1 Introduction

The L-PBF process enables fast production of highly complex parts without tools by melting metallic powder layer by layer. Although this process has been around for more than 30 years, there is only a limited choice of alloys that can be processed successfully. High-strength aluminum alloys are prone to hot cracking because of their solidification characteristics and the high cooling rates during L-PBF. This is why modified conventional alloys are the subject of current research with the aim to produce crack-free and dense samples by addition of Si or grain refiners and replacing volatile elements like Mg [3, 15, 20, 27].

However, specifically designed alloys for L-PBF are still needed. The first alloy of this tailoring approach was Scalmalloy® (Al–Mg–Sc–Zr), developed by Airbus between 2000 and 2010 and currently commercially available from the Airbus subsidiary APWORKS. Scalmalloy® has a tensile strength of 520 MPa and fracture elongation of 13% in heat-treated condition [2]. In 2019, the Scancromal® (Al–Cr–Sc–Zr) alloy concept was presented by Airbus where Mg is replaced by Cr. First trials using a AlCr2.6Sc0.72Zr chemistry revealed a tensile strength of up to 395 MPa and a very good fracture elongation of 21% [34]. Furthermore, the L-PBF meltpool dynamics are more stable due to the alloy’s higher surface tension, viscosity and less evaporation [5].

The melt process depends on various parameters including laser power P, scan speed \(v_{s}\), distance between scan tracks (hatch h) and build layer thickness l. These parameters are used to calculate an equivalent theoretical volume energy density (VED, Eq. (1)) or area energy density (AED, Eq. (2)) and can be used as guideline for parameter selection as it represents melt pool width [7]. That being said, VED is not able to capture realistically the threshold energy input value differentiating between keyhole and conduction melting mode [7]:

$$\hbox{VED}= \frac{P}{v_{s}\,h\,l},$$
(1)
$$\hbox{AED}= \frac{P}{v_{s}\,h}.$$
(2)

Besides VED, laser beam energy absorbance (converting laser power into melting heat) is a critical factor in L-PBF of aluminum and affects productivity. Absorbance depends on laser wavelength and is at 1000 nm about 50% for pure aluminum powder which is already higher than for solid aluminum (approximately 20%) [11]. Absorptivity is also influenced by alloying content, powder particle size, laser wavelength and spot size [26, 42]. When laser power (intensity) increases beyond a certain threshold value a capillary can form (keyhole) that will geometrically foster reflections of the laser and therefore increase principal absorptivity [44]. Although keyhole welding mode can lead to keyhole porosity it does also lead to a deeper penetration of the substrate and thus increases productivity by enabling thicker layer and/or faster scanning speeds during L-PBF.

The dynamics of melting and densification of the powder bed depend on partly temperature-dependent physical properties, such as surface tension, viscosity, heat capacity and thermal conductivity. A suitable method to comprehensively model the relevant physics of the melt pool is the Smoothed Particle Hydrodynamics (SPH) method [32]. SPH is a Lagrangian particle method that solves the Navier–Stokes momentum equation, the thermal energy balance equation, and the mass balance equation. Simulations of the application and subsequent melting of several powder layers complement the experimental investigations.

In this work, a Scancromal® variant is presented that aims for a more economic L-PBF alloy with moderate strengths and high build rates. Unlike in aviation, this is especially interesting for automotive applications where lower costs in combination with high productivity are more important than highest strengths. To investigate the processability of the new alloy, a parameter study was carried out with a build layer thickness of \(l={100}\,\upmu \hbox {m}\) including porosity analysis and hardness measurements. To evaluate if even higher build rates are possible, a 2D multilayer simulation of the process was done for \(l={150}\,\upmu \hbox {m}\).

2 Materials and methods

2.1 Powder and L-PBF process

The material used in this work was produced via centrifugal atomization in nitrogen by Toyo aluminum K.K., Japan. Chemical composition of the powder is given in Table 1. Powder particle distribution was measured with an optical image analyser Camsizer X2 by Microtrac Retsch GmbH on three powder samples resulting in a mean particle sphericity of 0.91 and particle size \(\hbox {d}_{10}\), \(\hbox {d}_{50}\), and \(\hbox {d}_{90}\) of \({36.0}\,\upmu \hbox {m}\), \({63.6}\,\upmu \hbox {m}\), \({104.2}\,\upmu \hbox {m}\), respectively.

Table 1 Chemical composition in wt.% of Scancromal® powder provided by manufacturer
Full size table

Samples were produced at Modell- und Formenbau Blasius GERG GmbH on a Concept Laser M2 upgraded to Series 5 system equipped with two 400 W fiber lasers with a wavelength of 1065 nm and a variable spot size of 50 to \({500}\,\upmu \hbox {m}\). The used parameters on this system are summarized in Table 2. All build processes were executed in nitrogen atmosphere with a maximum oxygen level of 0.1%. Nitrogen was chosen because it is cheaper than argon and no significant difference in density, nitrogen content or mechanical properties have been reported for Al–Si(-Mg) alloys processed in nitrogen atmosphere [46, 48].

Table 2 Summary of process parameters used during this study
Full size table

2.2 Parameter study

To study the influence of hatch, scan speed and laser power, 35 parameter sets were selected for the parameter study. For each set, three cubes measuring \(10\times 10\times 10\,\hbox {mm}^3\) were printed using three build jobs. One parameter set was repeated during the three build jobs to evaluate deviations between them. This results in 111 cubes. Note that one cube of the repeated parameters got lost during cutting off with a bandsaw.

2.2.1 Archimedes measurements

Archimedes density measurements were performed on all remaining 110 cubes in accordance to DIN EN ISO 3369 [18] in aceton due to its lower surface tension which is advantageous for rough surfaces [38]. Density of each sample \(\rho _{\text {sample}}\) was calculated following Eq. (3):

$$\rho _{\text {sample}} = \frac{m_{\text {air}} \, \rho _{\text {liquid}}}{m_{\text {air}} - m_{\text {liquid}}}.$$
(3)

\(\rho _{\text {liquid}}\) is the density of aceton, \(m_{\text {air}}\) is the determined specimen mass in air and \(m_{\text {liquid}}\) is the determined specimen mass in aceton. Relative porosity p (Eq. (4)) was calculated by \(1-D\), where D represents relative density as fraction of \(\rho _{\text {sample}}\) and the theoretical absolute density of the alloy \(\rho _{\text {theoretical}}\).

$$p = 1 - D = 1 - \frac{\rho _{\text {sample}}}{\rho _{\text {theoretical}}}$$
(4)

\(\rho _{\text {theoretical}}\) was calculated according to Aluminum Standards and Data 2017 [1] as \({2.793}\,\hbox {g cm}^{-3}\). The Mo content could not be taken into account in this calculation.

To compare productivity, build rates (BR, Eq. (5) [41]) were also evaluated in comparison to the achievable porosity:

$$\hbox{BR} = v_{s}\,h\,l.$$
(5)

2.2.2 Pore analysis

To get a deeper insight on pore morphology, optical micrographs of two cubes per parameter set were prepared using \(5\times\) magnification. For this micrographs, cross-sections parallel to the build direction were chosen (see blue highlights in Fig. 1) and evaluated with a Python script calculating number, size and sphericity of pores and relative density.

Fig. 1
figure 1

Cross-sections parallel and perpendicular to build direction (BD) for image analysis and hardness measurements

Full size image

2.2.3 Mechanical properties

Vickers hardness was measured for selected parameter sets according to DIN EN ISO 6507-1 [17] with an Instron Wolpert Testor 930/250. For this measurement, the third remaining cube was cut into the red highlighted pieces shown in Fig. 1 to measure hardness at nine points along cross-sections perpendicular to the build direction. To eliminate the effect of outliers, the highest and lowest values of each parameter and condition were excluded in the results given in this work. As-built condition was measured about 2 to 3 mm beneath the top layer. After aging in air for 4 h at \({330}\,^{\circ }\hbox {C}\) heat-treated specimens were measured about 4–5 mm beneath the top layer. Temperature and duration of the heat treatment were chosen based on peak aged condition for Scalmalloy® [39].

2.3 Microstructure

To show the microstructure, an optical micrograph was taken on a Leica DM1750 microscope under use of polarized light. The last preparation step of the micrographs was silica polishing (OP-S NonDry). No etching was applied. Besides micrographs, SEM and EDS analysis were conducted with 20 kV voltage and 12 mm working distance on a LEO 1430 SEM equipped with a Bruker XFlash 6-30 silicon drift detector.

2.4 Multilayer simulation

The L-PBF process simulations mimic the application and melting of up to six powder layers. For each layer, a powder bed is first generated with twice the height of the layer thickness to be built up. The simulations are performed in two dimensions (2D) as a suitable compromise between computational effort and predictive quality. In order to generate a realistic image of the volume fraction of the powder layer, the 2D approach requires that some of the particles have no contacts with neighboring particles. This corresponds to a 2D section through a three-dimensional (3D) powder layer at rest. The powder particles are discretized within the SPH method with a spatial resolution of \({2.5}\,\upmu \hbox {m}\). The width of the simulated build chamber section is 1.5 mm.

In the ray tracing method used, the laser is split into a large number of individual rays. The intensity follows a Gaussian distribution, with the spot diameter corresponding to four times the standard deviation. If a ray hits the material surface at a certain angle, absorption and reflection are calculated using Fresnel’s equations with the wavelength-dependent refractive index of the material [10]. The absorbed portion acts as a local heat source at the material surface due to the shallow optical penetration depth. The reflected portion of the ray is traced further until its remaining power is negligible. The use of the ray tracing method is particularly useful for the correct modeling of the locally introduced power during keyhole melting [43]. The modeling of the laser tracks in 2D is performed according to the scheme described in [25], taking into account power, scan speed and hatch.

Above the melting temperature, the modeled alloy liquefies. The flow in the melt pool is influenced by thermocapillary forces, gravity and, the recoil pressure if the gas transition is reached. The latent heats of the phase transitions are considered in the model. The cooling process is controlled by heat conduction to the build platform and the lateral walls as well as by convective cooling, thermal radiation and evaporation at the free surface to the build chamber.

The simulations in this work were performed using the SimPARTIX® software developed at Fraunhofer IWM [8]. The data on temperature-dependent material properties required for the SPH simulations were obtained by thermophysical measurements or by thermodynamic CALPHAD calculations using Thermo-Calc with the TCAL6 database.

3 Results

3.1 Parameter study

3.1.1 Archimedes measurements

35 parameter sets were printed in three build jobs and relative porosity was determined with Archimedes measurements. Mean porosity of all parameters ranged from 6.87 to 0.55%. An average scatter of ± 0.4% was experienced during all measurements. Three parameter sets were repeated in each build job. Relative porosity of these repeated ones varied between 0.70% and 0.55% showing a good reproducibility. The lowest mean porosity of 0.55% was achieved for \(h={100}\,\upmu \hbox {m}\) and \(v_s={1300}\,\hbox {mm s}^{-1}\) with an equivalent build rate of \({46.8}\,\hbox {cm}^{3}\,\hbox {h}^{-1}\).

To show the connection between porosity, build rate and VED, a polynomial fit was done for all parameters in Figs. 2 and 3. For both, build rate and VED, an optimum range is observed to achieve porosities below 1%. These optima range between 37.7 and \({59.9}\,\hbox {cm}^{3}\,\hbox {h}^{-1}\) for build rate and 22.7 and \({36.7}\,\hbox {J\,mm}^{-3}\) for VED.

Fig. 2
figure 2

Relative porosity (Archimedes) depending on build rate; average scatter range is \(\pm \,0.4\%\) experienced from all performed Archimedes measurements; \(R^2\) value of polynomial fit is 0.9347

Full size image
Fig. 3
figure 3

Relative porosity (Archimedes) depending on VED; average scatter range is \(\pm \,0.4\%\) experienced from all performed Archimedes measurements; \(R^2\) value of polynomial fit is 0.9461

Full size image

There are three outliers Figs. 2 and 3 that were excluded from the polynomial fit and marked with a black cross. These outliers show a significant higher porosity, although they are in the optimum ranges and will be discussed in Sect. 4.1.1.

3.1.2 Pore analysis

Micrographs of 380 W samples were prepared and analyzed in the range of h = 80 to \({150}\,\upmu \hbox {m}\) and \(v_s={1100}\) to \({1700}\,\hbox {mm s}^{-1}\). Relative porosity calculated by image analysis ranged from 3.22 to 0.10% with an average deviation of ± 0.12%. No cracks at all were observed in the evaluated micrographs.

The influences of scan speed and hatch on porosity are plotted in Fig. 4. The included micrographs in Fig. 4 show pore development and therefore porosity mechanism: looking at the micrographs of the red rectangle, a lower scan speed leads to more spherical shaped pores and a lower porosity. An optimum is observed at \(v_s={1300}\,\hbox {mm\,s}^{-1}\). Below this optimum, gas porosity dominates and porosity increases again. Looking at the micrographs of the black rectangle, higher hatch leads to more irregularly shaped pores and lack of fusion. Hatch has its optimum at \(h={100}\,\upmu \hbox {m}\), but shows similar pores for 80 and 120 \(\,\upmu \hbox {m}\).

Fig. 4
figure 4

Contour plot of mean relative porosity measured by Archimedes depending on scan speed and hatch for \(P={380}\) W; optical micrographs show pore development with sections of the biggest pores for the marked parameter sets

Full size image

Figure 5 shows a comparison of relative porosity by Archimedes, image analysis, pore properties and hardness results. Porosity by image analysis is slightly lower than by Archimedes (Fig. 5a,d), but the change in porosity with hatch and scan speed is similar and confirms the optimum parameter set (\(h={100}\,\upmu \hbox {m}\), \(v_s={1300}\,\hbox {mm\,s}^{-1}\)). The optimum is also confirmed by the lowest number of pore observations around the ideal parameters in Fig. 5e. In addition to that, mean pore sphericity correlates well with VED (Fig. 5b): high energy input by lower hatch and speed leads to more spherical pores as shown in Fig. 4.

Fig. 5
figure 5

Comparison of relative densities by a Archimedes and d image analysis as well as pore properties (b mean pore sphericity, e number of pore observations) and c hardness map in as-built condition and f comparison of hardness before and after heat treatment (\({330}\,^{\circ }\hbox {C}/{4}\,\hbox {h}\)); the black rectangle in c indicates the three parameter sets shown in f

Full size image

3.1.3 Mechanical properties

Hardness was measured for selected parameter sets with \(P={380}\) W and is plotted for as-built condition in Fig. 5c. The maximum hardness in as-built condition was reached by \(v_{s}={1300}\,\hbox {mm\,s}^{-1}\) and \(h={80}\,\upmu \hbox {m}\) and is 93.1 HV10. Minimum hardness in as-built condition belongs to the cube with the highest relative porosity within the measurements and is 72.8 HV10 for \(v_{s}={1700}\,\hbox {mm\,s}^{-1}\) and \(h={150}\,\upmu \hbox {m}\). Mean deviation of all hardness measurements was ± 2.4 HV10. In general, a low relative porosity leads to a high hardness and all samples with porosity below 1% reach > 90 HV10.

Fig. 6
figure 6

As-built sample for \(P={380}\) W, \(v_{s}={1300}\,\hbox {mm\,s}^{-1}\) and \(h={100}\,\upmu \hbox {m}\): a micrograph (\(10\times\) magnification, polarized light) of characteristic melt pools and b SEM SE picture of Al–Cr equiaxed dendrites within melt pool boundaries

Full size image

Hardness before and after aging for the three optimum parameter sets are displayed in Fig. 5c,f. After aging, hardness increased to over 125 HV10. While as-built hardness was slightly lower for \(h={100}\,\upmu \hbox {m}\) it is slightly higher in heat treated condition. The deviation between the three displayed samples lies within the observed scatter range and therefore no significant change in dependence of hatch can be detected. Overall, we see a significant hardness increase of about 37% after aging that will be discussed in Sect. 4.1.3.

3.2 Microstructure

A typical bimodal microstructure of L-PBF processed aluminum in as built condition is presented in micrograph in Fig. 6a. The micrograph reveals scan tracks that are clearly visible and consist of a melt pool and melt pool boundary. An accumulation of particles within the melt pool boundaries was detected (see rectangle Fig. 6a). These particles were revealed as equiaxed dendrites in Fig. 6b at higher magnifications. EDS analyses showed a doubled Cr content for the dendrites compared to the matrix around and the melt pools. The dendrites’ origin will be discussed in Sect. 4.2.

3.3 Multilayer simulation

To validate the simulation results, porosity of the simulated 2D-sections (three areas per parameter set) and the measured porosity via Archimedes of eight parameter sets are compared in Fig. 7. It is apparent that the simulation results are in good agreement with Archimedes measurements for a porosity beneath 1.5%. Only for higher porosities, a significant difference is observed. As these high porosities are not of interest, further simulations with \(l={150}\,\upmu \hbox {m}\) were started under the same conditions.

Fig. 7
figure 7

Validation of simulation results by comparing relative porosity of the simulation (porosity of three different areas of 2D simulation) and relative porosity of Archimedes measurements (\(l={100}\,\upmu \hbox {m}\)); error bars indicate standard deviation of simulated 2D-sections and minimum/maximum of Archimedes measurements

Full size image

To find a parameter set for the increased layer thickness by the simulations, laser power was increased as first step. 640 W was considered too high because of a very dynamic meltpool whereas 520 W showed lack of fusion on the edges which is why 560 W was chosen as laser power. Next, the influence of laser spot size was evaluated. Comparing the spot size of \({85}\,\upmu \hbox {m}\) and \({120}\,\upmu \hbox {m}\), less lack of fusion was observed with the latter. As last step, hatch and scan speed were varied with the chosen laser power and spot size. The final simulations are presented in Fig. 8 and show the simulated 2D-sections with declaration of porosity. The combination of \(v_s={1300}\,\hbox {mm\,s}^{-1}\) and \(h={100}\,\upmu \hbox {m}\) is identified as optimum parameter set regarding low porosity and high productivity and would lead to a build rate of \({70.2}\,\hbox {cm}^{3}\,\hbox {h}^{-1}\).

Fig. 8
figure 8

2D simulation results for varying hatch and scan speed with indication of resulting porosity (\(l={150}\,\upmu \hbox {m}\)); ± indicates standard deviation based on three different areas of 2D simulation

Full size image

4 Discussion

4.1 Parameter study

4.1.1 Porosity and pore morphology

Among the evaluated parameter sets in this study, a reliable process window for \(v_{s}={1300}\,\hbox {mm\,s}^{-1}\) and \(h={80}\)–\({120}\,\upmu \hbox {m}\) using \(P={380}\) W was determined. This process window had a maximum porosity of 1.13% (mean 0.76%) during this study. The process window is expected to be stable as an increase of porosity to >2% requires a significant change in hatch or scan speed (Fig. 4).

Mo could not be taken into account in the calculation of the theoretical absolute density by [1]. Because of this, a shift towards higher porosities in Figs. 2, 3 might be possible. However, as the image analysis of the micrographs shows a lower porosity than Archimedes (Fig. 5a, d), this influence is expected to be negligible.

There were three outliers for the lowest applied scan speed that had a significant higher porosity within the optimum VED range (Figs. 2, 3). To understand the reason for these outliers, additional micrographs were prepared. A comparison of the optimum parameter in Fig. 9b with the outliers Fig. 9a, c shows a high amount of big spherical gas pores. The melt pool shape cannot be identified reliably in the micrographs and therefore keyhole porosity cannot be confirmed by this. Pores caused by the keyhole effect usually are not perfectly spherical but more oval-shaped because of the formation in a dynamic melt pool. However, the pores shown in Fig. 9 are very spherical and hydrogen gas pores seem more realistic. Unfortunately, Al-powders are usually considerably contaminated with hydrogen. The origin of this contamination can be moisture on the surface of the powder particles (deposited during production, storage or handling) or trapped hydrogen in the particles themselves due to the atomization process. Drying would reduce this contamination but is usually not applied. Spherical pores by recombined hydrogen are a known issue in welding as well as L-PBF of aluminum because the solubility of hydrogen in liquid Al decreases rapidly during cooling [28, 47]. The three outliers in this study were produced with the lowest scan speed applied. The low scan speed leads to a high energy input and therefore a lower heat dissipation which gives the hydrogen enough time to recombine and form spherical pores as shown in Fig. 9a, c. When higher scan speeds are applied, no excessive formation of spherical hydrogen pores is observed (Fig. 9b). Nevertheless, the cause of the pores cannot be clarified with absolute certainty without measuring the gas content of the pores.

Fig. 9
figure 9

Comparison of micrographs with sections of the biggest pores, VED and porosity (Archimedes) of two outliers showing keyhole porosity (a,c) and dense sample (b) within the optimum VED range

Full size image

4.1.2 VED and build rates

Possible build rates and applicable VED values were determined during this parameter study for the novel economic version of Scancromal®. Figure 10 gives an overview of the achieved build rates of Al–Cr alloys (orange area) in comparison to Al–Mg (blue area) and Al–Si alloys (gray area). The shown Al–Cr values were determined via this parameter study and simulation and an additional Scancromal® variant (see Table 3).

Fig. 10
figure 10

Possible build rates and their required VEDs for Al–Cr, Al–Mg and Al–Si alloys in comparison to processed and simulated Scancromal®; for comparison VED was calculated from the parameter sets with the lowest achievable porosity in each study

Full size image
Table 3 Parameter set for a Scancromal® variant with higher alloying content (Al–4.0Cr–1.6Mo–0.72Sc–0.44Zr)
Full size table

The first finding of the overview in Fig. 10 is that Al–Mg alloys need a much higher VED and reach lower build rates when low layer thicknesses are used. APWorks has developed a high productivity parameter set for Scalmalloy® (Al–Mg) with a build rate of \({45}\,\hbox {cm}^{3}\,\hbox {h}^{-1}\) when using 600 W laser power [33]. Because there is no VED available for this productivity parameter set, a horizontal line indicates its build rate in Fig. 10. Even though this high productivity build rate is similar to this study, a significant higher laser power is required for it.

The second finding of the overview in Fig. 10 is that only the simulation with \(l={150}\,\upmu \hbox {m}\) and the additional Scancromal® variant achieve a higher build rate than Al–Si alloys. The reason for the higher build rate of the simulation is basically the layer thickness: both layer thickness and build rate increased by 50%. It will be evaluated in further studies if the simulation predicted a reasonable process window for the higher layer thickness. More interesting is the reason for the top build rate in Fig. 10: looking at layer thickness and laser power of AlSi10Mg, it is evident that laser power correlates with laser thickness and build rate. In contrast to Al–Si alloys, the additional Scancromal® variant (orange crosses) requires almost the same laser power for \(l={50}\,\upmu \hbox {m}\) and \({100}\,\upmu \hbox {m}\). While almost all parameters were held constant between this two parameter sets, build plate heating might contribute to increase the layer thickness. Despite build plate heating, no significant increase in laser power is necessary to increase layer thickness.

Last but not least, the two main differences between the Scancromal® variant in this study and the additional one are Cr content (2.9 and 4.0%), powder particle size (\(\hbox {d}_{90}={104.2}\,\upmu \hbox {m}\) and \({64.8}\,\upmu \hbox {m}\)) and build chamber atmosphere (nitrogen and argon). While particle sizes below \({70}\,\upmu \hbox {m}\) and argon are standard values in L-PBF, Cr content might be the main factor for high build rates. However, further investigation of productivity optimization would be necessary to validate if Cr is the reason for the high build rates.

4.1.3 Mechanical properties

Hardness measurements in Sect. 3.1.3 show a reliable process window where almost all parameter sets that achieve > 99% relative density do also achieve a mean hardness of 90 to 93 HV10 (as built). For Scalmalloy® a change of hardness was shown for different hatch and therefore different amounts of fine-grained regions [19] and for different scan speeds [40]. This change does not apply to the alloy and process window used in this work.

After aging, hardness increased by 37% to 125 HV10 which is most likely attributed to precipitation hardening by Al(Sc,Zr)-precipitations. In this investigation, the Sc content is 0.3%, whereas in many disclosed Scalmalloy® investigations, the Sc content varies from 0.65 to 0.8%. Consequently, the measured hardness evolution is about 40–50 HV lower than visible in heat-treated Scalmalloy® [4, 29]. Compared to AlSi10Mg (stress relieved and aged), hardness after heat treatment is comparable or up to 44 HV higher depending on the chosen heat treatment [22, 31].

A previously reported Scancromal® variant Al–2.6Cr–0.7Sc–0.3Zr achieved in as-built condition 70–75 HB and in peak aged condition 145–150 HB [34, 36]. Transformed to Vickers hardness according to DIN EN ISO 18265 [16], the previous variant is only slightly lower (10 HV) in as-built condition but about 25% higher in peak aged condition. This comparison also shows that the higher Cr and Mo content in the alloy treated in this work is responsible for about 10 HV hardness increase in as-built condition. An explanation for the lower peak aged hardness could be the reduced Sc content as well as an insufficient aging temperature. It is known that Zr affects precipitation kinetics of \(\hbox {Al}_{3}\hbox {Sc}\) phases and Al–Zr alloys reach peak hardness around \({400}\,^{\circ }\hbox {C}\) [12, 24]. Rating and comparing the hardness increase due to \(\hbox {Al}_{3}\)(Sc,Zr) precipitation shows in both AlCr(Mo)ScZr material cases a good correlation and coherence about the principal strength improvements possible by Sc alloying when using a single-step annealing.

4.2 Microstructure

As described in Sect. 3.2, equiaxed dendrites can be observed within the melt pool boundaries with an increased Cr content. These dendrites may be primary \(\hbox {Al}_{5}\hbox {Cr}\) phases as indicated by Thermo-Calc simulation in Fig. 11. It is possible that these dendrites form within the solute because of very high undercooling [13]. Al–Cr precipitates close to melt pool boundaries were reported before for a Scancromal® variant without Mo but were smaller and more evenly distributed [36]. Mo and Cr content as well as processed layer thickness with accompanying energy input and melt pool temperatures might lead to the much larger dendrites in this study. Build plate heating could also lead to lower cooling rates and therefore bigger primary AlCr(Mo) phases.

Fig. 11
figure 11

Volume fraction of stable phases of Scancromal® as a function of temperature as calculated by Thermo-Calc using database TCAL8

Full size image

4.3 Multilayer simulation

The residual porosities predicted by the simulations agree well with the measured values of the \({100}\,\upmu \hbox {m}\) layers in the range of the optimal process parameter window described by the AED. This confirms the applicability of the simulations to identify optimal alloy-dependent laser parameters. Discrepancies between simulated and measured porosities for values of AED below the stable process window could be due to an overestimation of the temperature caused by the limited size of the simulated build chamber section. The simplification of flow dynamics and temperature development associated with the 2D simulation may also be a reason for the quantitative deviations in porosity.

Beyond the scalar values of the porosity, the simulation provides detailed insights into the melt pool dynamics and the occurrence of lack-of-fusion defects. In addition, the simulated structures can be evaluated with respect to surface roughness, which, however, was not the focus of the present study.

5 Conclusion

A comprehensive parameter study including simulation, heat treatment and hardness measurement was performed on the novel Scancromal® variant with a composition of Al–2.9Cr–1.3Mo–0.3Sc–0.3Zr. In contrast to other studies, productivity was considered as an industrially interesting factor. The following conclusions can be drawn:

  • A stable process window has been determined using \({85}\,\upmu \hbox {m}\) laser spot size for \(P={380}\) W, \(l={100}\,\upmu \hbox {m}\), \(v_{s}={1300}\,\hbox {mm\,s}^{-1}\) and \(h={80}\)–\({120}\,\upmu \hbox {m}\).

  • High build rates have been achieved: with a layer thickness of \(l={100}\,\upmu \hbox {m}\), a build rate of \({47}\,\hbox {cm}^{3}\,\hbox {h}^{-1}\) and relative porosity (Archimedes) of 0.55% ± 0.4% was presented.

  • Simulations of \(l={150}\,\upmu \hbox {m}\) showed the possibility of a significant further increase to \({70}\,\hbox {cm}^{3}\,\hbox {h}^{-1}\) build rate while relative porosity should remain 0.6% ± 0.3%.

  • Hardness for \(l={100}\,\upmu \hbox {m}\) after heat treatment (\({330}\,^{\circ }\hbox {C}/{4}\,\hbox {h}\)) is 125 HV10 and up to 44 HV higher than stress relieved and aged AlSi10Mg.

To evaluate the reason for these high build rates, further research regarding melt pool behavior of Cr-containing aluminum alloys and productivity of other aluminum alloys for L-PBF is needed. In addition to that, heat treatment was chosen based on Scalmalloy® which might be not the ideal temperature or duration for specifically this alloy. Further research on heat treatments, mechanical properties and phase analysis for this novel alloy is currently in progress.