Adapting the Mn content within a Fe-Mn-Si-based shape memory alloy by in situ parameter variations in laser-based additive manufacturing

Shape memory alloys (SMAs) are functional materials capable of recovering their original shape after inelastic deformation. This shape memory effect (SME) results from a reversible solid phase transformation and can be observed in multiple materials with different base elements including nickel-titanium (Ni-Ti), copper, and iron.¹ Combining these functional materials with additive manufacturing methods, such as laser powder bed fusion (PBF-LB/M), allows the production of functionally integrated components with a high degree of design freedom.

While the well-studied Ni-Ti alloys exhibit the most pronounced shape memory effect and are used as conventionally processed parts in aerospace, biomedicine, and automotive industries, additive manufacturing of these alloys is challenging. Due to their high sensitivity to compositional changes leading to large differences in the functional properties, the processing window for these alloys in PBF-LB/M is narrow.² As a result, it is difficult to manufacture complex geometries with homogenous functional properties using additive manufacturing.³ 

Iron-based SMAs, such as Fe-Mn-Si, offer an affordable alternative to Ni-Ti-based alloys. Next to lower costs, they also come with better weldability and machinability, making them suitable for a broader range of applications. Moreover, it was shown that iron-based SMAs can be processed using PBF-LB/M leading partly to even higher functional properties compared to conventionally processed alloys.^4–6 However, laser-based processing can lead to evaporation of elements such as manganese (Mn) during the manufacturing process.^7,8 Variations in the content of this austenite stabilizing element affect the functional properties, as the shape memory properties of iron-based SMAs result from a stress-induced transformation of γ austenite to ɛ martensite and its reversion upon unloading (superelasticity) or during heating (SME).^7,9 Hence, the influence of the processing parameters on the Mn content and the resulting functional properties has to be known to manufacture parts with homogenous properties. If this relationship is known, the laser parameters can be actively used to produce parts with varying Mn content and different shape memory effects. This was demonstrated by Ferreto et al.,⁷ where different laser scanning speeds were used to alter the energy input during the manufacturing process of a Fe-Mn17-Cr10-Si6-Ni4 alloy. By lowering the scanning speed from 400 to 100 mm/s, the Mn content was reduced to 10.7%, promoting the formation of δ ferrite in the as-built state and decreasing the shape memory effect. This effect can also be used to manufacture parts with functionally graded structures as shown by Ferretto et al.¹⁰ So far, this has been only investigated for a manganese content of <17%. In this range, the microstructure consists of γ austenite and δ ferrite, while the relative proportions of the two phases can be influenced by the Mn content. Here, a lower Mn content limits the amount of austenite available for stress-induced martensite transformation, thus affecting the SME. Gärtner et al.⁶ showed that the SME is also influenced by the Mn content in a range between 23% and 28% Mn despite the microstructure being fully austenitic. In this case, the lower Mn contents lead to increased shape recovery. This possibly results from a change in the chemical driving force for the stress-induced γ-ɛ transformation, as an increasing Mn content lowers the martensite start temperature.^11,12 Consequently, there must be a Mn content, which still leads to a fully austenitic microstructure while keeping the martensite start temperature close to the deformation temperature, resulting in a high chemical driving force for the stress-induced phase transformation. Thus, processing a Fe-Mn-Si-based alloy with higher starting Mn content and evaporating, depending on the local energy input, different amounts of Mn in the manufacturing process can provide a possibility to generate a multitude of different functional properties using the same base material. For this purpose, the energy input can be altered by changing the laser process parameters such as laser power and laser scan speed or by utilizing multiple exposures where the same powder layer is rescanned without intermediate recoating.¹⁰ In addition to the energy input itself, the interlayer time, i.e., the combined time needed for exposure and recoating, has to be considered as well.¹³ A shorter interlayer time increases the temperature of the components due to faster recurring energy input.¹⁴ As a result, for double-exposed samples, the pause set between exposures is an additional factor to be considered.

In the present study, the influence of Mn content on the microstructure and shape memory properties of a Fe-30Mn-6Si-5Cr alloy was analyzed through targeted evaporation of Mn in the PBF-LB/M process. This can, on the one hand, help to explain the mechanisms of how Mn influences the shape memory effect of iron-based SMAs and, on the other hand, improve the composition of this specific alloying system and provide a pathway for functional grading in additive manufacturing.

In this work, a PBF-LB/M-processed Fe-Mn-Si-based SMA with high manganese (Mn) content is studied. The powder was gas atomized by Rosswag GmbH (76237 Pfinztal, Germany) and had a particle size between 15 and 63 μm. The chemical composition of the powder was determined by MICA Analysen GmbH (40548 Düsseldorf, Germany) for Mn, for chromium (Cr) and nickel (Ni) using atomic absorption spectroscopy, for silicon (Si) using gravimetry, and for niobium (Nb) using photometry. The carbon content was determined using IR absorption. The resulting composition is listed in Table I.

All in all, three sets of samples were manufactured in which different parameters were changed to influence the energy input in the manufacturing process. In two sets, the scan speed was varied from 100 to 800 mm/s, leading to a change of E_V and E_L. One sample set (S1) of these was manufactured using single exposure and one using double exposure (S2). For the double exposure, the first and second scans utilized the same process parameters. The pause between the first and second exposure was the same as the recoating time.

For the third sample set (S3), the E_V was kept constant at 97 J/mm³ and the E_L was changed between 0.29 and 2.6 J/mm in order to clarify to what extent a possible Mn loss can be attributed to the applied volume or line energy. Constant volume energy density with varying line energy was realized by varying both the scan speed and the hatch distance while keeping the other parameters constant. All parameters for the three sample sets are stated in Table II.

The samples were separated from the building plates using electrical discharge machining with a molybdenum (Mo) wire.

The component undersides were sanded with 180-grid ZrO₂ paper for the following analysis of the Mn content. The analysis was carried out by wavelength-dispersive x-ray fluorescence in a Bruker S8 Tiger 4kW (Bruker Corporation, Billerica, MA, USA). The measurements were obtained in the “Best Detection” mode and processed through standardless QuantExpress (Bruker AXS) software, which uses the fundamental parameters method.

The relative density of the samples was determined using microscopic images. For this purpose, the samples were separated in the build direction, embedded in epoxy and finally ground and polished. Adaptive thresholding was applied to the images to easily identify the pores within the samples. The relative density was then determined by calculation of the ratio of the total number of pixels within the sample outline to the number of pixels accounting for the pores within the sample.

Based on the Mn contents determined, the single exposure samples with scanning speeds of 200, 300, 400, and 500 mm/s were selected for analysis using electron backscatter diffraction (EBSD). In addition to these single-exposure samples, double-exposure samples with scanning speeds of 300 and 400 mm/s were also selected for EBSD analysis. The acceleration voltage was 25 kV for all samples tested in a Philips XL 30 ESEM with an EBSD detector (Koninklijke Philips N.V., Amsterdam, Niederlande). These selected samples cover a Mn content of 21.4–28.0 wt. % and a volume energy density range of 116.6–291.6 J/mm³ following Eq. (1).

The laser parameters used for the samples selected for the EBSD analysis were also used to manufacture demonstrator parts to analyze the SME. For this purpose, demonstrator parts consisting of bending beams with a length of 15 mm, a height of 4 mm, and a thickness of 1 mm as seen in Fig. 1(a) were manufactured. For fast and reproducible deformation, a custom bending device was manufactured [see Fig. 1(b)]. Bending the samples against the device allows a reproducible bending deformation of 15° along a circular arc. After the deformation, the demonstrators were heated in a muffle furnace (L9/11, Nabertherm GmbH, 28865 Lilienthal, Germany) with a heating rate of 10 K/min up to 400 °C. For even temperature distribution, the temperature was kept constant for 30 min before cooling down to room temperature (furnace cooling).

For determination of the deformations, the demonstrators were three-dimensionally scanned in the as-built, deformed, and re-heated state using a GOM Scan 1 MV 400 in combination with the corresponding rotation table (Carl Zeiss GOM Metrology GmbH, 38122 Braunschweig, Germany). The SME of the demonstrators was determined as the percentage recovery angle after reheating.

The Mn concentration in mass percentage determined by the XRF analysis is shown in Fig. 2 for all specimens produced. The Mn content of the single-exposed samples decreases with increasing energy input down to 16.7% at a scan speed of 100 mm/s. This trend can also be observed for the double-exposed samples where the lowest achieved Mn content was 15.7% at a scan speed of 200 mm/s. The difference in the Mn content between the single- and double-exposed samples is particularly noticeable at higher energy levels. The samples manufactured with a constant volume energy density of 97 J/mm³ show a slight decrease in the Mn content over the range of line energies tested, but the Mn evaporation in the samples is greater when the volume and line energy are both increased. This indicates the importance of the volume energy density for Mn evaporation. The Mn contents of the single-exposed samples with scanning speeds of 200, 300, 400, and 500 mm/s, of which the microstructure was analyzed as well, are 21.4%, 24.1%, 27.3%, and 28.0%, respectively. The double-exposed samples with scanning speeds of 300 and 400 mm/s have 21.6% and 24.3% Mn, respectively.

A linear regression of the manganese content over the volume energy density is shown in Fig. 3. The evaporation of manganese correlates strongly with the volume energy density introduced. The volume energy density calculation [see Eq. (1)] does not consider double exposure. If, for the sake of comparison between single and double exposure, the volume energy density were to be doubled for the double exposures, the gradients of the curves depicted in Fig. 2 would both be −0.02. Accordingly, it can be assumed that utilizing double exposure has, in this case, a similar effect on the amount of Mn evaporation than doubling the volume energy density. Here, it should be noted that this statement is not universal as typically the energy absorption of a powder layer and an already melted layer is different possibly leading to different melt pool temperatures.¹⁵ Even in the present case, it only applies if the pause time between the two exposures equals the recoating time of the single-exposure process. A change in the pause time would likely result in differences in the part temperature and, thus, in the Mn evaporation.¹⁶ 

When comparing the present results with the study of Ferretto et al.⁷, as displayed in Fig. 3, it is noteworthy that the course of the Mn loss over the volume energy density appears to follow a different slope. They investigated an iron-based SMA with a lower initial Mn content of 17 wt. %, indicating that the Mn content of the alloy and its general composition play a role in the Mn loss.

A linear regression of the manganese content at a constant volume energy density of 97 J/mm³ over the range of line energies (0.3–2.6 J/mm) tested is depicted in Fig. 4. The result illustrates that line energy has a clear influence on the Mn content. It can be hypothesized that the impact of line energies on Mn evaporation diminishes as the volume energy density increases. As the volume energy density increases, the temperature of the component rises and, thus, the temperature difference that must be overcome during exposure in order to cause vaporization of Mn decreases.

Whether the samples can be manufactured without part failure using the set parameters restricts the testable field of volume and line energies. The usable energies were explored to their limits in these experiments, showing that greater Mn evaporation can be achieved for the alloy in question by varying the volume energy density. However, the influence of the line energy on the quality of the components and their structure should not be neglected as quality differences could be seen within the samples produced with constant volume energy density and varying line energy.

Light microscopic images of selected samples with visible differences in the relative densities are shown in Fig. 5. Figure 6 illustrates the relationship between the relative density and scanning speed of all evaluated samples. The data indicate that higher scanning speeds result in samples with higher density. The higher porosity results from overheating of the material during the selective melting process leading to a higher amount of Mn evaporation and keyhole pores.¹⁷ 

Figure 7 shows the EBSD phase images. δ ferrite is shown in green and γ austenite in red. It is evident that only the sample single-exposed at 200 mm/s and the sample double-exposed at 300 mm/s, which resulted in a Mn content of 21.4% and 21.6%, respectively, show δ ferrite areas. The two-phase microstructure results from a primary solidification as δ ferrite and a subsequent partial solid transformation to γ austenite, as seen in similar alloys during additive manufacturing.¹⁰ Averaged over two EBSD images, the ferrite content for the single-exposed sample at 200 mm/s is 13.55%, while for the double-exposed sample at 300 mm/s, it is 35.85%. The large difference in the δ ferrite content despite the similar Mn content might be related to different cooling rates depending on the scanning strategy. While during the first exposure, the melt pool is always at least partly surrounded by powder, it is always in contact with solid material with higher thermal conductivity during the second exposure. As a result, the cooling rate during the second scan of the double exposure is faster and a higher amount of the initial δ ferrite is maintained, explaining the higher amount of ferrite in the double-exposed sample.^10,18 Another possible explanation could be inhomogeneities in the elemental composition leading to lower local Mn contents in the double-exposed sample. It is already known that the processing of Mn-containing alloys using PBF-LB/M can lead to an inhomogeneous Mn content in the melt pool.⁷ Double exposure could enhance these inhomogeneities due to the different melting properties of the powder and the solid material leading to different melt pool sizes. The resulting differences in Mn content might affect the ferrite content as well. For a Mn content of 24.3% (double-exposed sample, 400 mm/s), a purely austenitic microstructure can be observed, which is also present in the single-exposed sample with a scanning speed of 300 mm/s and a Mn content of 24.1%. Consequently, the tipping point for ferrite formation is between 24.1% and 21.6% Mn for the tested Fe-Mn30-Si6-Cr5 alloy.

Next to the ferrite content, the grain size is affected by the different laser parameters as well (see Fig. 7). Across all parameters, the average grain size area was between 13.6 and 72.3 μm² with a clear trend of higher energy input leading to smaller grains. This is a phenomenon observed already earlier by Ferretto et al.¹⁰ and results from a shift of the primary solidification mode from γ austenite to δ ferrite with a lower Mn content. The primary solidification as δ ferrite and the subsequent solid phase transformation to austenite during cooling leads to a grain refinement, which is especially pronounced when utilizing double exposure.

Overall, the different laser parameters can be used to influence the microstructure of a high-Mn Fe-based SMA by evaporation of Mn and subsequent changes in the solidification and transformation behavior. In this context, higher energy input leads to grain refinement, a higher proportion of δ ferrite in the as-built state, but also an increased amount of keyhole pores.

The 3D scans of the bending beams in their as-built, deformed, and reheated state are shown in Fig. 8 and their Mn contents as well as their SME are listed in Table III. For the single-exposed parts, the SME increases with decreasing Mn content. This is in agreement with the study of Gärtner et al.,⁶ where the Mn content of a similar alloy was changed in the base powder. The poorer SME of the double-exposed sample, despite the lower Mn content, is connected to the ferrite content in the sample, as the SME is the result of a stress-induced transformation of γ austenite to ɛ martensite and its reversion.^7,9 Next to the ferrite content, the process parameters affect the grain size as well. In the literature, the influence of the grain size on the SME is debated and conflicting information is present.^7,19 In the present study, there is no direct correlation of grain size and SME visible, as the highest SME was achieved with an average grain size of 22.4 μm² and samples with larger and smaller grain size result in a worse SME. It appears that the effect of the grain size, if present, is overlaid by the effects of Mn and ferrite content. Overall, these results support the hypothesis that a lower Mn increases the SME as long as the microstructure remains fully austenitic.

Samples varying in volume and line energy were manufactured from a gas-atomized Fe-30Mn-6Si-5Cr alloy using single and double exposure during PBF-LB/M. Samples with varying line energy and constant volume energy density were manufactured to assess the evaporation of Mn attributable to the energies. Mn content analysis using standardless XRF and EBSD analysis for grain size analysis and phase determination was carried out. A demonstrator was used to visualize the shape memory effect of selected parameters. The following conclusions can be drawn from the results obtained:

• A decrease in the Mn content with increasing volume energy density can be achieved by varying the scanning speed. A reduction in the Mn content from 29.8% within the alloy to 15.7% for a manufactured sample was realized.

• Mn evaporation correlates strongly with the volume energy density introduced. At a constant volume energy of 97 J/mm³, an influence of the line energy on the Mn content was also found.

• Compared to Ferretto et al.,⁷ there is a differing Mn trend over volume energy density. This indicates the dependence of the Mn loss function upon the initial Mn content of the alloy.

• The formation of δ ferrite is observable in samples with a Mn content below 21.6%. Conversely, samples exhibit a fully γ austenitic phase when the Mn content exceeds 24.1%, indicating that the critical threshold for δ ferrite formation is situated between 21.6% and 24.1% Mn.

• The additional energy introduced via the double exposure can be accounted for by doubling the volume energy density. However, this is likely connected to the selected pause time and is not generally applicable.

• The functional properties of the material were demonstrated and the differences depending on the Mn content are evident.

The authors gratefully thank Marten Grube and Enno Schirmer for technical support during additive manufacturing. They also thank Daniel Hallmann, Martje Brandt, and Martina Rickers for their work in sample preparation and imaging and Anne Rahfeld for providing access to the XRF measurement facilities and her knowledge of the method. Thanks to Karl Siebels, Leonie Seelig, and Axel Blankenburg for their support in the metal workshop. This work was funded by the Leibniz Association in the form of the Leibniz Junior Research Group “AURORA—Additive manufacturing of graded structures from iron-based shape memory alloys” (Project No. J119/2021).

The authors have no conflicts to disclose.

Maylin Homfeldt: Conceptualization (equal); Data curation (equal); Formal analysis (lead); Investigation (lead); Methodology (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Jonas Schmidt: Conceptualization (equal); Formal analysis (supporting); Investigation (supporting); Methodology (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Anastasiya Toenjes: Conceptualization (equal); Funding acquisition (lead); Project administration (lead); Resources (lead); Supervision (lead); Visualization (supporting); Writing – original draft (supporting); Writing – review & editing (lead).

The data that support the findings of this study are available from the corresponding author upon reasonable request.