A Comparative Study of Microstructure and Hot Deformability of a Fe–Al–Ta Iron Aluminide Prepared via Additive Manufacturing and Conventional Casting

Abstract

In this work, the microstructure and hot deformation behavior of laser powder bed fusion (L-PBF) and conventionally cast Fe-25Al-1.5Ta (at.%) alloys were compared. The L-PBF builds recrystallized comparably to the as-cast samples during hot deformation. Nevertheless, distinct differences were observed in the flow behavior characteristics between the as-cast and L-PBF samples. The L-PBF builds exhibited lower flow stress than the as-cast material over the entire deformation conditions tested. The average activation energy of hot deformation (Q) of 344 kJ mol⁻¹ was calculated for the L-PBF build and 385 kJ mol⁻¹ for the cast material. The lower Q indicates lower deformation resistance of the L-PBF sample. The peak work hardening rate (θ) in the L-PBF sample (1.72 × 10³ MPa) was significantly smaller than that of the as-cast sample (3.02 × 10³ MPa), suggesting that the dislocation glide in the L-PBF sample is less hindered during deformation. Possible sources of the observed differences in the deformation behavior between the L-PBF and cast materials will be discussed. Initial and post-deformation microstructures were characterized using an X-ray diffractometer (XRD) and ultra-high-resolution scanning electron microscopy (SEM) equipped with energy-dispersive X-ray spectroscopy (EDX) detector. The C14-(Fe, Al)₂Ta Laves phase (P6₃/mmc) was predominantly formed at the A2 α-(Fe, Al) matrix phase grain boundaries in both the as-cast and L-PBF materials. The XRD results suggest that the ordering transition from B2-FeAl to a D0₃-Fe₃Al phase occurs during casting, but rarely during ultra-high-cooling L-PBF processing. In summary, the L-PBF creates samples that are subject to less work hardening and require less deformation resistance, and thus, can be formed by a lower deformation force. It, in turn, reduces the loads imposed on the tooling and dies during the deformation processing, contributing to less wear and the high durability of dies.

1. Introduction

Iron aluminides, Fe–Al-based alloys containing either disordered A2 α-(Fe, Al), B2-ordered Fe–Al, or D0₃-ordered Fe₃Al as majority phases, have long been considered excellent candidates to replace heat-resistant steels or possibly even superalloys in high-temperature applications up to 800 °C, primarily because of their excellent oxidation and corrosion resistance even in aggressive environments [1,2,3]. They have a lower density and are less expensive than steels and superalloys; however, they exhibit low strength at temperatures above 800 °C [3].

Fe–Al–Ta alloys with strengthening (Fe, Al)₂Ta Laves phase were identified as more qualified for structural applications at and above 600 °C than its binary counterpart due to a superior creep resistance [4,5,6,7]. For example, the creep resistance of the Fe-25Al-2Ta (all compositions are given in atomic percent throughout the text unless otherwise stated) alloy at 650 °C surpasses that of the P92 (X10CrWMoVNb9-2) martensitic-ferritic steel, which is one of the most creep-resistant alloys developed for steam turbine applications [6]. The Fe-25Al-1.5Ta alloys produced by casting and spark plasma sintering revealed a higher degree of hot deformability over a wide temperature and strain rate range [8,9,10].

A significant problem in manufacturing Laves-phase strengthened iron aluminides by conventional casting is the strong coarsening tendency of the Laves phase precipitates at elevated temperatures, leading to a significant strength reduction [9,11]. In this context, state-of-the-art metal additive manufacturing (AM) promises to consolidate these aluminides with homogeneously distributed fine precipitates due to the short lifetime of the melt pool and fast solidification as well as cooling rates. Likewise, AM allows for the fabrication of near-net-shape complex geometries, which enables the realization of more efficient and lightweight designs, more sustainable part manufacturing, fast prototyping, and reduced machining costs [12,13,14].

Several attempts to produce iron aluminides by additive manufacturing have been reported [15,16,17,18,19,20,21,22]. However, to the authors’ best knowledge, no investigation has been conducted to compare the microstructure and deformation behavior of the AM and as-cast iron aluminides reinforced by Laves precipitation particles. In an earlier study by the authors [23], laser powder bed fusion (L-PBF), also known as selective laser melting (SLM), of the Fe-25Al-1.5Ta alloy was investigated in terms of processability and microstructure. Crack-free samples of anisotropic microstructures with densities higher than 99% were fabricated from gas-atomized and pre-alloyed Fe-25Al-1.5Ta powders. In the present work, a comparative study of the microstructure and deformation behavior of Fe₃Al-1.5Ta alloys prepared via additive manufacturing and conventional casting was conducted. The fundamentals of the process–microstructure-property relationships of the L-PBF alloy are provided by a detailed microstructural characterization using SEM/EDX, EBSD, OIM, XRD, and hot compression tests. The hot deformation behavior of L-PBF builds was studied compared to that of the as-cast material to evaluate the applicability of L-PBF as a valid alternative to standard processing routes. Such knowledge is vital, but the literature is insufficient considering the current attempt to use advanced aluminides in fossil-fuel power plants, aviation, and marine engines [24,25].

2. Materials and Methods

An alloy powder with a nominal composition of Fe-25Al-1.5Ta was produced by gas-atomization by NANOVAL GmbH & Co. KG, Berlin, Germany. The powder batch with a fraction size of +10/−45 µm and an average particle size of d₅₀ = 22.6 μm was used for further processing by L-PBF. Figure 1 shows a scanning electron microscopy (SEM) micrograph of the initial loose particles in the backscatter electron mode (BSE). The Fe-25Al-1.5Ta alloy powder particles were characterized by spherical morphology, with relatively smooth surfaces exhibiting a typical dendritic microstructure.

Table 1. The average concentration of the constituting elements measured by EDX on several powder particles.

Elements Concentration, at. %
Al	Fe	Ta
23.54 ± 3.30	74.77 ± 3.52	1.67 ± 0.42

presents the results of energy-dispersive X-ray spectroscopy (EDX) analysis performed on several powder particles, indicating that the average composition of the powder marginally deviated from the nominal composition.

The L-PBF experiments were performed with a 400 W AconityMIDI (Herzogenrath, Germany) laser source under an argon atmosphere. The laser power P, the scanning speed v_s, the layer thickness, and the hatching distance were 250 W, 1000 mm/s, 50 μm, and 90 μm, respectively. A stainless steel plate with a thickness of 10 mm preheated to 800 °C was used as a substrate material. Parallel stripes with a width of 5 mm were used with the scan vectors on successive layers rotating at an angle of 67°. Porosity analysis showed that the void volume in relation to the total volume of the as-built samples was ∼1%. A typical L-PBF build is shown in Figure 2a.

The phase identification was performed by a Bruker D8 ADVANCE X-ray diffractometer using Cu-K_α1 radiation (λ = 0.15406 nm) within the 2θ range of 20–120° with a step size of 0.1°. Crystallographic data [26,27] for the present phases in Fe-25Al-1.5Ta are listed in

Table 2. Crystal structure data for the present phases in the Fe-25Al-2Ta alloy.

Phase	Pearson Symbol	Space Group No.	Structure Designation	Prototype
α-Fe	cI2	$Im\overline{3}m$	A2	W
FeAl	cP2	$Pm\overline{3}m$	B2	CsCl
Fe3Al	cF16	$Fm\overline{3}m$	D03	BiF3
TaFe2Al, Heusler phase	cF16	$Fm\overline{3}m$	L21	MnCu2Al
TaFe2, Laves phase	hP12	P63/mmc	C14	MgZn2

.

The macrostructures of the L-PBF fabricated specimens sectioned parallel to the BD were investigated by light optical microscopy (LOM–VHX Digital Microscope, KEYENCE America). Microstructures of the samples were characterized by a field-free ultra-high resolution (0.9 nm at 15 keV, 1.3 nm at 1 keV) scanning electron microscopy (SEM) on a TESCAN AMBER (Brno, Czech Republic) equipped with an energy-dispersive X-ray spectroscopy (EDX—Oxford Instruments NanoAnalysis, AZtec version 5.0, Abingdon, UK) detector. The SEM analyses of the deformed specimens were carried out in the central regions of the samples where deformation had localized.

Laboratory scale isothermal compression tests were performed on a DIL805A/D/T deformation dilatometer. Specimens of Ø 5 × 8 mm were electric-discharge machined and heated to temperatures in the range of 900 °C to 1100 °C with a rate of 10 K/s and held for 3 min to obtain a homogeneous temperature distribution in the samples before the compression test. Subsequently, the specimens were deformed at strain rates from 0.0013 s⁻¹ to 0.1 s⁻¹ to a true strain of 0.8 in an argon atmosphere. Cooling was conducted in the air immediately after the compression tests to preserve the high-temperature microstructures. Typical machined and deformed specimens are shown in Figure 2b,c.

For comparison, the microstructure and hot deformation behavior of the as-cast material with the same chemistry are also reported. An ingot with a diameter of 30 mm was cast using an investment casting procedure by Access e.V. (Aachen, Germany) [9]. For more details on the microstructure of the as-cast and SPS materials, the readers are referred to [8,9].

3. Results and Discussion

3.1. Microstructure Characterization of the L-PBF Builds

Figure 3 exhibits the X-ray diffractogram of the L-PBF-produced Fe-25Al-1.5Ta sample. The primary diffraction pattern of the as-cast material is presented in the same figure. The L-PBF builds were characterized by fundamental reflections of the disordered A2 α-(Fe, Al) phase and characteristic superstructure peaks of (100) and (101) B2-type ordered structures occurring at around 21.29° and 30.79°, without reflections from the D0₃-type ordered phase. The characteristic reflections of (111)_D03 and (311)_D03 should occur at around 26.68° and 52.41° using Cu-K_α1 incident radiation; however, they were not present in the XRD pattern of the L-PBF builds. Apart from the matrix reflections, peaks of the C14 (Fe, Al)₂Ta Laves phase were found at around 2θ = 35.04°, 40.59°, and 58.94°.

In contrast to the L-PBF builds, the as-cast material exhibited the characteristic reflections of (311)_D03 occurring at around 52.41°, in addition to the primary reflections of the B2 and C14 phases. It should be mentioned that the L2₁ TaFe₂Al Heusler-type phase, which is the ternary equivalent of D0₃ Fe₃Al, had the same crystallographic structure as D0₃. Therefore, the primary X-ray diffraction peaks of the L2₁ Heusler phase will overlap with those of D0₃ [5] (see Table 2).

The phase identification results in the present study agreed well with the literature. The presence of the metastable B2 and the suppression of the equilibrium D0₃ at room temperature were typically found for the Fe₃Al-based materials subjected to the high cooling rate LENS fabrication [21,28]. Likewise, water quenching from 900 °C prevented D0₃ ordering in the binary Fe-27Al alloy at room temperature; however, it did not preserve the high-temperature A2 structure, and the sample had the B2 structure at room temperature, according to the neutron diffraction and XRD data [29,30]. Casting, in contrast, allows for slower cooling rates. Thus, the B2 to D0₃ transition may occur during cooling to room temperature. The phase identification results in the Fe-25Al-1.5Ta alloy produced with the investment casting procedure revealed the presence of D0₃-ordered and the C14 Laves phases [31]. Likewise, Fe₃Al samples in the as-cast polycrystalline bulk state constituted a partially ordered B2 structure (matrix) with dispersed clusters of the D0₃-ordered phase [32].

Due to the ultra-fast cooling during the L-PBF process, the disordered A2 phase is expected to be preserved at room temperature. Moreover, the L-PBF does not allow the transition of B2 to D0₃; therefore, the D0₃ ordering is suppressed, and B2 is retained at room temperature. The L-PBF builds are thus characterized by a disordered A2 matrix containing ordered B2 phase domains. This conclusion is in good agreement with the work of A. Balagurov et al. [32], who reported that the bulk polycrystalline (Fe_0.88Cr_0.12)₃Al alloy in the quenched state consists of a disordered A2 matrix in which dispersed clusters of the partially ordered B2 phase are embedded. The results suggest that the ordering transition from the B2 to a D0₃ phase may occur during casting but hardly occur during ultra-high cooling processing (e.g., L-PBF).

Figure 4 presents typical BSE-SEM micrographs of the L-PBF fabricated samples. The as-built sample exhibited the characteristic columnar grain morphology of additively manufactured metallic materials [33]. The microstructure primarily consists of large columnar grains elongated along the vertical BD by the epitaxial solidification and growth mechanism. A comparable grain morphology and texture was reported for FeAl and Fe₃Al parts fabricated by AM [19,20,22,34].

As the higher magnification image in Figure 4 shows, the Fe–Al matrix grains were decorated with a phase with white BSE contrast, as shown by arrows. According to previous studies [8,31] and the XRD results in Figure 3, this phase corresponded to the C14 Laves phase (Fe, Al)₂Ta with hexagonal crystal symmetry. Here, it was reported that the solubility of Ta in the binary Fe–Al phases is generally low and varies with Al content [35]. For example, the solubility of Ta in Fe-25Al alloys in equilibrium with the Laves phase is about 0.7% at 1000 °C and 0.5% at 800 °C and less at lower temperatures [35]; thus, the addition of small amounts of Ta to Fe(Al) solid solution phases leads to the precipitation of the ternary Laves phase, resulting in extended two-phase fields.

In contrast to the as-cast alloy [9,35], where the Laves phase precipitates are primarily located at the Fe–Al grain boundaries, the L-PBF builds in the present study also contained a significant fraction of Laves phase particles finely dispersed within the Fe–Al grains.

Table 3. The average concentration of elements in at. % within the Fe–Al–Ta matrix and (Fe, Al)₂Ta Laves phase determined by EDX in the L-PBF and in the as-cast samples.

Sample	Fe-Al-Ta Matrix			C14 Laves Phase
	Fe	Al	Ta	Fe	Al	Ta
L-PBF	77.8 ± 0.1	21.5 ± 0.1	0.7 ± 0.1	66.8 ± 0.7	9.1 ± 0.5	24.2 ± 1.2
As-cast	74.4 ± 0.28	24.8 ± 0.2	0.9 ± 0.1	59.4 ± 0.27	10.4 ± 0.29	30.2 ± 0.35

presents the average concentrations of the constituting elements of the Fe–Al–Ta matrix phase and the (Fe, Al)₂Ta Laves phase precipitates measured by EDX in both materials. The quantitative analysis of the matrix in the L-PBF area showed 77.6–77.9 at. % Fe, 21.4–21.7 at. % Al, and 0.6–0.8 at. % Ta, while the composition of the Laves phase was 66.3–67.8 at. % Fe, 8.7–9.8 at. % Al, and 22.4–24.9 at. % Ta. The results indicate that the Laves phase in both materials was considerably enriched in Ta compared to the matrix. An essential point to highlight is that both samples contained nearly the same amount of Ta; however, the as-cast sample included a comparatively higher Al content of 24.8 at. % than the L-PBF sample with only 21.5 at.%. This could be due to the vaporization of more volatile Al atoms due to high energy input during the L-PBF process. The possible influence of a lower Al content on the deformation behavior of the L-PBF specimens will be discussed later.

3.2. Hot Deformation Behavior and Microstructure Evolution

Figure 5 shows an overview of the flow curves and initial as well as the post-deformation microstructures of the L-PBF and the as-cast materials. The true strain–true stress curves of both materials (Figure 5a) exhibited a shape with a single maximum. This shape is typical for conditions where DRX leads to grain refinement. The flow stress rapidly increases to a peak value at the early stage of an imposed deformation and then gradually decreases to a steady state with increasing strain. The main flow softening mechanisms are dynamic recovery (DRV) processes and dynamic recrystallization (DRX) processes. They play a significant role in lowering the flow stress and thus in reducing the forces during the mechanical shaping processing. The microstructure characterization shown in Figure 5d–g indicates that the DRX microstructure in the L-PBF builds was similar to that in the as-cast material.

The rise in flow stress after a steady state at a true strain of 0.7–0.8 in several specimens was due to the dominance of friction and bulging of the specimens; therefore, the flow curves show the true stress only up to this point.

The critical point is that the L-PBF builds exhibited lower flow stress than the as-cast material across all of the tested deformation conditions. To compare the resistance to deformation between L-PBF and the as-cast materials, the activation energy of hot deformation (Q, [kJ mol⁻¹]) was calculated by [36]:

$$Q=R· {\left(\frac{\partial ln\dot{\epsilon }}{\partial ln\left[\sinh \left(\alpha \sigma \right)\right]}\right)}_T· {\left(\frac{\partial ln\left[\sinh \left(\alpha \sigma \right)\right]}{\partial \left(1/T\right)}\right)}_{\dot{\epsilon }}$$

(1)

where $\dot{\epsilon }$ is the strain rate (in s⁻¹); σ is the stress (in MPa); Q is the apparent activation energy of hot deformation (in kJ mol⁻¹); T is the absolute deformation temperature (in K); R is the universal gas constant (8.314 J mol⁻¹ K⁻¹), and A, α, and n are material constants independent of the deformation temperature.

The average activation energy was 344 kJ mol⁻¹ for the L-PBF build and 385 kJ mol⁻¹ for the cast material. The values of Q obtained in the present study were remarkably larger than those of Fe self-diffusion in binary Fe₃Al alloys (Q_A2 = 219 kJ mol⁻¹, Q_B2 = 234 kJ mol⁻¹, Q_D03 = 280 kJ mol⁻¹ [37]). This suggests that the Ta solutes significantly contribute to the pinning of dislocations and complicate the hot deformation process, leading to an enhanced Q for Ta-containing Fe₃Al alloys compared to the binary Ta-free counterpart.

The critical point is that a smaller value of Q for the L-PBF builds suggests that they exhibit a lower resistance to deformation and therefore require a lower force to deform. The lower deformation force is crucial to reduce the stress on the dies and tools during the deformation process and to extend their lifetime.

Figure 5b displays the work hardening rate (θ = dσ/dε) as a function of the true strain curves for both materials after deformation at 1000 °C/0.0013 s⁻¹. Both samples showed similar work hardening behavior with strain. The work hardening increased rapidly to a peak value at the early stage of imposed deformation up to the yield point and then decreased continuously at higher strain. Importantly, the peak θ in the L-PBF sample was significantly smaller (by about half) than that of the as-cast sample, indicating that the dislocation movements in the L-PBF sample were less pinned during deformation.

Figure 5c illustrates the work-hardening–true stress curves (so-called Kocks–Mecking plots) plotted for both materials after deformation at 1000 °C/0.0013 s⁻¹. The critical stress for the initiation of DRX was identified from the inflection point on the θ–σ curves [38]. The DRX initiated at a stress of 19.75 MPa for the L-PBF sample, corresponding to a true strain of 0.020 and 24.16 MPa for the as-cast samples, which corresponded to a true strain of 0.026. Therefore, it can be concluded that the L-PBF sample is subjected to DRX at a comparatively lower stress than the cast material.

The following section will discuss the possible sources of the observed differences in flow behavior and work hardening between the studied as-cast and L-PBF materials.

3.2.1. Grain Size and Morphology Effect on the Flow Behavior

The L-PBF sample revealed a heterogeneous grain size distribution, in contrast to the as-cast sample, which showed mainly equiaxed grains with a relatively homogeneous grain size distribution, as shown in Figure 5d,f. The average grain size was obtained from the EBSD data using a conventional mean-linear-intercept method. The grain size of the L-PBF sample ranged from 7 μm to 761 μm with an area-weighted average of 525 μm. One critical point to consider is the difference in the initial microstructure between the L-PBF and the as-cast samples. The as-built L-PBF samples revealed columnar grain morphology, while the as-cast sample exhibited equiaxed grains (see Figure 5d,f). In general, the samples with equiaxed grains showed higher flow stress and work hardening rates than those with columnar grains [39,40,41].

3.2.2. Ordering Effect on the Flow Behavior

Another point to consider is probably the retention of the room-temperature phase structure at the temperature of the imposed deformation. ThemoCalc computations suggested that the Fe-25Al-1.5Ta alloy remained in the (A2 + C14)-phase field region at the deformation temperature of 1000 °C (order–disorder transition of B2-to-A2 occurs at around 875–900 °C [8]). However, it is assumed that, at least at a very early deformation stage (the initial specimens were heated under argon atmosphere to an intended deformation temperature with a rate of 10 K/s and held for 3 min before compression testing), the alloy system can hardly equilibrate itself completely. The initial microstructure of the as-cast material at 1000 °C includes, at least partially, a mixture of (B2 + D0₃/L2₁) + C14 phases. In contrast, the L-PBF builds were characterized by a partially ordered matrix (A2 + B2) embedded with C14 Laves phase precipitates but without D0₃/L2₁ ordered regimes. The L2₁ precipitates coherent with the matrix imposed additional hardening in the cast sample. Dislocation slip within a disordered structure is much easier than in an ordered structure. Any cross slip in the ordered state requires the recombination of dissociated partials or their complete separation into single-fold bcc dislocations [42]. Both processes require additional energy and thus increase the flow stress for the as-cast material at the early stage of imposed deformation.

Another critical point is that defect diffusion slows down in ordered phases due to the decreased effective coordination number for nearest-neighbor defect jumps [43]. Order effects also impede processes based on self-diffusion; thus, the as-cast sample requires a higher hot deformation activation energy. Ordering also inhibits dislocation recovery to such an extent that dislocation rearrangement into subgrains does not occur. This contributes to the suppressed DRV process and higher flow stress in the as-cast alloys at the early stage of the imposed deformation.

3.2.3. Chemical Composition Effect on the Flow Behavior

Another point of discussion is the different chemical compositions of the cast and L-PBF samples. Both samples contained nearly the same amount of Ta; however, the as-cast sample included the comparatively higher Al content of 24.8 at. % than the L-PBF sample with 21.5 at. %. This may correspond to the vaporization of more volatile Al atoms during the L-PBF process due to the process-related high-energy input. The higher Al-content in the matrix of the as-cast material provides a remarkable solid solution hardening effect, which in turn, hinders dislocation annihilation and leads to dislocation pile-ups and higher flow stress.

The final remark is to consider the possible influence of a higher density of thermal vacancies retained in the L-PBF material as a result of rapid solidification. It was reported that the rapid solidification can quench in an extremely high density of thermal vacancies into Fe–Al alloys [44], which remarkably influences the strength and ductility of Fe–Al alloys from room temperature up to around 700 °C [45]. Nevertheless, a higher vacancy concentration promotes the annihilation of dislocations by recovery processes at higher temperatures, leading to the lower flow stress and activation energy of hot deformation in the L-PBF material.

4. Conclusions

In the present work, the microstructure and hot deformation behavior of laser powder bed fusion (L-PBF) Fe-25Al-1.5Ta (at. %) alloys were compared to the ones of the conventional cast material. The key conclusions of the current study are summarized as follows:

• The L-PBF builds were characterized by a phase mixture of A2 + B2 + C14. In contrast, the as-cast sample consisted of a phase mixture of B2 + D0₃/L2₁ + C14. Thus, the ordering transition from B2 to a D0₃ occurred during casting, but rarely during ultra-high-cooling L-PBF.
• The average activation energy (Q) of hot deformation was 344 kJ mol⁻¹ for the L-PBF build and 385 kJ mol⁻¹ for the cast material. A smaller value of Q for the L-PBF samples suggests a lower resistance to deformation. Therefore, these samples require a lower force to be deformed.
• The peak work hardening rate (θ) in the L-PBF sample (1.72 × 10³ MPa) was significantly smaller than that of the as-cast sample (3.02 × 10³ MPa), suggesting that the dislocation motions in the L-PBF sample were less pinned during deformation.
• The DRX initiated at comparatively lower stress in the L-PBF sample (19.75 MPa) than in the cast material (24.16 MPa) when deformation was carried out at 1000 °C/0.0013 s⁻¹.

In contrast to the casting process, L-PBF creates specimens that exhibit a lower work hardening rate and critical stress for the initiation of dynamic recrystallization. Furthermore, the L-PBF sample exhibits less deformation resistance and can therefore be formed with a lower deformation force. This, in turn, reduces the loads on the tools and dies, contributing to less wear and the high durability of dies.