Preprint for "Effect of process parameters on thermomechanical response and morphology of nitinol produced by laser powder bed fusion"

Effect of process parameters on thermomechanical response and morphology of nitinol produced by laser powder bed fusion Ondřej Červinek1*, Jakub Hurník1, Miroslav Šmíd2, Ondřej Zobač2, Melanie Todt3 and Daniel Koutný1 1 Institute of Machine and Industrial Design, Brno University of Technology, Technicka 2896/2, 61669 Brno, Czech Republic (*[email protected]) 2 Institute of Physics of Materials, Czech Academy of Sciences, Zizkova 513/11, 61600, Brno, Czech Republic 3 Institute of Lightweight Design and Structural Biomechanics, Technical University of Vienna, Gumpendorfer Straße 7, A-1060 Vienna, Austria Abstract This study represents a comprehensive investigation of the relationship between the process parameters of laser powder bed fusion (L-PBF) technology and the specific properties of nitinol. Based on a preliminary study carried out with variable laser power (40-400 W) and scanning speed (175-3000 mm·s-1), the most promising parameters were selected and used to produce bulk specimens as well as thin walls in different thicknesses between 0.5 and 2 mm. The porosity, dimensional accuracy, austenitemartensite phase transformation and mechanical properties were investigated. The results showed that for a volumetric energy density (VED) of 41-55 J·mm-3, a porosity of less than 0.3 % can be achieved, although after manufacturing brittle cracking may occur. For the VED range up to 60 J·mm-3, no significant influence of the Ni content on the transformation temperatures was found. For the walls made of L-PBF, the porosity tended to increase with decreasing nominal thickness, but the thickness deviations decreased. The phase transformation showed one exothermic and one endothermic peak with an austenite finish below room temperature for all tested configurations, resulting in superelastic behavior without noticeable temperature induced shape memory. Cyclic compression led to variable quasilinear pseudoelasticity with low hysteresis. The highest total strain after 50 cycles was 5.46 % with an associated accumulated residual strain of 2.79 %. The deterioration in recovery strain was greatest in the first few cycles and stabilized on average at 30th cycle. Keywords Nitinol, superelasticity, laser powder bed fusion, volumetric energy density, differential scanning calorimetry, cyclic loading 1. Introduction Nickel-titanium alloys become of great interest for high-end biomedical and aerospace applications due to their unique ability to undergo large deformations (up to 8∼10 % [1,2]) and return to their original, undeformed shape. Recoverability is achieved either by the application of heat, which we refer to as shape memory effect (SME), or by the release of stress, which we refer to as superelastic effect (SE). SME/SE are made possible by a reversible phase transformation between the austenitic phase (A) and the martensitic phase (M) [3,4]. Both effects allow a temporary transformation of the component shape produced from this material, creating an artificial actuator that can be triggered by external stimuli. The latest development takes advantage of the properties of nitinol and targets internally structured single-component actuators as a highly reliable replacement for complex mechanisms [5] such as those found in leading and trailing edges of airplane wings. To use them in real applications, a large actuation range had to be achieved in a relatively stiff component without damaging it, which was considered impossible until recently. However, the progressive development of additive manufacturing processes, such as laser powder bed fusion (L-PBF) has enabled the direct fabrication of nitinol metamaterials that allow the geometry and thermomechanical behavior [6] to be tailored to the desired actuation. Topology optimization methods can be used to optimize the internal geometry while the thermomechanical response of the parent material can be controlled by the parameters of the manufacturing process and posttreatment. Apart from the thermomechanical behavior, the parameters of the manufacturing process play an important role in the formation of porosity, microstructure formation and phase transformation temperatures [7,8]. Previous studies have shown that even different combinations of laser power and scanning speed with the same energy density delivered to the molten material change the material performance [9]. The influence of laser power on the nickel content in the molten pool is particularly pronounced due to the risk of evaporation of nickel at low temperatures [10], which can be avoided by a higher scanning speed, which reduces the energy input. In addition, the quality of surface forming of nitinol produced with L-PBF is significantly influenced by the scanning speed and the laser power ratio [8]. A ratio of more than 0.3 or less than 0.1 results in a considerable number of spherical pores [11]. In contrast, L-PBF fabrication of thin-walled nitinol parts has shown that a lower scanning speed is preferable for parts with small struts (0.2 mm-0.6 mm) to achieve higher geometric accuracy [12]. Post-processing, such as heat treatment, also has a major influence on the nitinol properties. The precipitation characteristics are highly dependent on the aging temperature, aging time, cooling rate and alloying [13]. Similar research indicates that the actuating performance of a beam actuator made of a nitinol shape memory alloy (SMA) improves with increasing heat treatment temperature [5]. For nitinol produced with L-PBF, it was shown that the specimen aged at 350°C and 18 hours exhibited near ideal superelasticity with 5.5% strain recovery and only 0.3% unrecoverable strain at the first cycle under a load of 1000 MPa [13]. Aging heat treatment leads to precipitation that is not related to the matrix phase. As a result, higher phase transformation temperatures can be achieved than with conventionally produced nitinol, which forms coherent precipitates after aging [14]. This affects the response to changes in ambient temperature [15] which is crucial in real applications. In general, L-PBF processed nitinol specimens consist mainly of austenite and a small amount of martensite at room temperature [16]. The phase transformation temperatures of L-PBF processed nitinol are higher than those of nitinol powder due to nickel evaporation during the L-PBF process [6,13]. With regard to applications in metamaterials, Iasnii et al. [17] have shown that the termination temperature of austenitic transformation in nitinol bulk rods is much lower compared to those of thin struts in lattice structures. In addition, Zhang et al. [18] found that the elastic modulus deteriorated and the residual strain increased when the applied maximum strain increased. Particularly affected are starting temperature of the forward martensitic transformation Ms under compressive stress and termination temperature of the martensitic transformation Af under recovery stress. Load history is another important feature that must be considered when developing actuators, as the recoverability decreases with the number of load cycles [9,19]. According to Saedi et al. [13], the superelastic behavior of nitinol specimens fabricated with L-PBF can be considered stabilized after the tenth cycle. In contrast, Henderson et al. [15] found stabilization up to the 40th cycle. According to Chen et al. [20], nitinol gradient lattice structures achieved a recoverable displacement rate of at least 99.15% after six cycles. This resulted in performance inconsistency in nitinols fabricated by LPBF with different process parameters. Further investigation of cyclic tensile loading with increasing loads and conventional stress-controlled fatigue showed that the higher proportion of untransformed martensite reduces fatigue life [21]. In addition, an increase in residual strain and a decrease in the elastic modulus can be observed when the maximum compressive strain increases under cyclic loading [18]. Regardless of the application, efficient development of internally structured actuators with tailored thermomechanical behavior requires a deep understanding of the relationship between the phenomenological behavior of nitinol and the parameters of the production process. This relationship can be interpreted based on material and geometry characteristics determined for a specific thin-walled production setup. Unfortunately, not much attention has been paid to the complex consideration of the relationship between the L-PBF manufacturing parameters and the resulting thermomechanical performance in relation to thin-walled components. This study addresses the lack of knowledge about the relationship between process parameters and the thermomechanical behavior of nitinol. Based on a preliminary study, potentially effective process parameters are identified and further investigated with respect to thermal, mechanical and geometrical aspects. The study provides an experimentally based, production-specific description of nitinol behavior and forms a solid basis for the development of internally structured actuators. 2. Materials and methods The aim of this study was to obtain an experimental description of the fabricationspecific thermomechanical behavior of a nitinol alloy produced using L-PBF technology. The following section describes the main analyzes carried out from the preliminary study towards the detection of the austenite-martensite phase transformation limit. 2.1 Powder characteristics Gas atomized nitinol metal powder (Carpenter Additive, Widnes, United Kingdom) was chosen for the manufacturing of the specimen. The shape of the particles was analyzed using a scanning electron microscope (SEM, Table 1). The particle size distribution was determined by laser size diffraction using the ASTM B822 standard. The tap density of the powder was determined to be 4.26 kg·m-3 using the ASTM B527 standard. The chemical powder composition of the nitinol alloy is given in Table 2. Table 1 – Size distribution of nitinol powder; nitinol powder morphology – SEM image Quantile Particle size [μm] Q10 13.3 Q50 30 Q90 51.2 Table 2 – Chemical composition of nitinol powder (*Other elements in total) Ni C Cr Co Cu H wt. % 55.75 0.0024 0.0042 <0.001 <0.0035 <0.0005 Fe Nb N O OET* Ti wt. % 0.0044 <0.001 <0.0005 0.058 <0.4 Bal. 2.2 Production of nitinol specimens The specimens were produced using an SLM 280HL machine (SLM Solutions Group AG, Lübeck, Germany). An ytterbium Gaussian-mode fiber laser YLR-400-WC-Y11 (IPG Photonics, Oxford, USA) with 400 W maximum power, a spot diameter of 82 µm and a wavelength of 1070 nm was used. A reduced 100 x 100 mm nitinol base plate and a standard silicon recoater blade were used to produce specimens. Table 3 gives an overview of the initial process parameters used. Table 3 – The L-PBF machine process parameters Laser Power P [W] Borders 40-400 Fill Contours 150 Hatching 40-400 Scanning speed V [mm·s-1] Borders 175-3000 Fill Contours 450 Hatching 175-3000 Hatch distance H [µm] 75-125 Layer thickness L [µm] 50 Scanning strategy [-] Bidirectional hatching (stripes) Interlayer orientation [°] 67 Platform heating [°C] 200 Atmosphere [-] Ar Residual oxygen content [%] <0.2 The following equation was used to determine the volumetric thermal energy density (VED) delivered to the powder layer during the manufacturing process: 𝑉𝐸𝐷 = 𝑃 𝑉 · 𝐻 · 𝐿 [𝐽 ∙ 𝑚𝑚−3] (2-1) where P is the laser power, V the laser speed, H the hatch spacing and L the layer thickness. Table 4 shows the types and geometric parameters of the test specimens produced in each series. The cylinders were fabricated to represent the uniaxial stress-strain state in the bulk nitinol material, while the walls were fabricated to represent the geometric features of a thin element, similar to those found in planar lattice structures. Table 4 – Types and geometric parameters of the test specimens Height [mm] Diameter/Thickness [mm] Number [-] Bulk cylinder 10 8 9 Thin wall 0.5;0.75;1;1.25;1.5;2 18 2.3 Material composition analysis A combination of SEM and energy dispersive X-ray spectroscopy (EDS) was used to compare the changes in material composition, in particular the amount of nickel evaporated during the L-PBF process. The SEM examination was performed with the Zeiss Ultra-Plus 50 (Carl Zeiss AG, Oberkochen, Germany). Chemical composition analysis was carried out using the Oxford EDS detector (Oxford Instruments, High Wycombe, United Kingdom). 2.4 Transformation temperatures The transformation temperatures of nitinol produced from L-PBF were determined using differential scanning calorimetry with heat flow detector (DSC HF). For this purpose, specimens with the dimensions 4 × 4 × 1 mm were cut from the bulk material cylinders using the electrical wire discharge cutting (EDWC) method. The NETZSCH DSC 204 F1 Phoenix device (Erich Netzsch GmbH & Co. Holding KG, Selb, Germany) with intercooler was used for the DSC analysis. Each specimen was placed in the aluminum crucible with lid and subjected to three temperature cycles ranging from 75 °C to -75°C and back at a constant heating and cooling rate of 10 K·min-1. The test was performed in an inert atmosphere with a flow rate of 70 ml·min-1 of pure 6N argon, as its lower thermal conductivity increases the sensitivity in the low temperature range. An empty aluminum crucible with lid served as reference material. The temperature was calibrated to the melting temperature of five pure substances. The NETZSCH Proteus® program was used to evaluate the measured DSC data. 2.5 Mechanical properties The mechanical properties were determined using a cyclic compression test on a Shimadzu AGX-V2 device (Shimadzu, Kjóto, Japan). In the first phase, the cylindrical specimens described in Section 2.2 were compressed vertically until engineering stress of 800 MPa was reached and then unloaded (the L-PBF fabricated nitinol should not reach the plastic yield [6,14]), as shown in Table 5. In this condition, 50 loading cycles were performed at a load rate of approximately 10-3 s-1. In the second stage, an additional 25 cycles were made with stress level increased by five load cycles to 1200 MPa. The measurement was carried out at room temperature, assuming superelastic behavior with no signs of thermally induced austenite-martensite phase transformation. The deformation of the specimens was recorded with an optical extensometer (Dantec Dynamics, Skovlunde, Denmark), which is based on the principle of digital 3D image correlation (DIC) and works with two 5 MPx cameras. The setup covered a measurement area of approx. 25 × 25 mm (base length 97.6 mm, stereo angle 27.26°). To cover the range, the system used 50 mm lenses with 10 mm intermediate rings. A customized imaging technique was used to reduce the acquired data. The correlation parameters are listed in Table 5. Table 5 – DIC correlation parameters; applied test parameters with time steps Searching radius 148 Px Facet size 31 Grid spacing 17 Px Maximum 3D residuum 0.6 Px 2.6 Microscope image analysis The Keyence VHX-2500 digital microscope (Keyence, Osaka, Japan) with the Z250R objective (x250 zoom) was used to obtain detailed information on the cross-sectional geometry and porosity of the specimens. The microscope images of the specimen cross-sections were analyzed using the plugin of the Java Script based open-source software ImageJ2 Fiji (National Institutes of Health, Bethesda, Maryland, USA). Images were converted to greyscale with a range of 0-255 (0-white, 255-black) to detect pores and borders. Then the threshold filter was applied to select a range of 110-255. In this way, the images were transferred into binary maps in which the porosity and dimensions of specimens were recognized based on binary criteria. 3. Results and discussion 3.1 Preliminary study In the first step, the process parameter window defined in section 2.2 was tested experimentally in the form of single tracks. The track properties were acquired as discrete results of 90 microscopic measurements of track width W, height H, depth D and contact angle α (Appendix Figure 7). As expected, a decrease in scanning speed V and an increase in laser power P leads to an increase in track width, depth and in some cases contact angle, while the height of the track decreases. A similar trend with close match at intermediate scanning speeds was also observed in the study by Bourke et al. [22] in a narrower process parameter window. A regression analysis using a quadratic polynomial function was performed to determine the relationship between the input process parameters and the resulting track properties, similar to Vaglio et al. [23]. Eq. 3-1 represents the trend of approximated track properties (Z with Z ∈ {W,H,D,α}) as a function of change in scan velocity (V) in mm·s-1 and laser power (P) in W. In addition, a continuous prediction of track properties was made based on possible combinations of input parameters. The polynomial parameters of the regression function (β0… β5) are given in the Appendix, Table 7. From this it can be concluded that a better approximation and thus a potential trend prediction was achieved for the track width and depth due to the relatively low mean square error. The poorer approximation of the track height can be attributed to the wide window of process parameters, which in extreme cases lead to considerable melt pool pulsations and a balling effect, which is reflected in a greater variation of the track height. Large variations in the measured height led to a deviation from the plane that approximates the data set, even when a second-order polynomial function is used. Based on the observed track continuity, uniformity, absence of cracks and the dimensions mentioned above, 12 groups of process parameters were selected for further investigation of bulk specimens (Appendix Figure 8). 𝑍 = 𝛽0+ 𝛽1𝑉 + 𝛽2𝑃 + 𝛽3𝑉2+ 𝛽4𝑃2+ 𝛽5𝑉𝑃 (3-1) 3.2 Bulk material analysis The analysis of the internal porosity of the bulk specimens showed low values overall (Figure 1 (a)). It increased to a maximum value of 0.37 % for V at an intermediate value of 1000 mm·s-1, P = 400 W and a high value of H = 125 µm, resulting in a peak VED of 64 J·mm-3. When the filter was applied to the sum of pores with an area of at least 10 µm2, a significantly lower maximum porosity of 0.24 % can be observed. In contrast, the lowest unfiltered and filtered porosity of less than 0.1 % each was achieved when H was reduced to 105 µm and P to 280 W, resulting in a VED of 53.3 J·mm-3. The comparison of the maximum and minimum porosity resulted in the same values for the specimens regardless of filtering. The more comprehensive comparison with review literature [24,25], in which different process parameters and Ni contents were taken into account, resulted in a higher relative density of the material. The only comparable density was obtained with a VED of 41 J·mm-3 [26] with a similar Ni content and similar parameters. In general, a very high relative density for VED in the range of 41-55 J·mm3 was achieved in [26] and in the present study, with a porosity of less than 0.3 %. However, it is worth noting that the layer thickness used in the source study was only 30 µm. Unfortunately, some of the specimens with the highest relative density were prone to macroscopic cracking after fabrication (Figure 1 (a) detail), which is why the process parameters combinations used for their fabrication were not investigated further. Macroscopic cracking of cylindrical specimens has also been described in the literature for process parameters that resulted in approximately equal VED values [25]. (a) (b) Figure 1 – (a) Comparison of the porosity of bulk specimens; (b) change in atomic content as a function of the L-PBF process parameters The EDS measurement results showed how the concentration of individual chemical elements changes when the material is subjected to the laser fusion process. Due to the very low concentration, some of the chemical elements were excluded from the comparison as their occurrence was negligible. Figure 1 (b) compares the average measured atomic content of nickel in the alloy processed with L-PBF with the concentration in the initial state. Due to the lower melting point temperature of nickel, its content in nitinol can decrease during the manufacturing process [27]. As a result, nickel evaporation can lead to a shift in transformation temperatures. However, despite the small variations in nickel content within the observed range, no significant decrease was observed with increasing VED. This is confirmed by the standard deviation in the graph, which overlaps with the nickel content in the premanufacturing state. According to Zhan et al. [28], a significant decrease in nickel content would require a VED value of over 65 J·mm-3. The authors of the study used a VED value of about 87 J·mm-3 with remelting to adjust the transformation temperatures and achieve adequate performance of mechanical properties. The observed increase in Ni content at a VED of about 45 J·mm-3 could be explained by the proximity to the detection limit of the EDS analyses [29]. 3.3 Thin walls analysis Of the 12 remaining groups of process parameters, the 3 most promising (Table 6), characterized by low porosity and the absence of cracks in the manufactured specimens, were selected for further investigation in the form of thin-walls with a nominal thickness of 0.5-2 mm. Figure 2 shows an example of the comparison of the nominal dimensions of the thin walls with the measured data. A comparison of specimens from different groups with different nominal thicknesses is shown. Figure 2 (a) shows a 0.5 mm thick specimen with a measured mean thickness below the nominal value, while Figure 2 (b) shows a 2 mm thick specimen with a mean thickness above the nominal value. The range of the measured data in Figure 2 (b), expressed by three times the standard deviation, is significantly larger than that of the data in Figure 2 (a), indicating that the variation in manufacturing dimensions increases when higher nominal thicknesses are used. Table 6 – Process parameters for bulk material cylinders and thin walls Group A B C VED [J·mm-3] 43 60 50 Laser Power P [W] Borders 200 280 360 Fill Contours 150 150 150 Hatching 200 280 360 Scanning speed V [mm·s-1] Borders 1250 750 1250 Fill Contours 450 450 450 Hatching 1250 750 1250 Hatch distance H [µm] 90 125 115 (a) (b) Figure 2 – Dimensional analysis of the thin wall of (a) group A with a nominal thickness of 0.5 mm; (b) group C with a nominal thickness of 2 mm Figure 3 shows the overall comparison of the mean measured thicknesses compared to the nominal values. It can be clearly seen that the measured specimens tend to be thinner at a nominal value of 0.5 mm, regardless of the group. The lowest measured thickness is observed for group B with a mean value of 465 µm. With this exception, the measured thickness was greater than the nominal thickness. The largest thickness deviation was measured at 2 mm nominal thickness with a maximum value of 2148 µm for group C (Figure 2 (b)). On the other hand, the smallest thickness deviations were found for group A across the entire range of nominal thicknesses tested. It is worth noting that the threefold standard deviation of the measured thickness does not overlap with the nominal thickness at 1.5 mm and 2 mm. Surprisingly, thicker walls lead to larger deviations, but these are more uniform in thickness variation. The mechanism responsible for the different wall thickness deviations in the different groups could be attributed to the different energy density supplied to the melt pool [30], as well as the lower heat dissipation path for thinner walls [31]. A higher temperature of the melt pools can be expected with them, which leads to a lower number of adhering particles. As a result, thicker walls tend to have greater thickness due to the larger number of adhering particles. Another parameter that played a key role here was the scanning strategy, which influences the mutual position of the contours and the scan vectors used for the core hatching [32,33]. In addition, as a result of the sliced geometry, some of the very short vectors could be omitted when the laser passed the scan path, especially in the case of very thin walls. [9] Ren Q, Chen C, Lu Z, Wang X, Lu H, Yin S, et al. Effect of a constant laser energy density on the evolution of microstructure and mechanical properties of NiTi shape memory alloy fabricated by laser powder bed fusion. Opt Laser Technol 2022;152:108182. https://doi.org/10.1016/j.optlastec.2022.108182. [10] Li G, Yu T, Wu P, Chen M. Molecular Dynamics Simulation of NiTi Shape Memory Alloys Produced by Laser Powder Bed Fusion: Laser Parameters on Phase Transformation Behavior. Materials (Basel) 2023;16. https://doi.org/10.3390/ma16010409. [11] Dadbakhsh S, Vrancken B, Kruth JP, Luyten J, Van Humbeeck J. Texture and anisotropy in selective laser melting of NiTi alloy. Mater Sci Eng A 2016;650:225–32. https://doi.org/10.1016/j.msea.2015.10.032. [12] Yang Q, Sun K, Yang C, Sun M, Peng H, Shen X, et al. Compression and superelasticity behaviors of NiTi porous structures with tiny strut fabricated by selective laser melting. J Alloys Compd 2021;858:157674. https://doi.org/10.1016/j.jallcom.2020.157674. [13] Saedi S, Turabi AS, Andani MT, Haberland C, Karaca H, Elahinia M. The influence of heat treatment on the thermomechanical response of Ni-rich NiTi alloys manufactured by selective laser melting. J Alloys Compd 2016;677:204– 10. https://doi.org/10.1016/j.jallcom.2016.03.161. [14] Lee J, Shin YC. Effects of Composition and Post Heat Treatment on Shape Memory Characteristics and Mechanical Properties for Laser Direct Deposited Nitinol. Lasers Manuf Mater Process 2019;6:41–58. https://doi.org/10.1007/s40516-019-0079-5. [15] Henderson E, Nash DH, Dempster WM. On the experimental testing of fine Nitinol wires for medical devices. J Mech Behav Biomed Mater 2011;4:261–8. https://doi.org/10.1016/j.jmbbm.2010.10.004. [16] Lu X, Li G, Liu L, Zhu X, Tu ST. Effect of ambient temperature on compressibility and recovery of NiTi shape memory alloys as static seals. Adv Mech Eng 2017;9:1–9. https://doi.org/10.1177/1687814017692287. [17] Iasnii VP, Junga R. Phase Transformations and Mechanical Properties of the Nitinol Alloy with Shape Memory. Mater Sci 2018;54:406–11. https://doi.org/10.1007/s11003-018-0199-7. [18] Zhang C, Jin J, He M, Yang L. Compressive Mechanics and Hyperelasticity of Ni-Ti Lattice Structures Fabricated by Selective Laser Melting. Crystals 2022;12. https://doi.org/10.3390/cryst12030408. [19] McCummiskey E, Dempster WM, Nash DH, Ashton TR, Stevenson DG. The determination and evaluation of nitinol constitutive models for finite element analysis. Appl Mech Mater 2007;7–8:81–8. https://doi.org/10.4028/www.scientific.net/AMM.7-8.81. [20] Chen W, Gu D, Yang J, Yang Q, Chen J, Shen X. Compressive mechanical properties and shape memory effect of NiTi gradient lattice structures fabricated by laser powder bed fusion. Int J Extrem Manuf 2022;4. https://doi.org/10.1088/2631-7990/ac8ef3. [21] Dora TRK, Goud R, Sahadevan A, Chand AH, Jha R, De PS, et al. Investigations into sample geometry effects on the superelastic and fatigue behavior of Nitinol: Modeling and experiments. Materialia 2021;20. https://doi.org/10.1016/j.mtla.2021.101256. [22] Bourke D, Selvam KT, Obeidi MA, Ahad IU, Brabazon D. Effect of powder and process parameters on in-situ alloying of nitinol during laser powder bed fusion. J Mater Res Technol 2024;30:7988–97. https://doi.org/10.1016/j.jmrt.2024.05.178. [23] Vaglio E, De Monte T, Lanzutti A, Totis G, Sortino M, Fedrizzi L. Single tracks data obtained by selective laser melting of Ti6Al4V with a small laser spot diameter. Data Br 2020;33:106443. https://doi.org/10.1016/j.dib.2020.106443. [24] Wei S, Zhang J, Zhang L, Zhang Y, Song B, Wang X, et al. Laser powder bed fusion additive manufacturing of NiTi shape memory alloys: a review. Int J Extrem Manuf 2023;5. https://doi.org/10.1088/2631-7990/acc7d9. [25] Xiang Z, Yang Q, Zhang T, Shen X, Chen J, Huang S. Effect of Process Parameters on Superelasticity of LPBF Ni-Rich Ni51.3Ti48.7 Shape Memory Alloy. Metals (Basel) 2024;14:961. https://doi.org/10.3390/met14090961. [26] Wang X, Yu J, Liu J, Chen L, Yang Q, Wei H, et al. Effect of process parameters on the phase transformation behavior and tensile properties of NiTi shape memory alloys fabricated by selective laser melting. Addit Manuf 2020;36:101545. https://doi.org/10.1016/j.addma.2020.101545. [27] Xue L, Atli KC, Picak S, Zhang C, Zhang B, Elwany A, et al. Controlling martensitic transformation characteristics in defect-free NiTi shape memory alloys fabricated using laser powder bed fusion and a process optimization framework. Acta Mater 2021;215:117017. https://doi.org/10.1016/j.actamat.2021.117017. [28] Zhan J, Wu J, Ma R, Li K, Lin J, Murr LE. Tuning the functional properties by laser powder bed fusion with partitioned repetitive laser scanning: Toward editable 4D printing of NiTi alloys. J Manuf Process 2023;101:1468–81. https://doi.org/10.1016/j.jmapro.2023.07.009. [29] Biffi CA, Tuissi A, Demir AG. Martensitic transformation, microstructure and functional behavior of thin-walled Nitinol produced by micro laser metal wire deposition. J Mater Res Technol 2021;12:2205–15. https://doi.org/10.1016/j.jmrt.2021.03.108. [30] Chernyshikhin S V., Pelevin IA, Karimi F, Shishkovsky I V. The Study on Resolution Factors of LPBF Technology for Manufacturing Superelastic NiTi Endodontic Files. Materials (Basel) 2022;15. https://doi.org/10.3390/ma15196556. [31] Jiang H, Wang X, Xi R, Li G, Wei H, Liu J, et al. Size effect on the microstructure, phase transformation behavior, and mechanical properties of NiTi shape memory alloys fabricated by laser powder bed fusion. J Mater Sci Technol 2023;157:200–12. https://doi.org/10.1016/j.jmst.2023.02.026. [32] Kus K, Breczko T. DSC-investigations of the effect of annealing tempe-rature on the phase transformation behaviour in Ni-Ti shape memory alloy. Mater Phys Mech 2010;9:75–83. [33] Li G, Yu T, Zhang N, Chen M. The effect of Ni content on phase transformation behavior of NiTi alloys: An atomistic modeling study. Comput Mater Sci 2022;215:111804. https://doi.org/10.1016/j.commatsci.2022.111804. [34] Martin AA, Calta NP, Khairallah SA, Wang J, Depond PJ, Fong AY, et al. Dynamics of pore formation during laser powder bed fusion additive manufacturing. Nat Commun 2019;10:1–10. https://doi.org/10.1038/s41467019-10009-2. [35] Shayesteh Moghaddam N, Saedi S, Amerinatanzi A, Hinojos A, Ramazani A, Kundin J, et al. Achieving superelasticity in additively manufactured NiTi in compression without post-process heat treatment. Sci Rep 2019;9:1–11. https://doi.org/10.1038/s41598-018-36641-4. [36] Zhu JN, Ding Z, Borisov E, Yao X, Brouwer JC, Popovich A, et al. Healing cracks in additively manufactured NiTi shape memory alloys. Virtual Phys Prototyp 2023;18. https://doi.org/10.1080/17452759.2023.2246437. [37] Liu T, Wu Z, Zhou W, Zhong M, Lin J, Yang Y. Quasilinear pseudoelasticity and small hysteresis in SLM-fabricated NiTi. J Alloys Compd 2023;933:167694. https://doi.org/10.1016/j.jallcom.2022.167694. [38] Nematollahi M, Saghaian SE, Safaei K, Bayati P, Bassani P, Biffi C, et al. Building orientation-structure-property in laser powder bed fusion of NiTi shape memory alloy. J Alloys Compd 2021;873:159791. https://doi.org/10.1016/j.jallcom.2021.159791. [39] Zhao S, Liang Q, Liang C, Wang D, Ji Y, Wang Y, et al. Quasi-Linear Superelasticity with Ultralow Modulus in Tensile Cyclic Deformed TiNi Strain Glass. Adv Eng Mater 2022;24:1–8. https://doi.org/10.1002/adem.202200239. [40] Yang H, An K, Wang YD. Anomalous high-temperature quasi-linear superelasticity of Ni-Fe-Ga-Co shape memory alloy. J Alloys Compd 2022;909. https://doi.org/10.1016/j.jallcom.2022.164808. [41] Dornelas VM, Oliveira SA, Savi MA, Pacheco PMCL, de Souza LFG. Fatigue on shape memory alloys: Experimental observations and constitutive modeling. Int J Solids Struct 2021;213:1–24. https://doi.org/10.1016/j.ijsolstr.2020.11.023. Appendix Figure 7 – Dependence of the single track (a) width; (b) depth; (c) height and (d) contact angle on the scanning speed and laser power Table 7 – Quadratic regression coefficients and intercept; MSE and R2 are the mean square error and coefficient of determination, respectively – Quadratic regression coefficients and intercept; MSE and R2 are the mean squared error and coefficient of determination, respectively (a) (b) (c) (d) Quadratic regression Intercept Coefficients Property β0 β1 β2 β3 β4 β5 MSE R2 Track width 1.61e2 -1.58e-1 1.05e0 3.49e-5 -9.12e-5 -1.03e-3 584.32 0.90 Track depth 5.79e1 -2.93e-1 2.26e0 8.88e-5 -4.28e-4 -2.48e-3 1940.1 0.85 Track height 3.72e1 1.90e-3 9.22e-2 -6.51e-7 -9.82e-6 -1.78e-4 128.27 0.05 Contact angle 1.35 e2 -3.12e-2 1.24e-1 6.63e-6 4.93e-6 -7.53e-5 154.35 0.29 Figure 8 – Section of the window with the process parameters used for the preparation of the individual single tracks (red – discarded, orange – considered but discarded, green – used further)