Copper alloys for additive manufacturing: Laser powder bed fusion of CuCr1Zr by using a green qcw-laser

Abstract Due to the high electrical conductivity in combination with high strength, the copper alloy CuCr1Zr plays an important role for industrial applications. Here, additive manufacturing of CuCr1Zr enables new fields of application due to a higher degree of freedom, as well as the materials design flexibility. However, the high thermal conductivity and the low absorptivity of the laser radiation of copper reduce the efficiency of the additive manufacturing process significantly. An improvement of the manufacturing process can be achieved by using a green laser source with a wavelength of 532 nm. The first parameter study via a design of experiments approach for the laser power bed fusion process of CuCr1Zr by using a green quasi continuous-wave laser is presented in this work. A maximal relative density of 99.6% was obtained for a laser power of 125 W, a scanning speed of 400 mm/s, and a hatching distance of 100 µm. A strong dependency of the relative density on the laser power and the hatching distance was observed. Moreover, the microstructure evolution comes along with the occurrence of zirconium oxides within the powder as well as within the additively manufactured samples.


Introduction
Due to a high thermal and electrical conductivity, technically pure copper (Cu), like CU-HCP or Cu-ETP, is used in many industrial applications such as electronics or aerospace industry (Colopi, Demir, Caprio, & Previtali, 2019).However, the low strength of copper causes limitations by loaded conditions, as in spring and contact applications (Deutsches Kupfer Institut Berufsverband e.V., 2012).Hereby, the strength of copper can be significantly increased via precipitation hardening (Tenwick & Davies, 1988).One promising age-hardenable copper alloy is CuCr1Zr, which features both, a high conductivity and sufficient strength (Artzt et al., 2020;Tenwick & Davies, 1988;Uhlmann et al., 2016).The rising strength results from precipitation of chromium solid solution (Cr), Cu 5 Zr or Cr 2 Zr during the aging process (Becker, 2014;Deutsches Kupfer Institut Berufsverband e.V., 2005;Tenwick & Davies, 1988).Based on the good ratio between conductivity and strength, CuCr1Zr can be used, e.g., for welding nozzles, or as current-carrying plugs (Deutsches Kupfer Institut Berufsverband e.V., 2005).Besides the material properties, applications of copper are determined by the available manufacturing routines.New production methods like additive manufacturing offer an opportunity to expand the scope of copper materials.Based on the higher geometric degree of freedom as well as the adjustment of material properties, laser powder bed fusion (LPBF) is suitable for a new geometrical and materials design of copper and its alloys.The provision of a suitable laser source for melting the copper powder represents one of the current problems for LPBF of copper.Common LPBF machines use a red laser with a wavelength of 1063 nm.However, the degree of absorption of copper in this range of wavelengths is only about 2% of a smooth surface (Becker, 2014;Blom et al., 2003).Consequently, only a low amount of the energy emitted by the laser source is effectively used for the required melting of powder particles.Therefore, common LPBF machines with a red laser require a high laser power up to 1 kW for the additive manufacturing of Cu (Colopi, Caprio, Demir, & Previtali, 2018;Colopi et al., 2019;Guschlbauer, Momeni, Osmanlic, & Körner, 2018;Ikeshoji, Nakamura, Yonehara, Imai, & Kyogoku, 2018).The energy transfer is supported by a multiple reflection at the rough powder bed surface and the temperature-dependent properties of heat conductivity and the absorptivity (Becker, 2014;Blom et al., 2003).Nevertheless, at a wavelength of 532 nm (green laser radiation), the absorptivity of copper increases to 40% (Becker, 2014;Sugioka, Meunier, & Piqué, 2010) and offers a higher energy input into the material.However, using a green laser within a LPBF machine does not represent the industrial standard.The reported relative density of CuCr1Zr with application of a red continuous wave laser (cw-laser) lies in the range between 97.65% and 99.8% (Becker, 2014;Guan et al., 2019;Jahns, Bappert, Böhlke, & Krupp, 2020;Uhlmann et al., 2016;Wallis & Buchmayr, 2019), whereas LPBF of CuCr1Zr using a green laser has not been reported yet.The present study should close this research gap by performing a parameter study with a green quasi continuous wave laser (qcw-laser) for CuCr1Zr and a subsequent investigation of the resulting microstructure.

Gas atomization
The production of CuCr1Zr powder is performed via gas atomization.An Indutherm AU3000 gas atomizer with a close-coupled application was used.
For CuCr1Zr powder production, a mixture of pure copper, zirconium (Zr), and CuCr10 was heated up inductively to 1,450 °C in a graphite crucible and atomized with a pressure of 10 bar.Afterwards, the powder was prepared via sieving (mesh size 63 µm) and air separation, resulting in a particle size distribution in the range of 10-60 µm.The particle size distribution was verified via dynamic image analysis using a Camsizer X2.The chemical composition of the powder was measured via wet chemical analysis using optical emission spectroscopy (Agilent 5110 ICP-OES) and the remaining oxygen concentration was measured by carrier gas hot extraction process with LECO RO 41DR.
The produced powder was analyzed with an analytical high-resolution scanning electron microscope (SEM, Zeiss Auriga), including energy dispersive X-ray spectroscopy (EDX, Oxford EDX SDD) and electron backscatter diffraction (EBSD, Aztec/Channel 5 by Oxford).Here, samples were ground with silicon carbide paper to a surface roughness of 18 µm.The following polishing involved a diamond suspension with diamond size of 9 µm, 3 µm, and 1 µm.Finally, the samples were treated with a colloidal silica suspension with a particle size of 0.06 µm and a pH-value of 10.This preparation was also applied for the LPBF samples.

Laser powder bed fusion
The CuCr1Zr powder was processed further by LPBF (AMCM EOS M290 customized) with a green qcw-laser.The pulse duration of laser is 1.2 ns and the repetition frequency is 125 MHz.Consequently, the powder bed is exposed each 8 ns for 1.2 ns, respectively.
Due to the research gap of LPBF with a green qcw-laser, a design of experiments (DOE) approach was implemented by using the Software Minitab.Here, an effect analysis was performed via a screening routine with the parameters laser power (P L ), scanning speed (v s ), and hatching distance (h s ).Thus, the process window is subdivided in three categories (see Figure 1).The first category marks the operating window.The process validation contains parameter combinations outside of the operating window and is termed as star points.For the screening routine, the relative density of 99.5% serves as the command variable.The DOE analysis results from the evaluation of the standardized effect of the selected parameter and its interaction, which is calculated via Minitab.Thereby, the standardized effect of a parameter indicates the deviation between the corresponding process parameter to the command variable.The calculation of the standardized effect goes via t-statistics.The t-value is defined as the ratio from a difference of the average value and the command variable to standard deviation.Hence, t-statistics specify the deviation from the command variable in relation to scattered sampling data (Stocker & Steinke, 2017).Consequently, different parameters and its interaction can be compared to each.A higher standardized effect results in a significance of the parameter.Based on the number of chosen parameters, a critical value for the significance can be selected from the t-statistics.By exceeding the critical value, the corresponding parameter or interaction can be assumed as significant to reach the command variable (Stocker & Steinke, 2017).At the beginning, the laser power was varied from 50 W to 400 W by applying a scanning speed of 400 mm/s, 600 mm/s, and 800 mm/s, respectively, at constant hatching distance of 100 µm.Afterwards, further hatching distances of 50 µm and 70 µm were used.For all iteration steps, cubic samples with a length of 8 mm were built using a layer thickness d s of 20 µm and a laser beam diameter of ∼50 µm.Furthermore, the scanning vectors are rotated after each layer by about 67°.The pre-heating of the building-platform (made of 316 l stainless steel) was 80 °C.
The relative density was measured according to VDI 3405-2 at polished cross sections of the exposure area and was calculated via quantitative images analysis of five images.The microstructure analysis was performed via analytical SEM in combination with EDX and EBSD.During EBSD, the microstructure was detected along the building direction.The analysis was performed with an acceleration voltage of 20 kV, a working distance of 15 mm, and a step size of 0.43 µm at a pre-tilted sample of 70°.Due to the high cooling rate during LPBF, a quenching of the remelted volume occurs so that a direct solution annealing takes place.Consequently, the as-built condition of the samples contained a supersaturated solid solution.Hence, the properties of the samples changed during subsequent heat treatment.Therefore, the microhardness HV0.1 measured by the Fischerscope HM2000 (Fischer GmbH) and the electrical conductivity via eddy-current principle (Sigmatest 2.069/FOERSTER Holding GmbH) were tested for the as-built condition and after aging at 480 °C for 2 h.

Powder production by gas atomization
Based on the particle size distribution (see Figure 2a), statistical values provide information about the quality of the separation process.Here, the upper and lower limit of the particle diameter is expressed by the d 10 -and d 90 -value, meaning that 10% and 90%, respectively, of the particles have a smaller diameter than the given value.Here, the d 10 has a value of 18 µm and the d 90 of 58 µm.Moreover, the measured median (d 50 ) has a value of 34 µm.
The chemical analysis yielded a Cr content of 1.07 wt.%, a Zr content of 0.25 wt.%, and an oxygen concentration of 22 ppm, respectively.Due to the low solubility of chromium in copper (Dies, 1967), the Cr precipitates are formed within the Cu matrix (see Figure 2b).An EDX point analysis verifies the presence of Cr phases.Furthermore, the nanostructure of the powder particles is decorated with nanoparticles (see Figure 2c) that are indicated to be zirconium oxide (ZrO 2 ).

Sample manufacturing by LPBF and process parameter definition
Figure 3 shows the distribution of the relative density depending on the selected parameters.Using a red cw-laser, the distribution of the relative density versus the laser power shows a continuous increase (Becker, 2014;Guan et al., 2019;Jahns et al., 2020;Uhlmann et al., 2016;Wallis & Buchmayr, 2019) (see Figure 3a).The relative density with a green qcw-laser shows a different behavior when using the same parameter set.For a laser power of P L = 50 W, a relative density of 91.5% is measured because particles are only partly melted (see Figures 3a and 4a).With increasing laser power, the relative density increases up to 99.6% at 125 W, see Figures 3a and 4b.By a further increase of the laser power, the relative density tends to decrease again to 95% at 200 W.At 200 W, the pores have an averages size of (52 ± 9.8) µm.Afterwards, the relative density rises up to 98% at 400 W and reaches an averages pore size of (95 ± 7.2) µm (see Figures 3a and 4b,c).
By increasing the scanning speed at a constant hatching distance of 100 µm, the distribution of the relative density shows a similar effect as increasing the laser power (see Figure 3b).All scanning speeds result in a maximum density at 125 W and decreases afterwards until 200 W.The hatching distance seems to be of a higher impact on the relative density (see Figure 3c).At hatching distances smaller than 100 µm, a decrease of the density after reaching a laser power of 100 to 150 W was observed, when using a scanning speed of 800 mm/s.Moreover, the approximate welding line has a width of 90-100 µm and a depth of maximal 52 µm (see Figure 5).
Based on the DOE, the influence of the parameters is represented by the standardized effect (see Figure 6).By passing the critical value of the standardized effect (here: 2.009), the respective parameter can be assumed as significant.According to standardized effect, the hatching distance is the most significant parameter with a value of 3.1483.Furthermore, a combination of laser power and hatching distance influences the relative density similarly.However, its standardized effect (1.1884) is lower than the critical value.Consequently, changing these parameters is not sensitive enough to regulate the relative density.According to the standardized effect, the impact of the interaction of the laser power and scanning speed (0.5627), the laser power (0.5120) and scanning speed (0.4067) is in a similar range.The lowest significance has the interaction of the scanning speed with the hatching distance.
Contrary to the powder, the LPBF sample did not reveal visible Cr precipitates via SEM within the microstructure (see Figure 7a).However, the analysis of the chemical composition of CuCr1Zr revealed a Cr-concentration of 1.06 wt.% which is in the same range as the powder.Consequently, Cr remains in solid solution within the matrix.Also, the Zr content of 0.24 wt.% is comparable to the powder.The oxygen concentration rises slightly to 34 ppm.malting stage.However, the formation of ZrO 2 nanoparticles is not fully analyzed and needs further investigation, e.g., through microstructure simulation of gas atomization and LPBF.
The microstructure formation is governed by preferential grain growth alomg the <100> direction during solidification in the direction of the maximum temperature gradient.Due to the inclination of the solid-liquid interface with respect to the plane perpendicular to the building direction (Figure 8d), the respective along the <100> texture is tilted as well (Haase, Tang, Wilms, Weisheit, & Hallstedt, 2017), in the present case by 23°.This was proven by EBSD measurements across the area shown in Figure 8a.The respective orientation distribution is given in Figure 8b and the pole figure in Figure 8c.

Properties
For electronic components, the electrical conductivity as well as the hardness of the material is of high importance.The properties of CuCr1Zr were investigated in as-built state (solution annealed condition), as well as after ageing.Due to the high amount of dissolved Cr, the electrical conductivity of both conditions is lower than that of the conventional material.The measured electrical conductivity of the sample in as-built condition is 11.1 MS/m, which is lower than that of the conventional CuCr1Zr, reported to be 20 MS/m (Deutsches Kupfer Institut Berufsverband e.V., 2005) (see Figure 9).Based on the formation of Cr, Cu 5 Zr, and Cr 2 Zr (Becker, 2014;Deutsches Kupfer Institut Berufsverband e.V., 2005;Tenwick & Davies, 1988), the conductivity rises to 39.5 MS/m aging for 2 h at 480 °C.However, the corresponding value of aged CuCr1Zr should be higher than 43 MS/m, according to Deutsches Kupfer Institut Berufsverband e.V. ( 2005).With 112HV0.1, the microhardness in the solution-annealed condition is above the reference value of 95HV (Deutsches Kupfer Institut Berufsverband e.V., 2005).This can be attributed to the higher solution strengthening potential since the Cr concentration dissolved in the supersaturated solid after LPBF is much higher.Moreover, the microhardness at aged condition overcomes the maximum value of 185HV (Deutsches Kupfer Institut Berufsverband e.V., 2005) with 220HV0.1.

Comparison of red cw-laser und green qcw-laser
By applying the optimum parameter for relative densities over 99.5%, a different temperature distribution is formed inside the melt pool.For heat conduction welding, the temperature increase ΔT with respect to room temperature can be estimated as a function of penetration depth by using the stationary heat conduction equation (Cline & Anthony, 1977;Graf et al., 2015).The calculation of ΔT can be impressed via where A is the degree of absorption, ρ the density, c p the specific heat capacity, κ the thermal conductivity, r the radius of the laser beam, τ = 2r/v the interaction time, and z the penetration depth.The result of Eq. ( 1) is the temperature distribution at the centre of the laser beam as a function of the penetration depth in z-direction.However, only the cw-mode can be expressed with Eq. (1).By using a cw-laser, a continuous exposure occurs during the interaction time.Consequently, the use of a cw-laser offers a continuous heat source, respectively.However, regarding the interaction time of the qcw-laser, the pulse duration and repetitions frequency must be considered.Thus, the exposure time is only a fraction of a cw-laser.Hence, the CuCr1Zr alloy is heated up stepwise during the interaction.Concerning the temperature dependency, the heat conductivity and absorption of the CuCr1Zr change after each exposure step.Hence, there is not a constant temperature increase.With rising temperature, the heating rate increases continuously due to a higher absorptivity and a decreasing heat conductivity at higher temperature.The temperature increase for relative densities of higher than 99.5% from Jahns et al. (2020) and the own work is shown in Figure 10.The calculation in Figure 10 results by applying Eq. ( 1) to the corresponding parameter set.The necessary physical properties were taken from Becker (2014) and Deutsches Kupfer Institut Berufsverband e.V. (2005).As a result, a maximal temperature increase of 2,503 °C for the red laser and 2,512 °C for the green laser is calculated at the centre of the laser beam (see Figure 10a).The corresponding melting depth is then ∼30.5 µm and 22.2 µm, respectively.However, the measured melting depth of the green laser indicates a higher melting depth, which can be attributed to the impact of the qcw-mode that cannot be represented by Eq. ( 1).Based on the shown temperature distribution, it can be assumed that using a green laser results in a more concentrated heat input generating a smaller melted volume as compared to the red.
Depending on the temperature distribution, different building rates  V and volume energies E V are adjusted for the red cw-laser and the green qcw-laser.Hereby, the building rate includes information about the remelted powder volume per time unit and can be estimated with the following equation (Becker, 2014) The building rates for appropriate process parameter for LPBF of CuCr1Zr are summarized in Table 1.Here, the values for  V are in the same range 0.8-1.2mm³/s by using a red cw-or a green qcw-laser.An increase of  V depends on the remelted melting depth.For defect-free samples, at least the outermost layer must be partially remelted.Hence, for a layer thickness of 20 µm, the melting depth should be in the range of 30-40 µm.For the calculated temperature, distribution of the red laser confirmed this behavior for defect-free samples (see Figure 10a).Due to a higher measured melting depth for the green qcw-laser, the layer thickness can probably increase slightly without a significant decrease of the relative density.In contrast to the building rate, the volume energy offers a greater difference between the two laser sources.For this purpose, the conventionally used equation has to be converted to a laser and material-dependent term.Therefore, the degree of absorption A of the powder bed is introduced to the equation.Consequently, the achieved value is called effective energy volume E V,eff and can be calculated via Depending on the parameter combination, the effective volume energy offers a reduction of the energy input from 40% to 60% by using a green qcw-laser.An exception is the process parameter of Uhlmann et al. (2016), where a pre-heating of 200 °C instead of 80 °C has been used.The advantages of a lower energy input are the reduction of residual stresses, reduced grain coarsening in the microstructure, and a weaker crystallographic texture (Caiazzo, Alfieri, & Casalino, 2020;Gokcekaya et al., 2021;Mishurova, Artzt, Haubrich, Requena, & Bruno, 2019;Yusuf & Gao, 2017).

Dependency of relative density on the laser power
The results of the DOE reveal an optimal laser power of 125 W for the green qcw-laser, where a maximal temperature difference of 2,512 °C is calculated.By increasing the laser power to 150 W, the evaporating temperature of copper is exceeded, which promotes a change of the welding mechanism from heat conductive welding to deep penetration welding (see Figure 10b).However, the keyhole exists only at a stable temperature beyond the evaporating temperature (Heider, Weber, Herrmann, Herzog, & Graf, 2015).Due to the periodic exposure, a repeated collapsing keyhole generates small pores that can be formed and cause a decrease in the relative density between 150 W and 200 W.At a laser power ≥200 W, the keyhole volume becomes larger and the pore sizes increases.It is assumed that with rising laser power, the temperature drop of the melt pool increases continuously with rising collapsing keyhole volume.Hence, more time is required to exceed the evaporating temperature again.Based on the continuous exposure, using a red cw-laser facilitates a steady increase of the temperature.As a result, the relative density rises equally by increasing the laser power.

Dependency of relative density on scanning speed
With increasing scanning speed, the interaction time decreases and less energy can be transferred to heat for both laser sources.Hence, the resulting maximum temperature inside the melting pool decreases with increasing scanning speed.However, Figure 10c indicates only a slight decrease of the temperature by doubling the scanning speed.As a result, a similar behavior of the relative density is observed by increasing the scanning speed.

Dependency of the relative density on the hatching distance
The measured relative densities at different hatching distance show the impact of the position of exposure.Whether the melting pool and the hatching distance are of the same order, only CuCr1Zr-powder particles are located along the scanning vectors.Hence, the interaction between the laser with the powder occurs at a surface with a homogenous roughness.With decreasing hatching distance, an increasing overlap of the weld pool was observed.Consequently, the interaction of the laser is no longer with a homogenous roughness.The weld pool of the previous scanning line contains depressions due to the collapsed keyholes as well as the characteristic surface roughness of a welding line.A rougher surface increases the absorptivity due to multiple reflection of the laser (Bergström, Powell, & Kaplan, 2007), promoting a more efficient transformation of input energy into heat required to melt the powder particles.Thus, the evaporating temperature of copper is achieved earlier and keyholes are formed more frequently leading to an intensification of collapse events.

Conclusions
During LPBF of CuCr1Zr with a green qcw-Laser, a maximum relative density was measured at P L = 125 W, v s = 400 mm/s, and h s = 100 µm.Moreover, the LPBF sample contains only ZrO 2 nanoparticles whereas the powder contained coarse chromium.Consequently, the chromium was completely dissolved within the matrix during LPBF.Based on the results of the DOE and the temperature distribution, the process evolution during the exposure time was reconstructed.Thereby, the evaporation temperature of copper can be passed with increasing laser power leading to keyhole formation.However, repeated and regularly collapsing keyholes can caused enhanced porosity.These effects are increased at increasing laser power.
Furthermore, with decreasing hatching distance, the exposure of a former scanning line causes multiple reflections due to a higher surface roughness.Consequently, the evaporation temperature of Cu can be achieved in a wide range of laser power and the corresponding keyholes create pores again.

Figure 1 .
Figure 1.schematic representation of a DoE showing the variation of laser power, scanning speed, and hatching distance in between the operating window as well as the average of the parameter and constellations outside the operating window (star points).

Figure 2 .
Figure 2. (a) particle size distribution of the CuCr1Zr powder, (b) representative morphology of a particle, (c) microstructure of CuCr1Zr, which consists of a Cu-matrix with Cr-rich precipitates as well as of Zro 2 nanoparticles.

Figure 3 .
Figure 3. results of the DoE illustrated by (a) comparison of the relative density of a red (λ = 1060 nm) cw-and a green (λ = 532 nm) qcw-laser (results from the red laser is given by (Jahns et al., 2020) at v s = 400 mm/s and at h s = 100 µm), (b) resulting relative density by variation of the scanning parameters (at h s = 100 µm), (c) measured relative density at a decreasing hatching distance (v s = 800 mm/s).

Figure 4 .
Figure 4. observed microstructure using a laser power of (a) 50 W, (b) 125 W, (c) 200 W, and (d) 400 W at a scanning speed of 400 mm/s and a hatching distance of 100 µm (light optical microscopy).

Figure 5 .
Figure 5. overview of (a) exposure area by optical microscopy with a melting line width between 90 µm and 100 µm and (b) welding depth of ca.52 µm (P l = 100 W; v s = 124 mm/s; h s = 120 µm).

Figure 6 .
Figure 6.standardized effect showing the significance of the process parameters.

Figure 7 .
Figure 7. illustration of (a) microstructure, where no Cr-phases are observed and (b) the Zro 2 nanoparticles and (c) EDX-line scan of nanoparticle showing a rising Zr-and o-concentration (P l = 125 W; v s = 400 mm/s and h s = 100 µm).

Figure 8 .
Figure 8. Microstructure analysis via EBsD of an additively manufactured CuCr1Zr sample showing (a) the measured microstructure, (b) the corresponding ipf-map in building direction, and (c) the pole figure illustrating an angle deviation of the <100> texture of 23°.

Figure 9 .
Figure 9. Hardness and electrical conductivity of additively manufactured CuCr1Zr at as-built and annealed (480 °C/2 h) condition (P l = 125 W, v s = 400, mm/s and h s = 100 µm) additional hardness and electrical conductivity values indicate references value taken from Deutsches Kupfer institut Berufsverband e.V. (2005).

Figure 10 .
Figure 10.Calculated temperature difference for (a) red (P l = 370 W; v s = 400 mm/s) and green laser (P l = 125 W; v s = 400 mm/s), (b) red (P l = 370 W) and green laser (P l = 125 W) by increasing the scanning speed from 400 mm/s to 800 mm/s, and (c) red (v s = 400 mm/s) and green (v s = 400 mm/s) laser by increasing the laser power [physical properties taken from Becker (2014); Deutsches Kupfer institut Berufsverband e.V. (2005)/r green = 25 µm and r red = 50 µm].

Table 1 .
overview of the building rate and effective volume energy for processes parameters that result in a relative density higher than 99.5% lpBf of CuCr1Zr by using a red cw-and green qcw-laser.