic scan coil is employed to direct it to selected areas of the metal layers in the XY plane. Important EBM process parameters such as electron beam power, scan speed, and focus of the electron beam usually depend on the speed function, beam current, and focus offset, respectively [70]. At first, a defocused beam pre-heats the powder bed by scanning the powder bed surface several times. Then, a high beam current and proper scan speed are employed to heat and sinter the powder. Similar to the SLM process, the build plate in EBM is then lowered and metal powder is fed and spread uniformly across the work area. The process of powder deposition, powder pre-heating, electron beam scanning, and build plate lowering is repeated until the part printing is completed. The EBM process operates under a vacuum and helium is fed to the work area to prevent the metal powder from electrical charging while improving the heat conduction and cooling of the melt [1,71]. Since electrons have different absorption and reflection mechanisms compared to photons, the EBM method is not affected by the optical reflectivity of materials and most of the energy in EBM is deposited within the materials [1,4,40,72]. Therefore, this method has a high potential for the processing of pure copper [73,74]. Furthermore, the sintering step of copper during the EBM process can be extremely short due to its high electrical conductivity. In fact, the copper powder might not be removed if preheating is too high or too long because it has high sintering activity [1,4,40]. Yang et al. [28] reported the fabrication of pure copper with orthotropic re-entrant auxetic structures using EBM. Because of the high thermal conductivity of copper, the dissipation of thermal energy during the melting process was severe, resulting in large variations in the dimensions and mechanical performance of the printed copper parts. Therefore, more process optimization was required to improve the surface quality and accuracy of the printed parts.
Since copper is quite sensitive to oxidation, its storage and handling should be considered properly [75]. Ramirez et al. [76] investigated the effects of Cu2O on the microstructure of the EBMprinted copper components with the use of a relatively impure Cu precursor powder (98.5%). The study results showed the formation of precipitation-dislocation architectures because of directional Since electrons have different absorption and reflection mechanisms compared to photons, the EBM method is not affected by the optical reflectivity of materials and most of the energy in EBM is deposited within the materials [1,4,40,72]. Therefore, this method has a high potential for the processing of pure copper [73,74]. Furthermore, the sintering step of copper during the EBM process can be extremely short due to its high electrical conductivity. In fact, the copper powder might not be removed if preheating is too high or too long because it has high sintering activity [1,4,40]. Yang et al. [28] reported the fabrication of pure copper with orthotropic re-entrant auxetic structures using EBM. Because of the high thermal conductivity of copper, the dissipation of thermal energy during the melting process was severe, resulting in large variations in the dimensions and mechanical performance of the printed copper parts. Therefore, more process optimization was required to improve the surface quality and accuracy of the printed parts.
Since copper is quite sensitive to oxidation, its storage and handling should be considered properly [75]. Ramirez et al. [76] investigated the effects of Cu 2 O on the microstructure of the EBM-printed copper components with the use of a relatively impure Cu precursor powder (98.5%). The study results showed the formation of precipitation-dislocation architectures because of directional solidification during EBM (Figure 7a). Interestingly, these precipitation-dislocation architectures improved the hardness of the parts from HV 57 for a base-plate to HV 88 for EBM-printed products. Similar phenomenon was also observed in their later study of the open-cellular copper structures (Figure 7b) produced by EBM of higher purity copper powder (99.8%) [26].  A comprehensive performance of pure copper parts fabricated by EBM using three different copper powder configurations was reported by Frigola et al. [40]. The printed copper parts exhibited good properties with a relative density of 98.6%, electrical conductivity of 97% IACS, thermal conductivity of 390 W/mK, and yield strength of 76 MPa. More interestingly, the copper cathode fabricated by the EBM process showed good performance during high power radio frequency (RF) testing, which was comparable to other cathodes conventionally machined from wrought oxygen free copper material.
Later, a much higher relative density (up to 99.95%) of EBM-printed copper parts was achieved by Lodes et al. [77]. The results in Figure 8 show that the samples fabricated at higher line energies were denser for all beam velocities. Part density was lower at the very low line energy input because the powder could not be melted completely, resulting in layer defects and porous printed parts. However, the use of too great an energy input could lead to local overheating and swelling of the sample. Additionally, overheated copper powder could stick to the melt surface during raking, resulting in binding defects and porous areas. At optimum line energy, the built sample was almost fully dense with no binding defects observed. A comprehensive performance of pure copper parts fabricated by EBM using three different copper powder configurations was reported by Frigola et al. [40]. The printed copper parts exhibited good properties with a relative density of 98.6%, electrical conductivity of 97% IACS, thermal conductivity of 390 W/mK, and yield strength of 76 MPa. More interestingly, the copper cathode fabricated by the EBM process showed good performance during high power radio frequency (RF) testing, which was comparable to other cathodes conventionally machined from wrought oxygen free copper material.
Later, a much higher relative density (up to 99.95%) of EBM-printed copper parts was achieved by Lodes et al. [77]. The results in Figure 8 show that the samples fabricated at higher line energies were denser for all beam velocities. Part density was lower at the very low line energy input because the powder could not be melted completely, resulting in layer defects and porous printed parts. However, the use of too great an energy input could lead to local overheating and swelling of the sample. Additionally, overheated copper powder could stick to the melt surface during raking, resulting in binding defects and porous areas. At optimum line energy, the built sample was almost fully dense with no binding defects observed. Raab et al. [78] fabricated 3D copper parts with high relative densities (up to 99.95%) and good physical performance using EBM. At volume energies higher than 60 J/mm 3 , an electrical conductivity of up to 55.82 MS/m (96.24% IACS) and thermal conductivity of 400.1 W/mK could be achieved. The study results also suggested that pores and impurities such as oxides or carbides inside the printed copper parts might lower their electrical and thermal performance due to the reduced conduction cross-section. Later, Guschlbauer et al. [79] reported the fabrication of EBM-printed copper parts with excellent physical properties at different process parameters, as shown in Figure 9. At optimum conditions, the printed parts could obtain a relative density greater than 99.5%, an electrical conductivity of more than 58 MS/m (>100 IACS), a thermal conductivity of more than 411 W/mK, and tensile strength of 177 MPa. The uniform distribution of oxides in the copper matrix of the printed parts and no elevated oxygen content observed at the grain boundaries suggested that oxygen might have no significant role in crack formation during the tensile test. Instead, cracks might be generated by internal stresses formed during the manufacturing process [79].  Raab et al. [78] fabricated 3D copper parts with high relative densities (up to 99.95%) and good physical performance using EBM. At volume energies higher than 60 J/mm 3 , an electrical conductivity of up to 55.82 MS/m (96.24% IACS) and thermal conductivity of 400.1 W/mK could be achieved. The study results also suggested that pores and impurities such as oxides or carbides inside the printed copper parts might lower their electrical and thermal performance due to the reduced conduction cross-section. Later, Guschlbauer et al. [79] reported the fabrication of EBM-printed copper parts with excellent physical properties at different process parameters, as shown in Figure 9. At optimum conditions, the printed parts could obtain a relative density greater than 99.5%, an electrical conductivity of more than 58 MS/m (>100 IACS), a thermal conductivity of more than 411 W/mK, and tensile strength of 177 MPa. The uniform distribution of oxides in the copper matrix of the printed parts and no elevated oxygen content observed at the grain boundaries suggested that oxygen might have no significant role in crack formation during the tensile test. Instead, cracks might be generated by internal stresses formed during the manufacturing process [79]. Raab et al. [78] fabricated 3D copper parts with high relative densities (up to 99.95%) and good physical performance using EBM. At volume energies higher than 60 J/mm 3 , an electrical conductivity of up to 55.82 MS/m (96.24% IACS) and thermal conductivity of 400.1 W/mK could be achieved. The study results also suggested that pores and impurities such as oxides or carbides inside the printed copper parts might lower their electrical and thermal performance due to the reduced conduction cross-section. Later, Guschlbauer et al. [79] reported the fabrication of EBM-printed copper parts with excellent physical properties at different process parameters, as shown in Figure 9. At optimum conditions, the printed parts could obtain a relative density greater than 99.5%, an electrical conductivity of more than 58 MS/m (>100 IACS), a thermal conductivity of more than 411 W/mK, and tensile strength of 177 MPa. The uniform distribution of oxides in the copper matrix of the printed parts and no elevated oxygen content observed at the grain boundaries suggested that oxygen might have no significant role in crack formation during the tensile test. Instead, cracks might be generated by internal stresses formed during the manufacturing process [79]. Figure 9. (a) Processing window. Green marks represent cuboids higher >99.5% relative density, blue marks display relative densities lower <99.5% and red marks are dense but uneven cuboids due to too much energy, and (b) electrical conductivity of the cuboids in (MS/m). The black, solid lines represent constant line energies [79], with permission from Elsevier, 2019. Figure 9. (a) Processing window. Green marks represent cuboids higher >99.5% relative density, blue marks display relative densities lower <99.5% and red marks are dense but uneven cuboids due to too much energy, and (b) electrical conductivity of the cuboids in (MS/m). The black, solid lines represent constant line energies [79], with permission from Elsevier, 2019.

Binder Jetting
Binder jetting (BJ) is a 3D printing process in which one or two inkjet printheads are used to selectively deposit a liquid binding agent on the top of a powder bed to join powder materials [21,80,81]. The schematic of BJ process is shown in Figure 10 [21]. In Step 1, a roller is employed to spread metal powder uniformly across the platform to form a thin layer. Then, an inkjet printhead moves along the X-and Y-directions to selectively distribute binder droplets into the powder layer. The binder bonds the powder particles to produce a printed cross-sectional layer. Then, the build platform lowers (Z direction) for a small distance, the roller spreads a new powder layer, and the binder injection process repeats.

Binder Jetting
Binder jetting (BJ) is a 3D printing process in which one or two inkjet printheads are used to selectively deposit a liquid binding agent on the top of a powder bed to join powder materials [21,80,81]. The schematic of BJ process is shown in Figure 10 [21]. In Step 1, a roller is employed to spread metal powder uniformly across the platform to form a thin layer. Then, an inkjet printhead moves along the X-and Y-directions to selectively distribute binder droplets into the powder layer. The binder bonds the powder particles to produce a printed cross-sectional layer. Then, the build platform lowers (Z direction) for a small distance, the roller spreads a new powder layer, and the binder injection process repeats. When all the layers are deposited, the final glued part, usually called the ''green part'', is formed. Then, the binder is cured at elevated temperature to improve the green part strength for easy handling. The printed part experiences further sintering to achieve its final strength and density. During the sintering process, the binder is removed and the loosely packed powder is bonded together through powder sintering and densification. In the BJ process, many processing and postprocess sintering parameters such as the binder saturation, spreader speed, layer thickness, and sintering profile affect the final density and properties of the printed parts [82][83][84][85][86][87]. Compared to SLM and EBM, BJ has several unique characteristics for fabricating complex pure copper parts: i) no requirements of support structures, expensive energy sources, and enclosed chamber; ii) scalable technology; iii) no concerns over a material's reactivity, melting point, thermal conductivity, and optical reflectivity; iv) leveraged understanding of well-studied powder metallurgy processes due to the separation of part creation and powder sintering steps [21].
Bai et al. [21] first studied the 3D printing of pure copper via BJ of copper powders with median diameter of 75.2 µm, 15.3 µm, and 16.5 µm. Large powder size is usually preferred in the BJ process for powder spreading while small powder is crucial for sinterability, surface quality, and resolution. PM-B-SR2-05, a thermosetting polymer, was used as a binding agent in the process due to its minimal ash residue after debinding. The crucial printing parameters in the green part production were binder saturation ratio and layer thickness. The binder saturation ratio (the percentage of air space between the powder particles occupied by the binder volume) needs to be sufficient for the fabrication of a strong green part.
After sintering, green parts with a maximum relative density of 85.5% were achieved. The shrinkage and sintered density were found to be proportional to the temperature in a range of 1060-1090 °C. A purity of 97.3% and improvement of 25.3% in the sintered density could be achieved by sintering the printed parts in a reduced atmosphere, as shown in Figure 11. The printed sintered copper parts could only have a tensile strength of 116.7 MPa due to their high porosity. One main challenge of the BJ process is to improve the densification and part sinterability while satisfying the size requirements of the copper powder [21]. When all the layers are deposited, the final glued part, usually called the "green part", is formed. Then, the binder is cured at elevated temperature to improve the green part strength for easy handling. The printed part experiences further sintering to achieve its final strength and density. During the sintering process, the binder is removed and the loosely packed powder is bonded together through powder sintering and densification. In the BJ process, many processing and post-process sintering parameters such as the binder saturation, spreader speed, layer thickness, and sintering profile affect the final density and properties of the printed parts [82][83][84][85][86][87]. Compared to SLM and EBM, BJ has several unique characteristics for fabricating complex pure copper parts: (i) no requirements of support structures, expensive energy sources, and enclosed chamber; (ii) scalable technology; (iii) no concerns over a material's reactivity, melting point, thermal conductivity, and optical reflectivity; (iv) leveraged understanding of well-studied powder metallurgy processes due to the separation of part creation and powder sintering steps [21].
Bai et al. [21] first studied the 3D printing of pure copper via BJ of copper powders with median diameter of 75.2 µm, 15.3 µm, and 16.5 µm. Large powder size is usually preferred in the BJ process for powder spreading while small powder is crucial for sinterability, surface quality, and resolution. PM-B-SR2-05, a thermosetting polymer, was used as a binding agent in the process due to its minimal ash residue after debinding. The crucial printing parameters in the green part production were binder saturation ratio and layer thickness. The binder saturation ratio (the percentage of air space between the powder particles occupied by the binder volume) needs to be sufficient for the fabrication of a strong green part.
After sintering, green parts with a maximum relative density of 85.5% were achieved. The shrinkage and sintered density were found to be proportional to the temperature in a range of 1060-1090 • C. A purity of 97.3% and improvement of 25.3% in the sintered density could be achieved by sintering the printed parts in a reduced atmosphere, as shown in Figure 11. The printed sintered copper parts could only have a tensile strength of 116.7 MPa due to their high porosity. One main challenge of the BJ process is to improve the densification and part sinterability while satisfying the size requirements of the copper powder [21]. In a later study, Bai et al. [80] reported the benefits of using bimodal powder mixtures in the BJ of copper parts. In the process, five powder mixtures formed from copper powder with different diameters were employed as the process feedstock. The study results showed that the bimodal powder mixtures could improve the powder packing density by 8.2% and powder flowability by 10.5%, compared to the monosized powder counterparts. More importantly, the printed green part and the sintered density could increase by up to 9.4% and 12.3%, respectively, while a decrease of 6.4% in sintering shrinkage was observed due to the use of the bimodal powder mixtures. Another benefit was that less energy input might be required for the process using the bimodal powder mixtures as it was less sensitive to the sintering conditions.
Since coarse powder and loosely packed powder beds are used in the BJ process, the final printed copper parts usually contain high porosity with low mechanical properties. One effective approach to address this issue is to apply post-process heat treatments, such as hot isostatic pressing (HIP) [88,89]. Kumar et al. [36] applied HIP to the final printed copper parts with three different dimensions (Type A: 16 mm × 16 mm × 4 mm type B: 32 mm × 8 mm × 4 mm, and type C: 16 mm × 8 mm × 4 mm). As seen in Figure 12a, the relative density of the sintered parts could reach 99.47% after HIP treatment. This result was supported by a significant decrease in pores observed in the sectioned parts shown in Figure 12b, suggesting that HIP could densify the printed copper parts to nearly full density. In a later study, Bai et al. [80] reported the benefits of using bimodal powder mixtures in the BJ of copper parts. In the process, five powder mixtures formed from copper powder with different diameters were employed as the process feedstock. The study results showed that the bimodal powder mixtures could improve the powder packing density by 8.2% and powder flowability by 10.5%, compared to the monosized powder counterparts. More importantly, the printed green part and the sintered density could increase by up to 9.4% and 12.3%, respectively, while a decrease of 6.4% in sintering shrinkage was observed due to the use of the bimodal powder mixtures. Another benefit was that less energy input might be required for the process using the bimodal powder mixtures as it was less sensitive to the sintering conditions.
Since coarse powder and loosely packed powder beds are used in the BJ process, the final printed copper parts usually contain high porosity with low mechanical properties. One effective approach to address this issue is to apply post-process heat treatments, such as hot isostatic pressing (HIP) [88,89]. Kumar et al. [36] applied HIP to the final printed copper parts with three different dimensions (Type A: As seen in Figure 12a, the relative density of the sintered parts could reach 99.47% after HIP treatment. This result was supported by a significant decrease in pores observed in the sectioned parts shown in Figure 12b, suggesting that HIP could densify the printed copper parts to nearly full density. In a later study, Kumar et al. [90] investigated the combined effects of bimodal powder mixtures and HIP treatment on the BJ printed copper parts. In the process, the copper parts were printed from three powder configurations: two with 17 and 25 µm powders and one with a bimodal powder mixture of 5 and 30 µm powders. The improvement in density shown in Figure 13a indicated that the printed parts with bimodal configuration had the best densification effects with a relative density of 97.32% after HIP. The results also suggested that HIP was only an effective post-treatment when the sintered BJ parts had a minimum relative density of 90%, which could be achieved by using bimodal powder mixtures. The relative density of 99.47% in their previously study [36] could not be replicated due to the problems with clogged inkjet nozzles and differences in the powder quality. More importantly, due to the reduction in porosity after HIP application, both tensile strength and ductility of the bimodal samples increased from 144.9 to 176.35 MPa and from 18.81% to 31.28%, respectively. In a later study, Kumar et al. [90] investigated the combined effects of bimodal powder mixtures and HIP treatment on the BJ printed copper parts. In the process, the copper parts were printed from three powder configurations: two with 17 and 25 µm powders and one with a bimodal powder mixture of 5 and 30 µm powders. The improvement in density shown in Figure 13a indicated that the printed parts with bimodal configuration had the best densification effects with a relative density of 97.32% after HIP. The results also suggested that HIP was only an effective post-treatment when the sintered BJ parts had a minimum relative density of 90%, which could be achieved by using bimodal powder mixtures. The relative density of 99.47% in their previously study [36] could not be replicated due to the problems with clogged inkjet nozzles and differences in the powder quality. More importantly, due to the reduction in porosity after HIP application, both tensile strength and ductility of the bimodal samples increased from 144.9 to 176.35 MPa and from 18.81% to 31.28%, respectively. The effects of binders on the BJ process of pure copper were studied by Bai et al. [91] using two different nanoparticle binder systems: colloidal organic binder (a conventional polymer binder with suspended copper nanoparticles) and inorganic nanosuspension (an inorganic copper nanosuspension ink). They found that the jetted nanoparticles could produce denser green parts. However, there were no significant differences in the sintered density compared to copper parts printed by conventional organic binder. This might stem from the offset of the reduced sintering densification to the density obtained in the green part. However, the inkjetted nanoparticles could reduce the grain size of the sintered parts by hindering the grain growth during the sintering process, leading to their improved tensile strength. Moreover, the use of these metal binders could enhance the metal purity and structural integrity of the sintered parts. Besides, a higher weight percentage of copper nanoparticles in the printed copper parts could be achieved at higher binder saturation ratios, resulting in better green part strength as well as higher green part and sintered part density.
In another study, Bai et al. [92] used metal-organic-decomposition (MOD) ink as a binder in the BJ process of copper. As the MOD ink is particle-free [93,94] it can avoid the common issues found in the BJ process using particle suspensions such as particle oxidation, sedimentation, and nozzle clog problems. To validate this concept, copper organometallic complex with a metal content of 7.4 wt.% was synthesized for BJ printing of pure copper (Figure 14a). During the fabrication process, an overhead heater was used to dry the MOD ink between layers by removing the solvent. Then, copper nanoparticles were precipitated and sintered from organometallic complex to bond the copper powder when the printed parts were cured in a reducing atmosphere (Figure 14b). However, the green parts using MOD ink were weaker than those printed with the conventional polymeric binder. Although the sintered copper parts printed by MOD ink had a denser core section, their outer shell was more porous. This porous shell may stem from the powder loss due to weak powder bonding. Consequently, the relative density of the copper parts printed by MOD ink was lower than that of the conventional binder counterparts. The effects of binders on the BJ process of pure copper were studied by Bai et al. [91] using two different nanoparticle binder systems: colloidal organic binder (a conventional polymer binder with suspended copper nanoparticles) and inorganic nanosuspension (an inorganic copper nanosuspension ink). They found that the jetted nanoparticles could produce denser green parts. However, there were no significant differences in the sintered density compared to copper parts printed by conventional organic binder. This might stem from the offset of the reduced sintering densification to the density obtained in the green part. However, the inkjetted nanoparticles could reduce the grain size of the sintered parts by hindering the grain growth during the sintering process, leading to their improved tensile strength. Moreover, the use of these metal binders could enhance the metal purity and structural integrity of the sintered parts. Besides, a higher weight percentage of copper nanoparticles in the printed copper parts could be achieved at higher binder saturation ratios, resulting in better green part strength as well as higher green part and sintered part density.
In another study, Bai et al. [92] used metal-organic-decomposition (MOD) ink as a binder in the BJ process of copper. As the MOD ink is particle-free [93,94] it can avoid the common issues found in the BJ process using particle suspensions such as particle oxidation, sedimentation, and nozzle clog problems. To validate this concept, copper organometallic complex with a metal content of 7.4 wt.% was synthesized for BJ printing of pure copper (Figure 14a). During the fabrication process, an overhead heater was used to dry the MOD ink between layers by removing the solvent. Then, copper nanoparticles were precipitated and sintered from organometallic complex to bond the copper powder when the printed parts were cured in a reducing atmosphere (Figure 14b). However, the green parts using MOD ink were weaker than those printed with the conventional polymeric binder. Although the sintered copper parts printed by MOD ink had a denser core section, their outer shell was more porous. This porous shell may stem from the powder loss due to weak powder bonding. Consequently, the relative density of the copper parts printed by MOD ink was lower than that of the conventional binder counterparts. Figure 14. (a) Comparison between jetting a nanoparticle suspension (with particles) and metalorganic-decomposition (MOD) ink (particle-free) as the binder and (b) SEM images of the powder particle formed by sintered copper nanoparticles [92], with permission from Elsevier, 2019.

Ultrasonic Additive Manufacturing (UAM)
Ultrasonic additive manufacturing (UAM) is a manufacturing method using thin metallic foils or tapes as feedstock to fabricate 3D parts [95][96][97][98]. In this method, an ultrasonic wave and mechanical pressure are applied on metallic tapes at room temperature to bond the interfaces of the stacked tapes by diffusion [99]. 3D parts can be built by bonding the stacked tapes layer-by-layer without the use of any heat source. The metallic tapes are usually cut into designed shapes before ultrasonic consolidation bonding. To achieve good finishing, polishing is usually applied during or after the consolidation process. Figure 15 illustrates the working process of an UAM equipment [95].  (a) Comparison between jetting a nanoparticle suspension (with particles) and metal-organic-decomposition (MOD) ink (particle-free) as the binder and (b) SEM images of the powder particle formed by sintered copper nanoparticles [92], with permission from Elsevier, 2019.

Ultrasonic Additive Manufacturing (UAM)
Ultrasonic additive manufacturing (UAM) is a manufacturing method using thin metallic foils or tapes as feedstock to fabricate 3D parts [95][96][97][98]. In this method, an ultrasonic wave and mechanical pressure are applied on metallic tapes at room temperature to bond the interfaces of the stacked tapes by diffusion [99]. 3D parts can be built by bonding the stacked tapes layer-by-layer without the use of any heat source. The metallic tapes are usually cut into designed shapes before ultrasonic consolidation bonding. To achieve good finishing, polishing is usually applied during or after the consolidation process. Figure 15 illustrates the working process of an UAM equipment [95]. (a) Comparison between jetting a nanoparticle suspension (with particles) and metalorganic-decomposition (MOD) ink (particle-free) as the binder and (b) SEM images of the powder particle formed by sintered copper nanoparticles [92], with permission from Elsevier, 2019.

Ultrasonic Additive Manufacturing (UAM)
Ultrasonic additive manufacturing (UAM) is a manufacturing method using thin metallic foils or tapes as feedstock to fabricate 3D parts [95][96][97][98]. In this method, an ultrasonic wave and mechanical pressure are applied on metallic tapes at room temperature to bond the interfaces of the stacked tapes by diffusion [99]. 3D parts can be built by bonding the stacked tapes layer-by-layer without the use of any heat source. The metallic tapes are usually cut into designed shapes before ultrasonic consolidation bonding. To achieve good finishing, polishing is usually applied during or after the consolidation process. Figure 15 illustrates the working process of an UAM equipment [95].  The building process starts with the laying out and stacking of a metallic tape on a base plate. Then, a sonotrode applies ultrasonic vibrations laterally to the tape when it rolls along the tape length [99]. This bonds the metallic tape to the previously built parts to form a seam weld. During this process, the rapid "scrubbing action" between the faying surfaces generates the bonding. Typically, the first tape is welded onto the base plate, followed by the bonding of the next tape. After all the layers are built, the final part is cut from the base plate and polished for desired surface finishing. Therefore, the UAM process can produce metallic parts of complex shapes and designs. During the bonding process, frictional heat at the bonded interfaces is produced, leading to the increased temperature of the consolidated region. Therefore, a short cooling period between the manufacturing of each layer is required to avoid thermal residual stress [95].
Welding copper tapes are usually difficult due to material hardening and oxidation issues, leading to the requirement of high power used in the process (up to 9 kW) [95,96,100]. Sriraman et al. [96] studied the bonding characteristics formed during a very high-power UAM of copper tapes. In the study, C11000 copper tapes (170 mm × 25 mm × 0.15 mm) were successively bonded via ultrasonic seam welding at room temperature. The welding process was conducted at an amplitude of 36 µm, static force of 6.7 kN, and travel speed of 30 mm/s. The final built part was comprised of 13 layers to reach a height of about 2 mm. Figure 16a shows the interfaces of the built sample with wavy features. It might stem from the violent plastic flow of the material which was formed along the ridges and valleys between the contacting layers during the scrubbing process. Hardness distribution shown in Figure 16b suggests that the material might be uniformly softened and the hardness values throughout the built part were about 11-23% lower than those of the as-received foils [96]. The building process starts with the laying out and stacking of a metallic tape on a base plate. Then, a sonotrode applies ultrasonic vibrations laterally to the tape when it rolls along the tape length [99]. This bonds the metallic tape to the previously built parts to form a seam weld. During this process, the rapid "scrubbing action" between the faying surfaces generates the bonding. Typically, the first tape is welded onto the base plate, followed by the bonding of the next tape. After all the layers are built, the final part is cut from the base plate and polished for desired surface finishing. Therefore, the UAM process can produce metallic parts of complex shapes and designs. During the bonding process, frictional heat at the bonded interfaces is produced, leading to the increased temperature of the consolidated region. Therefore, a short cooling period between the manufacturing of each layer is required to avoid thermal residual stress [95].
Welding copper tapes are usually difficult due to material hardening and oxidation issues, leading to the requirement of high power used in the process (up to 9 kW) [95,96,100]. Sriraman et al. [96] studied the bonding characteristics formed during a very high-power UAM of copper tapes. In the study, C11000 copper tapes (170 mm × 25 mm × 0.15 mm) were successively bonded via ultrasonic seam welding at room temperature. The welding process was conducted at an amplitude of 36 µm, static force of 6.7 kN, and travel speed of 30 mm/s. The final built part was comprised of 13 layers to reach a height of about 2 mm. Figure 16a shows the interfaces of the built sample with wavy features. It might stem from the violent plastic flow of the material which was formed along the ridges and valleys between the contacting layers during the scrubbing process. Hardness distribution shown in Figure 16b suggests that the material might be uniformly softened and the hardness values throughout the built part were about 11-23% lower than those of the as-received foils [96]. As seen in Figure 16c, the grain size of the copper foils at the foil interface reduced from 25 to 10 µm after the processing [96]. Moreover, a larger fraction of grains with high-angle misorientations were observed in the built part (Figure 16d). The findings indicated that there was recrystallization As seen in Figure 16c, the grain size of the copper foils at the foil interface reduced from 25 to 10 µm after the processing [96]. Moreover, a larger fraction of grains with high-angle misorientations were observed in the built part (Figure 16d). The findings indicated that there was recrystallization followed by the movement of grain boundaries across the interfaces, resulting in metallurgical bonding between the layers. Additionally, the initial coarse grain structure might experience plastic deformation and recovery during the welding process. Their later study on the thermal transients during the process suggested a strong correlation between the interfacial temperature and plastic deformation heating [95]. Moreover, vibration amplitude had strongest effects on interfacial temperatures as dynamic plastic shear strains at the asperities were higher with increasing amplitudes. Although the UAM process could produce fully dense parts, void defects might be formed within the material's internal structure during the fabrication process, lowering its electrical and mechanical performance. Table 1 compares the characteristics of all 3D printing methods of pure copper. In both direct SLM and EBM, high energy power is required to heat the copper powder to its melting point for part fabrication. However, the required energy of direct SLM is much higher than that of EBM due to the low energy absorption of pure copper with conventional lasers (wavelength of 1 µm) [41,59,[101][102][103]. Moreover, the high reflected laser radiation in direct SLM of pure copper can damage the system's components. These issues can be addressed by (i) improving the optical absorption of copper by using a green laser with a shorter wavelength (515 nm) or chemically modifying the surface of the copper powder and melt; and (ii) applying advanced multi-variate ray tracings to eliminate the back focal points on the locations of the laser mirrors [51]. The required energy in the EBM of copper is lower than that in direct SLM because optical reflectivity of the materials do not affect the EBM process [4,40]. However, the EBM beam size is larger than the laser spot size in direct SLM, limiting the fine features size of the EBM-printed parts. Due to rapid melting/solidification of copper powder, the copper parts fabricated by both SLM and EBM have many issues in the full melting process, such as thermal residual stresses and thermally-induced deformation [2,27]. These issues can be minimized by properly designed anchors or support structures which dissipate heat and prevent part deformation [1,41]. Table 2 compares the relative density, electrical, mechanical and thermal properties of the printed copper parts fabricated by different methods. Performance of the parts fabricated by UAM is not presented because no data are available in the literature. As can be seen, EBM can fabricate copper parts with the best value of relative density (99.95%), electrical conductivity (102% IACS), tensile strength (177 MPa), and thermal conductivity (411.89 W/mK). In contrast, copper parts from indirect SLM have the poorest performance with a relative density of 84.8% and strength of only 8 MPa. Interestingly, parts printed by JB have a high relative density and strength of 99.7% and 176.35 MPa, respectively, but no data for electrical and thermal conductivity are available. For direct SLM, the printed parts have a high relative density (99.6%) and good physical performance with an electrical conductivity of 88% IACS, tensile strength of 149 MPa, and thermal conductivity of 336 W/mK.
Heat transfer components such as micro reactors, radiators, exchangers, or coolers could have improved function and efficiency by additive manufacturing approaches. An effective heat transfer in a liquid cooling system usually has a reliable and complex structure with high surface-volume ratio so that the flowing liquid can have many contacts with the surface area of the components. Moreover, their structure needs to be reliable to provide high pressure and resistance against fluid crossing between channels or fluid leakage. Notably, heat exchangers fabricated by conventional processes such as soldering or diffusion bonding are costly and insufficiently reliable because of their complex assembly [12,13]. Due to the single-step process of 3D printing, the printed components have no joints and, therefore, lower the risk of leakage [12,13]. Moreover, higher efficiency of the heat transfer can be achieved by maximizing the surface-volume ratio because many complex internal structures or intricate channels in confined space can be designed and fabricated by 3D printing, as shown in Figure 17a [13]. Furthermore, many materials with high thermal conductivity such as copper or copper alloys can be used for 3D printing of heat exchangers. Figure 17b presents an example of a compact cooler unit fabricated by SLM. The component (46 mm × 59 mm × 15 mm) has many cooling fins with a thickness of 0.3 mm to ensure efficient heat exchange. A high surface-volume ratio and turbulent flow of the coolant are achieved by optimizing the ribbing within the component [13]. In another approach, open-cellular copper structures with densities ranging from 0.73 to 6.67 g/cm 3 have been fabricated by EBM, as shown in Figure 17c [26]. The printed structures had hardness values of more than 75% of commercial Cu parts, exhibiting great potential for complex, multi-functional electrical and thermal management systems and heat exchange devices. Figure 17. (a) Heat exchanger designs with extremely intricate inner geometry [13], (b) miniature cooler [13], and (c) reticulated copper mesh fabricated by EBM [26], with permission from Emerald Publishing Limited, 2019.

Induction Heat Coil
Generally, induction heat coils have various shapes and sizes which are often highly customized for different applications [27]. Moreover, they usually have complex structures with curved paths or helical shapes and hollow feature for water cooling. Therefore, the production process of induction coils is usually tedious and time-consuming because they are usually fabricated manually with customized design. These features make induction coils a good candidate for 3D printing due to its advantages of fabricating complex and highly customized parts with little lead time. Induction coils are usually made of highly pure copper because the coil materials must have high efficiency fpr conducting electricity. Martin et al. [27] suggested that direct SLM, EBM and UAM are suitable for fabricating induction coils with higher precision compared to typical hand-made coils.

Radio Frequency Cathode
3D printing can be used to fabricate photoinjector cathodes with high RF performance. Frigola et al. [40,107] tested RF performance of the EBM printed copper cathode in a Pegasus 1.6 cell photoinjector. No other heat treatment was performed on the printed cathode before the final machining. The study results showed that during high power RF testing, the EBM printed copper cathodes had similar performance to wrought oxygen-free (OFE) copper cathodes fabricated by conventional methods. During two hours of RF conditioning, a peak electric field of 70 MV/m, a photoelectron beam with a charge of 60 pC, energy of 3.3 MeV, and quantum efficiency of ~2 × 10 −5 were achieved with stable operation.

Conclusions and Future Direction
This paper presents a comprehensive review of research on the 3D printing methods for fabricating highly pure copper, including SLM, EBM, BJ and UAM. The advantages, challenges, and performance of the copper parts fabricated by each method have been identified and compared. Additionally, potential applications of the 3D printed copper parts in thermal management systems, heat exchange devices, RF cathodes, and induction heat coils have been demonstrated. Figure 17. (a) Heat exchanger designs with extremely intricate inner geometry [13], (b) miniature cooler [13], and (c) reticulated copper mesh fabricated by EBM [26], with permission from Emerald Publishing Limited, 2019.

Induction Heat Coil
Generally, induction heat coils have various shapes and sizes which are often highly customized for different applications [27]. Moreover, they usually have complex structures with curved paths or helical shapes and hollow feature for water cooling. Therefore, the production process of induction coils is usually tedious and time-consuming because they are usually fabricated manually with customized design. These features make induction coils a good candidate for 3D printing due to its advantages of fabricating complex and highly customized parts with little lead time. Induction coils are usually made of highly pure copper because the coil materials must have high efficiency fpr conducting electricity. Martin et al. [27] suggested that direct SLM, EBM and UAM are suitable for fabricating induction coils with higher precision compared to typical hand-made coils.

Radio Frequency Cathode
3D printing can be used to fabricate photoinjector cathodes with high RF performance. Frigola et al. [40,107] tested RF performance of the EBM printed copper cathode in a Pegasus 1.6 cell photoinjector. No other heat treatment was performed on the printed cathode before the final machining. The study results showed that during high power RF testing, the EBM printed copper cathodes had similar performance to wrought oxygen-free (OFE) copper cathodes fabricated by conventional methods. During two hours of RF conditioning, a peak electric field of 70 MV/m, a photoelectron beam with a charge of 60 pC, energy of 3.3 MeV, and quantum efficiency of~2 × 10 −5 were achieved with stable operation.

Conclusions and Future Direction
This paper presents a comprehensive review of research on the 3D printing methods for fabricating highly pure copper, including SLM, EBM, BJ and UAM. The advantages, challenges, and performance of the copper parts fabricated by each method have been identified and compared. Additionally, potential applications of the 3D printed copper parts in thermal management systems, heat exchange devices, RF cathodes, and induction heat coils have been demonstrated.
In the near future, the 3D printing of pure copper will most likely grow further to address all the current issues of the printing methods while exploring more potential applications of the printed copper parts. For example, the TruPrint Laser Metal Fusion (LMF) 3D printer equipped with green laser source has been developed to overcome the low absorption of pure copper to conventional laser sources and component damages [108]. Heraeus, a German technology group, has addressed the technical issues and optimized the process of 3D printing highly-conductive copper components, which opens up new industrial fields of application such as in mobility, electronics, robotics, hydraulics, medicine and aerospace [109].