Robust γ-TiAl Dual Microstructure Concept by Advanced Electron Beam Powder Bed Fusion Technology

Abstract

The dual microstructure concept for gamma titanium aluminides (γ-TiAl) processed via electron beam–powder bed fusion (PBF-EB) provides a huge potential for more efficient jet turbine engines. While the concept is feasible and the mechanical properties are promising, there are still some challenges. For an industrial application, the heat treatment window has to match the conditions in industrial furnaces. This study shows how the required heat treatment window can be achieved via advanced PBF-EB technology. Through using an electron beam with 150 kV acceleration voltage, the difference in aluminum between the designed aluminum-rich and aluminum-lean regions of the part is increased. Moreover, the aluminum content within each of these regions, respectively, is more homogenous compared to the 60 kV acceleration voltage. This combination provides a heat treatment window of 25 °C, enabling the industrial application of the dual microstructure concept for γ-TiAl.

1. Introduction

Additive manufacturing (AM) in general and electron beam–powder bed fusion (PF-EB) in particular show new opportunities compared to conventional manufacturing technologies. For instance, complex geometries can be manufactured near net shape and the processing parameters, and hence the microstructure is locally adjustable. During PBF-EB, the build platform is lowered by the layer thickness, and powder is applied. Subsequently, the powder bed is pre-heated by a strongly defocused electron beam to sinter the powder particles and control the process temperature. Then, the parts are molten by a focused electron beam. Finally, an electron image of the built layer can be taken for process monitoring if electron optics are available. A comprehensive review of the PBF-EB process is given in [1].

Due to the inertia-free deflection of the electron beam, PBF-EB possesses a remarkably large degree of freedom to locally adjust melting parameters. Firstly, adjusted melting parameters were used for local texture control of nickel-based superalloys in PBF-EB [2,3]. Similar approaches are used for microstructure control in laser-direct metal deposition (DMD) [4] and laser beam–powder bed fusion (PBF-LB) [5].

Another material class suited for locally tailored microstructure via PBF-EB is gamma titanium aluminide (γ-TiAl). The microstructure of γ-TiAl alloys can be adjusted by heat treatment. For the industrial application of the Ti-43.5Al-4Nb-1Mo (at%) TNM alloy, two microstructures are of interest: the fully lamellar (FL) microstructure and the nearly lamellar (NL) microstructure. If the heat treatment takes place in the α phase field, coarse α grains are formed. During cooldown, fine γ-lamellae grow into the α grain [6,7]. Lastly, the α transforms into α₂ [6,7]. Hence, a microstructure with coarse α₂/γ colonies consisting of fine lamellae is formed, which is called a fully lamellar microstructure [7,8]. If the heat treatment is conducted in the α + γ phase field, the remaining globular γ_glob grains prevent the α grains from growing. Therefore, only small α₂/γ colonies form with γ_glob in between. This microstructure is classified as nearly lamellar [7,8]. Since both microstructures have different characteristics, the mechanical properties are different as well. While the FL possesses an increased creep resistance [9], the NL displays enhanced ductility [10].

By locally adjusting the aluminum content of the produced parts via PBF-EB, different microstructures can be achieved at the same heat treatment temperature [11,12]. Al-lean regions are heat treated in the α phase field, while Al-rich regions stay in the α + γ phase field [11]. As a result, the mechanical properties of different sections of the part can be tailored to the local requirements [13]. For instance, the creep-loaded airfoil of a γ-TiAl turbine blade benefits from a FL microstructure, while a NL microstructure is a better fit for the root section of the blade [14]. This is known as the dual microstructure concept for γ-TiAl turbine blades [11,13,14].

If the processed powder possesses a homogenous aluminum distribution, the local aluminum content in the final part is governed by aluminum evaporation during PBF-EB, which mainly depends on the melt pool characteristics. In principle, aluminum evaporation is enhanced by increasing the peak temperature, size, and lifetime of the melt pool [11,15,16,17,18]. These melt pool characteristics can be manipulated by the process parameters.

Defocusing the electron beam slightly during melting reduces the power density within the beam, which leads to smaller and cooler melt pools and hence less Al evaporation [16,17]. Similar results are achieved by reducing the area energy or volume energy [12,18]. Knörlein et al. showed first that this dual microstructure concept is feasible for the γ-TiAl TNM alloy by adjusting the area energy and focus of the electron beam locally [11]. Further, the mechanical properties of specimens manufactured this way fit the requirements of the dual microstructure concept [13]. This will enable the next generation of γ-TiAl turbine blades towards higher application temperatures and consequently increase the efficiency of jet turbine engines [14].

The heat treatment window presented by Knörlein et al. is rather narrow [11]. According to the published data, defocusing the electron beam is not sufficient to produce a dual microstructure on the Arcam A2X 60 kV machine (GE Additive, Boston, MA, USA). By additionally increasing the area energy, a dual microstructure is achieved for heat treatment at 1260 °C [11]. Considering there is no heat treatment window solely relying on the focusing of the electron beam, it can be assumed that even with additional adjustment of the area energy the heat treatment window is only a few degrees wide, which will prevent industrial applications in the future.

A possible solution is adjusting the area energy over a wider range. However, the adjustment of the area energy for 60 kV acceleration voltage is restricted by the processing window of γ-TiAl [19]. The lower limit of area energy is given by the formation of misconnections, which are detrimental to the mechanical properties of γ-TiAl, and hence have to be avoided at all costs. Therefore, a safety factor is often added to produce γ-TiAl parts instead of operating at the minimal required area energy for dense samples. On the other hand, swelling of the samples sets the upper edge of the available area energies. Moreover, beam widening on the Arcam A2X for higher beam currents [20] limits the available process parameters, especially when considering scaling the beam powers to larger parts. Taking all these factors into account, it is remarkable that a dual microstructure is achieved for complex parts with conventional PBF-EB machines [11]. Nevertheless, advanced PBF-EB machines are necessary to enhance the heat treatment window of the dual microstructure concept of γ-TiAl for industrial application.

Progress in machine development makes PBF-EB machines with up to 150 kV acceleration voltage available, compared to 60 kV of conventional machines [1]. A recent study with a 150 kV prototype machine shows that the processing window of the TNM alloy is increased compared to 60 kV acceleration voltage, due to the larger penetration depth of electrons with higher acceleration voltage [19]. Moreover, higher beam powers are available for 150 kV. These results are supported by data from another research group investigating Ti-6Al-4V with 90 kV acceleration voltage [21]. This work demonstrates how the enhanced melting with 150 kV finally provides an industrial heat treatment window for the dual microstructure TNM alloy.

2. Materials and Methods

All samples are manufactured from gas-atomized Ti-44.5Al-4Nb-1Mo-0.1B (at%) powder produced by Allegheny Technology (Pittsburgh, PA, USA). The composition is the same as the TNM alloy with an increased aluminum content by 1 at% to compensate for the evaporation during the PBF-EB process. The particle size distribution is 60 µm (D10) to 145 µm (D90) with a D50 value of 95 µm and the oxygen content is 1250 ppm measured via EMGA 620W (HORIBA, Kyoto, Japan). The powder possesses a gas porosity of 0.1%.

The layered cuboids with the dimensions 15 mm × 15 mm × 17.5 mm (XYZ) are produced on the prototype PBF-EB machine HELIOS: the prototype is developed using pro-beam GmbH & Co. KGaA (Gilching, Germany) and is equipped with a standard electron beam welding gun capable of 150 kV acceleration voltage and up to 45 kW power. The electron beam is controlled by a script, developed at the Chair of Material Science and Technology for Metals at the Friedrich-Alexander-Universität (Erlangen, Germany). Subsequent to lifting the powder tank and lowering the build tank by the layer thickness, the rake moves the powder from the powder tank to the build area. Excess powder is moved into an overflow tank. Before melting with a cross snake hatch pattern, a standard pre-heating of the powder bed to 1000 °C is performed. The beam quality is described in Section 3.1. For further information about the PBF-EB machine and the process, the reader is referred to [19].

All specimens are built with 150 kV acceleration voltage (U_b), 900 W beam power (P_b), 6 mA beam current (I_b), 100 µm line offset (l_o), and 100 µm layer thickness. The area energy (E_a) is adjusted to 1.20 J/mm² and 2.00 J/mm² by setting the beam velocity (v_b) to 7500 mm/s and 4500 mm/s, respectively (see Equation (1)).

$${\mathrm{E}}_{\mathrm{a}}=\frac{{\mathrm{U}}_{\mathrm{b}}\ast {\mathrm{I}}_{\mathrm{b}}}{{\mathrm{v}}_{\mathrm{b}}\ast {\mathrm{l}}_{\mathrm{o}}}=\frac{{\mathrm{P}}_{\mathrm{b}}}{{\mathrm{v}}_{\mathrm{b}}\ast {\mathrm{l}}_{\mathrm{o}}}$$

(1)

Scanning strategy is a standard cross snake hatching with 90° rotation each layer and no contours. Both parameters are within the dense processing window, which is published in a previous work [19]. The focus current is varied between 2980 mA and 3060 mA in 20 mA steps, with each 3.5 mm build height within one cuboid. In total, six cuboids per area of energy are produced in two build jobs. After manufacturing, the as-built cuboids are heat-treated in a vacuum furnace LHTM 250/300 (CarboliteGero GmbH und Co. KG, Neuhausen, Germany). Overall, five heat treatments are conducted at temperatures from 1230 °C to 1270 °C in 10 °C increments. The temperature is held for 30 min followed by fast cooling with argon.

Finally, the samples are cut, ground, and polished with a suspension of 50 mL OPS, 50 mL distilled water, 10 mg KOH, and 10 mL 30% H₂O₂. Gas porosity and misconnections are determined on light optical images of the cross sections taken with a Zeiss AxioImager.M1m (Carl Zeiss AG, Jena, Germany). The defects are measured on all six cuboids, three per build job. For each parameter, images with 100× magnification are taken and analyzed by an automated script. The analyzed xz-plane is 12 mm × 8 mm large. The aluminum content is quantified via electron probe microanalysis (EPMA) with the Jeol JXA 8100 (JEOL, Akishima, Japan). The overall aluminum content is determined on three cuboids, two of the first build job and one of the second, via EPMA. The aluminum content is measured in the middle of the cross-section in a 1.5 mm wide band over the total height of the samples. Therefore, the measurement area is 1.5 mm × 3.5 mm per parameter. The resolution of the measurement is 15 µm. In addition, more detailed EPMA measurements are conducted to evaluate the homogeneity within one cuboid. The measured area is located in the middle of the larger measurement for each parameter, respectively. The area is 500 µm × 500 µm large and the local resolution is 2.5 µm. Microstructure images are acquired with the scanning electron microscope (SEM) Helios NanoLab 600i (FEI Company, Hillsboro, OR, USA) by a circular backscatter electron detector (CBS).

The beam diameter and beam profile for each focus current are measured with a pinhole sensor based on a Faraday cup and the EB Vision software (v22.8.2.8571) both provided by pro-beam GmbH & Co. KGaA. The beam diameter is defined as D₈₆ by the second moment of area, which is the same as the 4σ diameter for a perfect Gaussian beam power distribution, following ISO 11146-1.

3. Results

3.1. Electron Beam Characteristics

Figure 1 shows the beam profile measurement for the used electron beam with 150 kV acceleration voltage, 6 mA beam current, and 900 W beam power. In an under-focused state, the beam shows a top-hat beam profile, which resembles the shape of the emitting tungsten filament. The top-hat beam profile is characterized by a near-uniform power distribution. The description top-hat is derived from the 2D beam profile as can be seen for 2991 mA in Figure 1. By increasing the focus current, also called focusing the electron beam, the beam diameter gets smaller, and the beam profile is changed to a Gaussian power distribution (see 3021 mA and 3041 mA). The 3001 mA focus current shows an intermediate state, in which a maximum of the power density starts to form, but the Gaussian profile is not yet fully visible. When increasing the focus current further to 3061 mA, the electron beam is defocused while a Gaussian beam power distribution is maintained. This state is called over-focused. The results fit well with other measurements in the literature [22].

While looking at the power distribution within the electron beam is the most accurate way to investigate the electron beam characteristics, this information is not always available. The beam power density can be used as an alternative if only the beam diameter is known. The beam power density (q_b) is the beam power (P_b) divided by the area of the beam, which can be calculated from the beam diameter (d_b), see Equation (2).

$${\mathrm{q}}_{\mathrm{b}}=\frac{{\mathrm{P}}_{\mathrm{b}}}{\mathsf{\pi }\ast ({{\mathrm{d}}_{\mathrm{b}}/2)}^2}$$

(2)

3.2. Gas Porosity

To analyze the defects in the microstructure, micrographs of all samples are analyzed. Since the micrographs showed no misconnections, only the gas porosity is shown in Figure 2. The determined gas porosity of the samples with 1.2 J/mm² and 2.0 J/mm² area energy is plotted over the five different focus currents in Figure 2. The average of the six cuboids is displayed with a standard variation. Additionally, the used beam diameter is displayed.

The samples with higher area energy (2.0 J/mm²) show less porosity than the samples with lower area energy (1.2 J/mm²). This correlation is also reported by other authors [11,18]. A higher area energy leads to a longer-lasting melt pool [18], resulting in more time for the entrapped gas to rise to the surface.

In the second step, the influence of focusing the electron beam is analyzed. Samples with a larger beam diameter (defocused) possess a higher gas porosity than samples with a small beam diameter (focused). This fits the literature data as well [16,17]. The focused electron beam increases the melt pool lifetime and hence reduces the gas porosity. Interestingly, an electron beam with a similar diameter, but a different beam profile (3000 mA vs. 3060 mA), results in a different gas porosity. This will be discussed later on.

3.3. Aluminum Content

To investigate the overall aluminum content for each parameter, a large area (1.5 mm × 3.5 mm) with a low resolution (15 µm) is analyzed via EPMA. The samples with a lower area energy (1.2 J/mm²) show higher overall aluminum contents (see Figure 3). Specimens built with 2.00 J/mm² possess a lower aluminum content since a higher area energy enhances aluminum evaporation [19].

Moreover, a defocused electron beam leads to less aluminum evaporation for each area energy, respectively. The defocused electron beam distributes the 900 W used in this study over a larger area. Consequently, the resulting melt pool is cooler and the aluminum evaporation is reduced.

Both a lower area energy and a defocused electron beam favor a cooler, smaller, and shorter living melt pool, which reduces the aluminum evaporation during the process. These results are in good agreement with the literature [11,12,15,16,17,18].

The standard deviation of the three analyzed cuboids (two of the first build job, one of the second build job) is mostly smaller than 0.1 at%, indicating good reproducibility within one build job as well as in between two build jobs.

Figure 4 displays the aluminum content of the detailed EPMA measurement on a smaller area (500 µm × 500 µm) with a higher resolution of 2.5 µm. Additionally, the median aluminum content of the evaluated area is given below the mapping. The standard deviation within one area is 0.7 at% for each area, respectively. The aluminum contents of the detailed EPMA measurements are in good agreement with the measurements of the larger areas above (Figure 3). Since the mappings show 500 µm × 500 µm of the xz-plane, four complete 100 µm layers should be visible. Although small fluctuations are visible, the overall aluminum distributions seem rather even. Small fluctuations can in part be attributed to the different composition of the α₂ phase and γ phase, since some microstructural features (see Section 3.4) are larger than the local resolution of this EPMA measurement.

Furthermore, the median aluminum content of each line perpendicular to the build direction is displayed as a black line in Figure 5. The grey area shows the 0.05 percentile to the 0.95 percentile of the aluminum content within each line, respectively. The median aluminum content of the whole area (see Figure 4) is the center of each x-axis. Independent of the used parameter, the median aluminum content only scatters slightly around the area median, indicating a good homogeneity. The range of aluminum content within one line is mostly smaller than ±1.5 at%. Moreover, no distinct boundaries or large differences between the layers can be found in the aluminum content, which highlights the homogeneity of the samples further.

3.4. Microstructure

Table 1. Overview of experimental results. The microstructure after heat treatment (NL = nearly lamellar (blue background), FL- = almost fully lamellar (white background), and FL = fully lamellar (orange background)) is given for 1.2 J/mm² and 2.0 J/mm² for all five focus currents. Additionally, the measured aluminum content and beam diameter as shown above are listed. Considering the actual measured temperatures during heat treatment, a heat treatment window of 25 °C is identified. Suggested parameters for dual microstructure parts are marked by thick borders.

heat treatment target [°C]			1230	1240	1250	1260	1270	Al content [at%]	beam diameter [µm]
heat treatment measured [°C]			1223	1234	1242	1259	1266
1.2 J/mm² area energy	focus current [mA]	3060	NL	NL	NL	NL	FL-	44.2	576
		3040	NL	NL	NL	FL-	FL	43.9	388
		3020	NL	NL	NL	FL	FL	43.8	386
		3000	NL	NL	NL	FL	FL	43.9	559
		2980	NL	NL	NL	NL	FL	44.1	830
2.0 J/mm² area energy	focus current [mA]	3060	NL	FL-	FL	FL	FL	43.3	576
		3040	NL	FL	FL	FL	FL	43.0	388
		3020	NL	FL	FL	FL	FL	43.1	386
		3000	NL	FL	FL	FL	FL	43.1	559
		2980	NL	FL	FL	FL	FL	43.3	830

gives an overview of the microstructure analysis after heat treatment. Additionally, the aluminum content and beam diameter are displayed as presented above. To evaluate the experiments, it is crucial to examine the actual measured temperature during heat treatment, not the target temperature. Most of the parameters show either nearly lamellar (NL) or fully lamellar (FL) microstructure, which can be attributed to the homogeneous aluminum distribution within the samples (see Figure 4). Samples with 1.2 J/mm² area energy and hence high aluminum contents are mostly NL and only transform to FL for high temperatures. On the other hand, specimens with 2.0 J/mm² possess less aluminum and hence display a FL microstructure for lower temperatures. A few samples show an almost fully lamellar (FL−) microstructure, consisting mostly of FL with small regions of NL. This microstructure results from a heat treatment slightly above the γ-solvus, which is the border between the α phase field and the α + γ phase field. During heat treatment near the γ-solvus, even small deviations in the aluminum content provoke different microstructures, and hence small regions with a higher aluminum concentration remain NL, while most of the sample is transformed into an FL microstructure. Exemplary SEM images of the microstructures after heat treatment taken with a CBS detector are shown in Figure 6.

Looking at the measured temperatures, a heat treatment window from 1234 °C to 1259 °C for the dual microstructure concept can be established. To generate dual microstructure parts, adjusting either the focus or the area energy is a viable option. In general, every combination of NL and FL with any of the tested parameters and heat treatments is possible. However, to reliably produce dual microstructure parts despite small deviations in aluminum content and heat treatment, adjusting the area energy and focus current at the same time is recommended (see marked parameters in Table 1).

4. Discussion

4.1. Influence of Area Energy and Focusing of the Electron Beam

The dependence of the gas porosity and aluminum content on the focus current and area energy is in good agreement with the literature as described above [11,12,15,16,17,18]. In general, a higher area energy and a stronger focused beam (smaller beam diameter) lead to a larger and longer-lasting melt pool with increased temperature [16,17,18,20]. These melt pool characteristics enhance aluminum evaporation and reduce gas porosity. These trends from the literature are confirmed by the results of this study for 150 kV acceleration voltage.

However, the slightly under-focused beam (3001 mA), which shows neither a top-hat profile nor a Gaussian beam power distribution, leads to similar results as the focused Gaussian beam profiles (3021 mA and 3041 mA), especially considering the aluminum content (Figure 3 and Figure 4). This is somehow unexpected since the beam diameter is larger (559 µm) compared to the focused electron beam (386 µm and 388 µm). Similarly, the top-hat profile (2991 mA) achieves similar results as the defocused Gaussian beam (3061 mA), although the beam diameters are dramatically different. These results imply that the calculation of the electron beam diameter following ISO 11146 works well for Gaussian power distributions, while the beam diameter calculation for non-Gaussian profiles seems to be inaccurate. Nevertheless, this shows the potential of innovative process strategies for PBF-EB by manipulating not only the electron beam diameter but also the power distribution within the electron beam.

4.2. Quasi-Binary TNM Phase Diagram

The quasi-binary phase diagram Ti-xAl-4Nb-1Mo (at%) proposed by Knörlein et al. [11] is confirmed by the results of this study (see Figure 7). In particular, the location of the γ-solvus line, which is essential to achieve a dual microstructure, is approved by the experimental results.

The narrow α phase field of the TNM alloy, which is roughly 1 at% wide needs to be stressed. This makes it difficult to conduct a heat treatment for a dual microstructure since the formation of the FL microstructure is hindered if regions of the sample are still in the α + γ phase field (Al-enriched) or already in the α + β phase field (Al-lean). This will be important for the discussion of the improved heat treatment window for 150 kV.

4.3. How Advanced PBF-EB Technology Increases Robustness

Finally, this study reveals improvements in the dual microstructure concept when using 150 kV acceleration voltage. The heat treatment window is enhanced from a few degrees Celsius [11] to 25 °C (see Table 1), which is crucial for industrial applications due to temperature differences in industrial furnaces. Since the difference in aluminum content determines the success of a dual microstructure heat treatment [11], this was the presumed reason for the increased heat treatment window. Studies conducted with 60 kV acceleration voltage reported aluminum differences of 0.7 at% [23] and 0.9 at% [11]. The 150 kV acceleration voltage enables an aluminum difference of 1.0 at% to 1.2 at%. The primary reason for this increased difference is the reduced area energy, which is required to produce dense samples [19]. As a result, the aluminum evaporation of the low energy parameter is reduced and the difference between the parameters increases. However, the increase is too small to account for the largely widened heat treatment window. An additional factor for the formation of a dual microstructure is the homogeneity of the aluminum content, which will be discussed in the following.

Several studies conducted by other research groups with 60 kV report a banded or layered microstructure for Ti-48Al-2Nb-2Cr (at%) [17,18,24] and TNM [11,23]. On the contrary, no such banded microstructures are found in this study, indicating more homogeneous samples. Even the almost fully lamellar (FL-) samples show vastly fully lamellar microstructures with separated, small islands of nearly lamellar microstructure. Neither of the FL- specimens displays bands or layers of NL microstructure. This is supported by the detailed EPMA measurements (Figure 4), which show a homogeneous aluminum distribution over several building layers.

When directly comparing the aluminum content of parameters used for dual microstructure samples with 60 kV and 150 kV (Figure 8), the difference in homogeneity is visible. The 60 kV samples are part of the work published by Knörlein et al. [11]. The authors of the study kindly provided the raw data of the electron probe microanalysis. The aluminum content is shown in an area of 1.5 mm × 1.5 mm with a resolution of 15 µm. It is noted that every 15 µm × 15 µm data point contains a different share of Al-rich γ and Al-lean α₂, which in part is responsible for the Al differences. To fairly compare the homogeneity and not the absolute aluminum content, the median aluminum content of each area is calculated. Subsequently, the data are plotted with a color bar ±2.0 at% around the median value for each of the four samples, respectively. Gas pores are also depicted in blue. While the 60 kV samples show large aluminum depleted (blue) and enriched (red) clusters, especially for the focused parameter with high area energy, the aluminum is homogeneously spread in the 150 kV samples. Interestingly, the standard deviation over the whole area is similar for all four samples (0.7 at%–0.8 at%).

To quantify the homogeneity of the samples, the median aluminum content is calculated for each line perpendicular to the build direction. The median aluminum content is chosen to reduce the influence of the gas pores on the calculated value of the affected lines. While the 60 kV samples display several distinct peaks over the build height, the line median of the 150 kV samples mostly scatters around the median aluminum content of the whole area. Interestingly, the range of measured aluminum (0.05 percentile to 0.95 percentile) is similar for all four samples.

Therefore, the key to a larger heat treatment window is avoiding the Al-rich or Al-lean clusters, which are illustrated by the peaks of the aluminum line medians of the 60 kV samples. To understand why these clusters are so harmful, a closer look at the microstructure evolution during the heat treatment is necessary.

To form a fully lamellar microstructure, a heat treatment in the narrow α phase field is necessary. During heat treatment, the Al-lean regions transform to α first. Subsequently, the α grains grow until they reach a competing α grain or are restricted by thermodynamically stable γ (Al-rich regions). In the case of the 150 kV samples, the Al-rich regions (red dots in Figure 8) are small and hence are quickly overgrown by α grains, which are in the range of 100 µm to 200 µm (see Figure 6). On the contrary, the Al-rich clusters in the 60 kV sample have a similar size or are larger than the α grains. Therefore, the α grains formed in the Al-lean clusters are limited in their growth by the Al-rich clusters and a layered microstructure is formed. A similar argument with opposite roles can be made for the nearly lamellar microstructure.

While this explains the larger heat treatment window of the 150 kV samples, the reason for the formation or prevention of these clusters remains unclear. Most authors argue that aluminum evaporates at the top of the melt pool and hence a gradient in aluminum is formed within the melt pool [17,18,25]. However, the layers or bands presented in the literature are often in the same magnitude as the layer thickness, which contradicts this explanation. Moreover, the melt pool during PBF-EB is highly dynamic, which should at least lead to some homogenization within the melt pool. Further, all studies with 60 kV acceleration voltage are performed on Arcam A2X machines, so a machine dependency cannot be excluded for sure.

Since the microstructure and its homogeneity are mainly controlled by the melt pool, there has to be an impact of the presented advanced PBF-EB technology on the melt pool formation. However, due to the nature of the melt pool with its small size, short lifetime, and high temperatures, it is very hard to study it experimentally. To gain a thorough understanding of the impact of the higher acceleration voltage on the melt pool and hence the formation of the final microstructure, a combined experimental and numerical approach should be chosen for future experiments.

5. Conclusions

Electron beam-powder bed fusion with 150 kV acceleration voltage can enable the production of dual microstructure γ-TiAl turbine blades in an industrial environment. This is achieved by two improvements:

• Reducing the necessary area energy for dense samples decreases the aluminum evaporation for the defocused parameter with low area energy. Consequently, the difference in aluminum content between the two dual microstructure parameters is increased to ≥1 at% compared to 0.7 at%–0.9 at% for 60 kV.
• A more homogenous microstructure is achieved by 150 kV acceleration voltage. Although the underlying mechanisms are not yet fully understood, the higher acceleration voltage leads to a more homogenous aluminum distribution without the formation of clusters or bands.

The combination of a larger aluminum difference and a more homogeneous microstructure enhances the heat treatment window from a few degrees Celsius for 60 kV to 25 °C for 150 kV. This heat treatment window is crucial to realize the dual microstructure concept for complex parts in industrial furnaces.