1. Introduction
Shape memory alloys exhibit superelasticity that recovers the original shape when external stress that caused deformation is removed or the shape memory effect that returns to the original shape in a certain temperature range. Most shape memory alloys exhibit the shape memory effect due to the reversible phase transformation between the parent phase and the stress-induced martensite phase. For the case of superelasticity, it exhibits thermoelastic transformation in which the reverse transformation from the stress-induced martensite phase caused by deformation to the parent phase is thermally activated upon the removal of stress because the austenite transformation starting temperature (As) is close to room temperature or low. The shape memory effect involves non-thermoelastic transformation in which reverse transformation begins when the deformed alloy is heated because the As temperature is high than room temperature. Significant applications of these shape memory properties are expected in civil engineering structures, such as prestressing and damping structures1-4). Alloys that can be used as shape memory alloys include Ni-Ti-based alloys as well as Cu-Zn and Cu-Al-based alloys that are based on Cu5,6). For Ni-Ti-based shape memory alloys, which have been widely used, many studies have been conducted due to their high shape memory effect, superelasticity, high corrosion resistance, and biocompatibility7), but their utilization is not proper for large civil engineering structures because of expensive raw materials and manufacturing process. Therefore, research has been conducted on alloys to replace them. Among them, iron- based shape memory alloys that are far less expensive than conventional shape memory alloys have the potential to serve as replacements.
Research on iron-based shape memory alloys has been conducted since Sato et al. discovered the significant shape memory effect from Fe-Mn-Si-based alloys in 19828-10). Fe-Mn-Si-based shape memory alloys exhibit the shape memory effect through the reversible phase transformation between the FCC-γ phase and the HCP-ε martensite phase. For these alloys, it is possible to obtain optimized properties by adjusting the transformation temperature, flow stress, and strain through the addition of various alloying elements (e.g., Mn, Co, and Ni)10,11). Fe-Mn-Si-based alloys, however, do not exhibit superelasticity as they are classified as non-thermoelastic martensitic transformation12-14).
In general, Fe-Mn-Al-based alloys also do not exhibit superelasticity. According to research by Omori et al., however, the addition of Ni causes reversible martensitic transformation between the BCC-α phase, a parent phase with a body-centered cubic structure, and the FCT-γ′ phase, a stress-induced martensite phase with a face-centered tetragonal structure, and it exhibits high superelasticity by inhibiting the dislocation generated during martensite transformation through the coherent precipitation of the nano-sized precipitates of the B2 phase in the A2 disordered matrix13-15). In addition, Fe-Mn-Ni-Al alloys exhibit very small entropy changes during martensite transformation due to the magnetic effect. Since the stress required to initiate transformation hardly changes despite the temperature change, these alloys can exhibit superelasticity in a stable manner over a very wide temperature range3,4). These properties contrast with conventional superelastic alloys for which the applicable temperature range is limited because the required stress increases alongside the increase in temperature16,17).
Additive manufacturing (AM) is a manufacturing technology that builds geometry by stacking materials in a layer-by-layer manner to implement a three-dimensional model designed with CAD. Conventional machining methods are unfavorable for the fabrication of complex geometry and exhibit low design freedom and significant material loss during processing. AM, on the other hand, has few geometry restrictions, minimizes the waste of materials, and can implement precise structures18,19). The AM process is particularly favorable for the fabrication of shape memory alloys that exhibit the shape memory effect and superelasticity. Since shape memory alloys are prone to defects during the casting process and their transformation temperature or superelastic properties significantly vary depending on their composition and microstructure, the metal AM process that can control thermal history and composition during the process is very effective20-22).
Recently, research on AM based on Fe-Mn-Al-Ni- based superelastic alloy powder has been increasing. Studies have reported the properties and microstructure characteristics of the alloys fabricated through various AM processes, such as laser powder bed fusion (L- PBF) and laser Direct Energy Deposition (L-DED) 23-27). The L-DED method performs deposition by supplying metal powder in real time and locally melting it using a high-energy-density laser. Since this process has a high deposition speed compared to PBF and enables deposition on substrates of various shapes, research has been actively conducted28-30).
According to a previous study, Fe-34Mn-14Al-7.5Ni alloys exhibit excellent mechanical properties because the nano-sized precipitates of the B2 phase are well precipitated13).
For AM of iron-based superelastic alloys, however, high-temperature solidification cracks (e.g., interlayer delamination) frequently occur due to solute segregation and residual stress at the boundary of the melt pool, making it difficult to fabricate specimens with no defect13,14). T0.9-0.99 represents the temperature difference of a section in which the fraction of the solid phase increases from 90% to 99%. According to Kou et al., the sensitivity to solidification cracking by the residual liquid phase increases as T0.9-0.99 increases because grains are not completely bonded in this section31). Therefore, this study aimed to reduce the solidification cracking sensitivity by fabricating specimens with Mn contents of 36 at.% and 42 at.% using the L-DED process and decreasing T0.9-0.99 to address the solidification cracking problem of iron-based super elastic alloys. This study verified the effectiveness of increasing the Mn content by investigating the effect of the Mn content on microstructure and super-superelastic properties before fabricating large specimens.
2. Materials
2.1 Materials
In this study, the powders of two compositions with different Mn contents and fixed Al and Ni contents were prepared by mixing pure metal powders, such as Fe (spherical, 99.9%), Mn (irregular, 99.9%), Al (spherical, 99.9%), and Ni (spherical, 99.9%) (EML, Republic of Korea). Each alloy was named as 36Mn and 42Mn using the Mn content. The mid-carbon steel plate S45C (0.45 wt.% C) was cut in a size of 100 × 50 × 10 mm3 to be used as the substrate. The chemical compositions of the two alloys deposited using the L-DED process are presented in Table 1. As a preliminary study, T0.9-0.99 was measured from Scheil calculation using Thermo-calc (Thermo-calc 2024b, Thermo- calc, Sweden), a commercial thermodynamics software program, and the TCFE 10 database. No fast diffusion element was set, and it was confirmed that T0.9-0.99 decreased as the Mn content increased as shown in Fig. 1.
2.2 Additive manufacturing
Fig. 2 shows the schematic of the L-DED process used in this study. Ytterbium Fiber Laser with a maximum laser power of 500 W was used in this study. The MX-LAB (InssTek Inc., Republic of Korea) equipment with a laser spot size of 0.3 mm was used, and argon gas (99.99%) was used for metal powder spraying and shielding. Cuboid specimens for superelasticity evaluation were prepared through AM in a size of 3 x 3 x 10 mm3. A laser power of 300 W, a scan speed of 840 mm/min, a hatch spacing of 0.3 mm, and a layer thickness of 0.15 mm were used as 15mm AM conditions. As for the deposition method, the cross deposition method that rotates the scan direction by 90° for each layer was used. For homogenization after deposition, heat treatment was performed at 1,200°C for six hours followed by air cooling. Aging treatment was then performed at 200°C for three hours for the precipitation of the β-NiAl phase, a nano-sized precipitate.
3. Method
3.1 Microstructure analysis
For microstructure analysis, scanning electron microscopy (SEM, JSM7200F, JEOL, Japan) equipped with energy dispersive spectroscopy (EDS, NanoAnalysis, Oxford, UK) and electron backscatter diffraction (EBSD, Nordlysnano, Oxford, UK) was used. The specimens were subjected to mechanical polishing using #400, #800, #1200, and #2000 polishing papers and then fine polishing using 3 μm and 1 μm diamond suspensions. Final polishing was performed through a 0.04 μm colloidal silica suspension.
3.2 Shape recovery performance (cyclic compression test)
Cyclic compression tests at room temperature were conducted in the 10mm longitudinal direction (deposition direction) through a universal testing machine (QUASAR50, Galdabini, Italy). The strain rate was set to 3×10-4 s-1, and loading-unloading data were obtained repeatedly in real time while the displacement was increased by 1% from 1% to 5%. The inflection points in the loading-unloading curve represent the critical transformation stress (σt) and critical slip stress (σs). The intersection of the tangents at both ends of the inflection points was regarded as the critical stress. In addition, the superelastic displacement (ds) was represented by the difference between the maximum displacement and the residual displacement (dr) for each cycle as shown in Fig. 3.
4. Result and Discussion
Fig. 4 shows the microstructures of the 36Mn and 42Mn alloys in the as-built state. Even though EBSD analysis was conducted in the same specimen for both alloys, there were significant differences in phase fraction and microstructure between the top and bottom areas of the specimen. For the 36Mn alloy, the bottom area of the specimen near the substrate exhibited a dual-phase structure in which the FCC-γ and BCC-δ phases were mixed (Figs. 4 (a and c)) while the top area showed the BCC-δ grain single-phase structure elongated in the axial or deposition direction. In the case of the 42Mn alloy, however, the FCC-γ phase was mostly observed in the bottom area while the top area of the specimen exhibited a dual-phase structure composed of BCC-δ and FCC-γ phases (Figs. 4 (b and d)). This appears to be due to the volatilization of Mn caused by the heat accumulated in the specimen during the fabrication of the specimen using the DED process. Mn is a representative FCC stabilization element, and it is known that a considerable amount of Mn volatilizes during AM that uses laser due to its high saturated vapor pressure32). Therefore, it appears that the FCC-γ phase was mainly detected from the bottom area of both specimens with relatively low potential for Mn volatilization while the fraction of the BCC phase increased in the top area of the specimens due to the reduction in the phase stability of FCC-γ caused by Mn volatilization.
Fig. 4
EBSD analysis results of 36Mn and 42Mn alloys. (a and c) 36Mn alloy, (b and d) 42Mn alloy. (a-1, b-1, c-1, and d-1) IPF map along building direction and (a-2, b-2, c-2, and d-2) phase map. (a and b) bottom area near the substrate and (c and d) area near the top surface
Meanwhile, it appears that the difference in phase fraction between the 36Mn and 42Mn alloys was caused by the difference in Mn content, and that the high Mn content of the 42Mn alloy caused the high fraction of the FCC-γ phase in both the top and bottom areas of the specimen compared to the 36Mn alloy. According to Fig. 1 that shows the solid phase according to the temperature, however, the supernormal phase was the BCC-δ phase for both alloys and no FCC phase was observed until solidification was complete. To identify the cause of this discrepancy, the phase fraction according to the temperature was calculated at the target compositions of the two alloys using the Thermo- calc software and the results are shown in Fig. 5. According to Figs. 5 (a and b), the BCC-δ phase was formed at high temperature for both alloys and the FCC-γ phase was formed as the temperature decreased. As can be seen from Fig. 5(c), the transformation start temperature of 36Mn (955ºC) was lower than that of 42Mn (960ºC), and the FCC fraction of the 36Mn alloy at 500ºC was also lower than that of 46Mn. Although the calculated difference in the fraction of the FCC-γ phase between the 36Mn and 42Mn alloys is not large, the driving force for the phase transformation from the BCC-δ phase, a supernormal phase, to the FCC-γ phase is expected to be higher for 42Mn. Meanwhile, Kim et al. reported that the BCC phase, a supernormal phase, can be transformed into FCC, a low-temperature phase, due to the inherent heat treatment (IHT) effect in which the heat generated during the deposition of a new layer affects the previously deposited layers in the AM of Fe-Mn-Si-based shape memory alloys using L-PBF33). In this study, it also appears that a significantly higher fraction of BCC-δ, a supernormal phase, was transformed into FCC-γ due to considerable IHT during the AM of the 42Mn alloy, which is expected to exhibit high driving force for the BCC-δ → FCC-γ phase transformation. For the accurate analysis of the cause, however, further research is required, including the analysis of free energy changes between the two phases and real-time phase transformation behavior.
Fig. 5
Equilibrium phase fraction as a function of temperature. (a) 36Mn alloy, (b) 42Mn alloy, and (c) FCC-γ phases fraction
Fig. 6 shows the microstructures of the 36Mn and 42Mn alloy specimens subjected to air cooling after heat treatment at 1,200°C for six hours. As can be seen from Figs. 6 (a and c), there was no significant change in grain structure after heat treatment. Meanwhile, according to Figs. 6 (b and d), both specimens had a BCC single-phase structure with no residual FCC phase after heat treatment. This is in good agreement with the result that both the 36Mn and 42Mn alloys exhibited a single phase of BCC-α at 1,200ºC as can be seen from Fig. 5, and both alloys are expected to exhibit higher superelastic properties than the as-built phase because Fe-Mn-Al-Ni-based superelastic alloys develop superelastic properties due to the reversible phase transformation between BCC-α, a parent phase, and the FCT-γ′ phase, a stress-induced martensite phase. For the precipitation of B2-NiAl that significantly affects the superelastic properties of alloys, both specimens were subjected to aging treatment at 200°C for three hours and then cyclic compression tests.
Fig. 6
IPF map (a and c) and phase map (b and d) of 36Mn and 42Mn alloy after heat treatment at 1200°C for 6 h. (a and b) 36Mn, (c and d) 42Mn
Fig. 7 shows the cyclic compression test results for the 36Mn and 42Mn alloys. The critical transformation stress (σt), critical slip stress (σs), superelastic displacement (ds), and residual displacement (dr) for each cycle were listed in Table 2.
Fig. 7
Stress-deformation curves of 36Mn and 42Mn alloys from cyclic compression test. (a) Overall, (b)1st cycle, (c) 2nd cycle, (d) 3rd cycle, (e) 4th cycle, and (f) 5th cycle
Table 2
Critical stresses for phase transformation and slip for each cycle
As shown in Fig. 7 (b), there was no significant difference in compressive behavior between the 36Mn and 42Mn alloys and they exhibited almost identical elastic behavior during the first compression. From the second cycle, there was a difference between the curves of the two specimens (Fig. 7 (c)). According to the literature, the primary inflection point on the loading-unloading curve is the critical stress of phase transformation34). In the case of the 42Mn alloy, phase transformation began at a relatively lower stress compared to the 36Mn alloy, and the maximum stress at a maximum displacement of 2% was also lower. For the third cycle (Fig. 7 (d)), the difference in compressive behavior became more obvious and the maximum stress at a maximum displacement of 3% differed by approximately 80 MPa between the 36Mn alloy (495 MPa) and the 42Mn alloy (412 MPa). Since there is no significant difference in σc between the two alloys, the difference in maximum stress appears to be associated with the fraction of the FCT-γ’ phase. As can be seen from Figs. 4 (b, d) and 5, the phase transformation from BCC-α to FCT-γ’ was more active for 42Mn compared to the 36Mn alloy due to the high Mn content, thereby increasing the fraction of the FCT-γ’ phase, which has lower yield strength than the BCC-α phase. ds also decreased by approximately 0.16% from 2.662% of the 36Mn alloy to 2.503% of 42Mn. This also seems to be because the stability of the FCT-γ’ phase increased due to the high Mn content of the 42Mn alloy and the fraction of the FCT-γ’ phase subjected to reverse transformation decreased compared to the 36Mn alloy.
In the case of the 42Mn alloy, an additional deflection point was observed from the fourth cycle (Fig. 7 (e)). The difference in maximum load between the two alloys at a maximum displacement of 4% increased to 186 MPa, and the difference in ds also increased to approximately 0.4%. This appears to be due to the second inflection point at 436.8 MPa, which occurred during the compressive deformation of the 42Mn alloy. Fe-Mn-Al-Ni-based superelastic alloys develop superelastic properties due to the phase transformation from the BCC-α phase, a parent phase, to FCT-γ′, a stress-induced martensite phase, and the reverse transformation. In the case of the 42Mn alloy, it appears that the residual displacement dr significantly increased because the yield phenomenon by dislocation slip at 436.8 MPa (approximately 3.15% displacement) allowed some of the deformation to exhibit superelasticity by being accommodated by phase transformation, but the rest could not exhibit superelastic property by reverse transformation upon the removal of the load due to the permanent deformation caused by dislocation slip. The difference in maximum stress at the maximum displacement also appears to have resulted from the reduction in the slope of the stress-displacement curve caused by the yield phenomenon in the 42Mn alloy.
For the fifth cycle shown in Fig. 7(f), yielding was also observed at approximately 749.9 MPa from the 36Mn alloy and a significant increase in residual displacement (dr) was confirmed. This supports the fact that the second inflection point for both alloys is the critical stress point of slip that causes permanent deformation. Meanwhile, it was confirmed that the residual displacement (dr) increased alongside the increase in compressive displacement even before the occurrence of the yield phenomenon despite being smaller in quantity compared to after the occurrence of the yield phenomenon. This is believed to be because dislocation slip that causes permanent deformation does not occur only above a certain critical stress, but it gradually increases alongside the increase in stress and compressive deformation and explosively increases above the critical stress (σs) to become the dominant deformation mechanism.
5. Conclusion
In this study, alloys with high Mn contents were fabricated by mixing elemental powders and their microstructures and superelastic properties in the as-built state were investigated to improve the buildability of Fe-Mn-Al-Ni alloys vulnerable to solidification cracking. The effect of the Mn content on the microstructure formation of alloys in the as-built state could be verified, and a significant reduction in superelastic property alongside the increase in Mn content could be confirmed. Based on this, it was possible to identify factors to be considered additionally during the design of Fe-Mn-Al-Ni alloys optimized for additive manufacturing (AM) in the future. The conclusions and key findings of this study are summarized as follows.
1) For the Fe-Mn-Al-Ni alloys deposited using the laser Direct Energy Deposition (L-DED) process, the fraction of the FCC-γ phase in the as-built state increased as the Mn content increased. This appears to be because FCC, a low-temperature phase, was transformed from the BCC phase, a supernormal phase, due to the inherent heat treatment (IHT) effect. Both the 36Mn and 42Mn alloys exhibited the BCC phase when air-cooled after heat treatment at 1,200°C for six hours.
2) In the case of the 42Mn alloy with a high Mn content, the stability of the FCT-γ’ phase increased due to the high Mn content and thus the amount subjected to reverse transformation into BCC-α decreased, thereby exhibiting lower superelastic displacement (ds) compared to 36Mn under the same compressive deformation.
3) The critical stress for slip that causes permanent deformation (σs) was also lower for the 42Mn alloy compared to 36Mn and thus permanent deformation began at lower stress, thereby decreasing the superelastic displacement (ds) and increasing the residual displacement (dr).
The above results confirmed that the addition of Mn increased the FCC phase in the as-built state but decreased mechanical and superelastic properties. There- fore, we will conduct further research on methods to promote reverse transformation by improving the stability of the BCC-α phase of the equilibrium state while maintaining low T0.99-0.9 values and high FCC fractions during solidification through further alloy design and an in-depth analysis of solidification behavior that involves non-equilibrium states.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
