Tailoring grain sizes of the biodegradable iron-based alloys by pre-additive manufacturing microalloying

We demonstrated the design of pre-additive manufacturing microalloying elements in tuning the microstructure of iron (Fe)-based alloys for their tunable mechanical properties. We tailored the microalloying stoichiometry of the feedstock to control the grain sizes of the metallic alloy systems. Two specific microalloying stoichiometries were reported, namely biodegradable iron powder with 99.5% purity (BDFe) and that with 98.5% (BDFe-Mo). Compared with the BDFe, the BDFe-Mo powder was found to have lower coefficient of thermal expansion (CTE) value and better oxidation resistance during consecutive heating and cooling cycles. The selective laser melting (SLM)-built BDFe-Mo exhibited high ultimate tensile strength (UTS) of 1200 MPa and fair elongation of 13.5%, while the SLM-built BDFe alloy revealed a much lower UTS of 495 MPa and a relatively better elongation of 17.5%, indicating the strength enhancement compared with the other biodegradable systems. Such an enhanced mechanical behavior in the BDFe-Mo was assigned to the dominant mechanism of ferrite grain refinement coupled with precipitate strengthening. Our findings suggest the tunability of outstanding strength-ductility combination by tailoring the pre-additive manufacturing microalloying elements with their proper concentrations.

www.nature.com/scientificreports/ applications, biodegradable bone implants were developed to avoid the need of secondary surgery to remove the implants 14 . However, the main issue of earlier observed biodegradable iron coronary stent results from the low degradation rate 15 . The development of biodegradable materials that have high mechanical properties with a suitable degradable rate is imperative for implant applications 16 . It has been reported that the reduced grain size in the biodegradable iron coronary stent would increase the corrosion rate 17 . Meanwhile, additive manufacturing is a fabrication process where the parts are built layer by layer with a very localized heating from a laser source [18][19][20][21] . With the assistance of rapid development of the additive manufacturing, the Fe-based biodegradable implants have drawn much attention because of great possibilities in building complex geometry and customized parts 22,23 . Tailoring the enhanced mechanical properties of pure Fe via grain refinement by the selective laser melting (SLM) process was explored 24,25 . In fact, there have been many effective grain refinement methods reported for the Fe-based alloys and the steels [26][27][28] . However, achieving grain refinement by undercooling is full of challenges due to recalescence effect during phase transformation and it is even more difficult for a large batch process 29,30 . Up to date, the post heat treatments on the additive manufacturing parts are well summarized 31 .
Microalloying element molybdenum (Mo) has been effectively used to increase the strength of steel and its potential applications as artificial implants have recently been explored 32,33 . Our primary objective is to tailor the microstructure in simultaneously improving both the mechanical properties and degradation rate of the Fe-based alloys by the addition of trace amounts of biodegradable Mo. In the present study, we investigated the design of microalloying elements for tunable mechanical properties via controllable grain sizes in two horizontally built SLM biodegradable Fe-based alloys of BDFe and BDFe-Mo. To understand the dominant mechanisms governing the mechanical properties in the two SLM-built BDFe samples, we carried out monotonic tension experiments. Since thermal stability of the feedstock materials is especially important during the building process 34,35 , the quality of pre-additive manufacturing microalloying BDFe and BDFe-Mo powders during multiple heating and cooling sequences is one of the prerequisites to be examined in this study.

Results
Thermal stability during continuous heating. Figure 1 shows the coefficient of thermal expansion (CTE) and lattice constant change using temporally coherent X-ray diffraction upon continuous heating from room temperature (RT) to 900 °C in the two BDFe powders in comparison with the previously reported pure Fe. Two BDFe powders disclosed similar CTE values in Fig. 1a, except those in the temperature range from 50 to 100 °C in which the CTE value of BDFe-Mo was higher than that of BDFe. Moreover, compared with the pure Fe 36 , the two BDFe powders showed lower CTE values. The increased chromium (Cr) and silicone (Si) concentrations were reported to reduce the CTE of the ferritic iron 37 while the increased carbon (C) and manganese (Mn) contents were found to increase the CTE of the Fe alloys [38][39][40] . The latter one is supposed to be the dominant contribution to the slightly higher CTE value of the BDFe-Mo although there exist complex competitive interactions between the individual microalloying elements. A slight decrease in the CTE values of both BDFe powders from 500 to 900 °C relates to the transformation into Fe's magnetic phase from 570 to 877 °C 41,42 . As shown in Fig. 1b, the lattice constants upon continuous heating from RT to 900 °C were similar in the two BDFe powders and they were lower compared with the reported lattice constant of pure Fe 43 , which was in accordance with the variation trend of CTE.
Thermal stability of pre-additive manufacturing powders is very important for a better quality of the as-built parts fabricated via additive manufacturing 20 . Armentani et al. reported that a butt welding with higher CTE tends to produce higher residual stresses 44 , which may cause serious limitation on the practical use of the additive manufactured parts due to their influence on plastic deformations, microcracks, and load resistance 45,46 . In addition, material with lower CTE was found to possess even lower CTE value after SLM process 47 . Therefore, www.nature.com/scientificreports/ the pre-additive manufacturing BDFe powders with lower CTE are expected to enhance the thermal stability and possess good mechanical properties in the SLM-built parts.
Phase and microstructure stability during thermal cycling. Thermal cycling test was conducted to identify the phase stability of two BDFe powders, which would be experienced during the SLM process. Figure 2a shows the variation of lattice constant as a function of temperature during the repeated heating and cooling cycles in which no evident difference was obtained between the two BDFe powders. To further acquire the phase transformation during multiple heating and cooling sequences, the evolution of diffraction patterns at RT after each thermal cycle was recorded in Fig. 2b  phases. In addition, more minor peaks were obtained in the BDFe rather than in the BDFe-Mo. We analyzed the integrated intensities of the major peaks of bcc Fe, the minor peaks of Fe 2 O 3 and Fe 3 O 4 to investigate their evolutions during the heating and cooling processes, depicted in Fig. 3. As shown in Fig. 3a and  www.nature.com/scientificreports/ phases during the successive heating and cooling sequences, we quantified the phase fraction by calculating their integrated area fractions, as listed in Table 1. It can be seen from Table 1 that the amounts of oxidized Fe 2 O 3 and Fe 3 O 4 in the BDFe were much higher than those in the BDFe-Mo during thermal cycling, indicating better oxidation resistance of the BDFe-Mo powder. The existence of Mo and the increased Si and Cr contents might hinder the diffusion of Fe cations towards the metal/ oxide interface and thus significantly reduce the high temperature oxidation in the BDFe-Mo 48 . The tendency that transforms to the more stable Fe 3 O 4 phase was also found 49 .
Tunable mechanical performance. Figure 4 presents the tensile properties of the two horizontally built SLM BDFe alloys in comparison with other biodegradable alloys. It can be seen from the uniaxial tensile engineering stress strain (S-S) curves in Fig. 4a that a salient strength enhancement was evidently obtained in the BDFe-Mo alloy. Specifically, the macroscopic yield strength (YS) and ultimate tensile strength (UTS) of the BDFe-Mo was 1193 and 1200 MPa, which were 2.5 times greater than those of the BDFe (451 and 495 MPa, respectively). The elongation to fracture was reduced by 23% from the BDFe (17.5%) to the BDFe-Mo (13.5%). Figure 4b depicts the tensile strength and elongation of the two BDFe alloys compared with the Fe, Zn, and Mgbased biodegradable alloys fabricated using the conventional methods and SLM 5,50-57 . Both the SLM-built BDFe alloys revealed superior tensile properties with an especially remarkable mechanical strength observed in the BDFe-Mo. Several possible reasons governing the strengthening mechanism in the BDFe-Mo might be the grain boundary strengthening, the increased dislocation density, and the precipitation of second phase 24 , which will be unraveled for the design of pre-additive manufacturing microalloying elements regarding to the improved strength-ductility combination.

Discussion
Grain boundary strengthening effect. One of the general reasons leading to the grain boundary strengthening in laser process manufacturing is the high speed of laser scanning or fast cooling rate. During the SLM process, pure Fe experiences multiple ferrite (α) → austenite (γ) → ferrite (α) phase transformation during the successive cycles of rapid heating and cooling, resulting in hierarchical microstructure with 0.1-0.3 μm sized subgrains, separated by low-angle grain boundaries 25,58 . The optimal laser power of 150 W and faster laser scanning speed of 1.2 m/s were expected to significantly promote the increased nucleation rate and to achieve microstructure refinement of the SLM-built BDFe alloys during rapid solidification process. Although the two BDFe samples experienced the same SLM process, a more significant strengthening effect was attained in the BDFe-Mo, generally originated from the major difference in microstructural evolution.  www.nature.com/scientificreports/ The microstructural evolution in the three orthogonal planes (XY, YZ, and XZ) was characterized using electron backscatter diffraction (EBSD) to clarify the grain shape, grain size, and second phase precipitation of the two SLM-built BDFe specimens, as described in Fig. 5. EBSD analysis was taken at the corner of the tensile specimens (as illustrated by the red marked square in Fig. 5a), where it is assumed not to be experienced the plastic deformation. Both BDFe samples exhibited a random distribution of majorly equiaxed grains accompanying with some slightly elongated grains, in agreement with similar observations on the grain shape appeared during SLM fabrication process. The building direction in the SLM-built BDFe samples was along the Z-axis, as shown in Fig. 5b,c. The formation of slightly elongated grains was ascribed to temperature gradients in the melted pre-additive manufacturing powder layer along the building direction during rapid solidification process [59][60][61][62][63] .
It can be seen from Fig. 5b,c that the grain size of the BDFe-Mo was much smaller than that of the BDFe. The imageJ software was employed to determine the average grain sizes of the two BDFe samples in the three orthogonal planes, reported in Table 2. The distribution of grain size was not uniform in the three orthogonal planes in which the grain sizes in the XZ and YZ planes were similar but they were smaller than those in the XY plane for both BDFe specimens. To calculate the strength enhancement as a contribution of grain boundary strengthening in the two BDFe samples, the empirical Hall-Petch (H-P) relationship was analyzed based on Eq. (1) 24,64 .
where �σ cr is the increment of yield strength, k is the H-P slope of the contribution of grain boundary strengthening, and D is the average grain size. Following the previously reported work, the value of k was chosen as 20 MPa/mm 1/265 . The calculated yield strengths using the average grain sizes of the three orthogonal planes were determined to be 392.19 and 1,105.62 MPa which were slightly smaller than the macroscopic yield strengths of 451 and 1,193 MPa in the BDFe and BDFe-Mo obtained from the S-S curves, respectively.
To elucidate the other possible reasons contributing to the strengthening behavior in the SLM-built BDFe specimens, the microstructures of both samples in the XY plane in Fig. 5d,e were coupled with their crystal orientation maps, as shown in Fig. 5f,g in the same magnification. In accordance with the XRD results, both BDFe samples revealed the ferrite grains together with the second phase precipitate at the grain boundaries which was identified as the cementite (Fe 3 C). In the Fe-based alloys, the metal carbides precipitate out of the supersaturated ferrite 66,67 and form at the grain boundaries 68 . The growth of Fe 3 C precipitates was assigned to the increased C content coupled with the partitioning of Si occurring around 300 °C as the SLM-built specimens experienced the consecutive reheating cycles 69 . The precipitation of Fe 3 C at the grain boundaries caused the pinning effect in inhibiting the growth rate of matrix grains and thus refined the ferrite grains. The existence of more grain boundaries due to the increased contents of C and Si promotes the ferrite nucleation for achieving a uniformly finer grained microstructure in the BDFe-Mo. Moreover, the increased dislocation density during the successive heating and cooling sequences derived from a large difference of CTE between the matrix and Fe 3 C contributed to the strength enhancement 69 . The precipitation strengthening of Fe 3 C at the grain boundary greatly enhanced with the increased C concentration and this strengthening mechanism significantly contributed to the macroscopic yield strength 70 . The results of microstructure investigation indicated that the ferrite grain refinement accompanying with the precipitation strengthening were the major strengthening mechanisms for the superior mechanical strength in the BDFe-Mo.
Since the primary reason affecting the grain boundary strengthening between the two SLM-built BDFe specimens is the difference in their constituent microalloying elements, an examination towards the chemical distribution of individual alloying elements via X-ray fluorescence (XRF) mapping was necessary. The XRF maps in Fig. 6a exhibited a localized chemical inhomogeneity of the constituent alloying elements where the Fe-Mn, Fe-Cr rich regions, and Cr-C-Si segregation were obtained in the BDFe specimen. The heterogeneous distributions of alloying elements in the BDFe-Mo were likely similar to those in the BDFe in which the Fe-Mn rich region was associated with a more uniform distribution of Cr, C, and Si (Fig. 6b). Furthermore, a preferred segregation of Mo with C was visible in the BDFe-Mo specimen. Low mobility of Mn-induced large undercooling during phase transformation from the austenite to the fully ferrite resulted in strong nucleation of ferrite during cooling in the Fe-Mn alloy 71 . The dissolved Cr in the Fe matrix impeded the dislocation movement and thus strengthened the materials, due to the lattice distortion caused by a large atomic size difference between Fe and Cr 72 . The addition of Mo in the BDFe-Mo has been found to facilitate the solid solution strengthening effect and possibly induce significant ferrite grain refinement due to the delay of dynamic recrystallization 73 . The grain refinement may also occur with the existence of Mo and the increased Mn content, which could lower the temperature for phase transformation from the austenite to the ferrite 74 . Furthermore, the formed Mo-C rich region in the BDFe-Mo may attribute to the precipitation hardening 73 . The impressive mechanical strength behavior in the BDFe-Mo inferred that an appropriate increase in the concentration of microalloying elements gave rise to the grain refinement of the ferrite and precipitation strengthening of metal carbides.
In order to better understand thermal behaviors affecting the grain growth during the SLM process, the evolution of heat energy stored in the BDFe powders was explored using differential scanning calorimetry (DSC) analysis. Figure 7 presents the heat absorption and heat release of the BDFe powders during continuous heating and cooling up to 1000 °C. The DSC plots disclosed similar broad exothermic peaks in both BDFe powders upon heating, however, the exothermic peaks were more obvious in the BDFe rather than in the BDFe-Mo. This Table 2. The calculated strength enhancements using the average and dominant grain sizes of the three orthogonal planes in the two BDFe samples in comparison with the macroscopic yield strengths obtained from the S-S curves. www.nature.com/scientificreports/ result indicated more heat energy released in the BDFe and implied more evident grain growth in the BDFe 75 . Similar features were obtained in both BDFe powders upon cooling and the BDFe was more endothermic than the BDFe-Mo, indicating more heat absorbed in the BDFe. The peaks located at 772 °C and 891 °C upon cooling correspond to the Curie temperature and the austenite to ferrite transformation temperature 75 . The ferrite phase with a Fe 3 C observed in the two SLM-built BDFe alloys was beneficial for enhancing high mechanical strength.

Summary
The SLM-built BDFe alloys with optimal fabrication parameters revealed good mechanical properties. Compared with the BDFe, the BDFe-Mo exhibited remarkably superior YS of 1193 MPa and UTS of 1200 MPa together with a relatively fair elongation of 13.5%. Such a prominent strengthening behavior in the BDFe-Mo was assigned to a significant ferrite grain refinement coupled with the precipitate strengthening mechanism, tailored by the higher optimal concentrations of C, Cr, Mn, and Si accompanying with the addition of Mo. Designing the pre-additive manufacturing microalloying elements with their appropriate contents is promising for additive manufactured parts owning excellent strength-ductility combination without post heat treatments.

Materials and methods
Feedstock materials. The two kinds of spherical powders, BDFe with Fe purity higher than 99.5% and BDFe-Mo with Fe purity higher than 98.5%, were used in this study. The elemental compositions of the two powders were examined by X-ray fluorescent method using Rigaku ZSX Primus IV machine and tabulated in Table 3. The two powders were produced by gas atomization method using VIGA equipment technology, which enabled the spherical powder particles to have less amount of interstitial impurities. The particle sizes of the two powders were in the range of 10-60 µm and were determined by air-flow powder classifier.  www.nature.com/scientificreports/ Selective laser melting process. A selective laser melting process was carried out using AM100, Industrial Technology Research Institute (ITRI) self-made machine. The working chamber was filled with argon (Ar) gas with outlet pressure of 2 bar in order to maintain the oxygen level below 0.1% for preventing iron oxidation during fabrication process. This machine provides a single laser beam with the spot size of 60 µm and laser power up to 500 W. A S45C steel was used as a base plate with the area of 100 × 100 mm 2 on the building platform.
Pin-loaded tension test specimens were fabricated in horizontal built orientation using laser power of 150 W, laser speed of 1.2 m/s with the layer thickness of 30 µm. The fabrication process also uses stripe laser scanning strategy with a hatch space of 60 µm. The pin-loaded tension test specimens were prepared with an overall length of 60 mm, gauge length and width of 20 mm and 10 mm, respectively, following ASTM standard E8/E8M-09 76 . For statistical considerations, three pin-loaded tension test specimens of each BDFe powder were measured.

Electron backscatter diffraction. Microstructure observation via electron backscatter diffraction (EBSD)
was performed using the JEOL JSM-6500F/OXFORD Nordlys electron microscope operated at 15 kV and was analyzed using AZTEC software. The undeformed region of the tensile sample was cut into three distinct specimens for the representative microstructures in the three orthogonal planes (XY, YZ, and XZ). The measurement was made at the ambient temperature of 20 °C with a relative humidity of 50%.
Continuous heating and thermal cycling using temporally coherent X-ray diffraction. Insitu synchrotron X-ray diffraction measurements were conducted at Taiwan Photon Source (TPS) 09A beamline (BL) using temporally coherent X-ray diffraction of the National Synchrotron Radiation Research Center (NSRRC). This station was designed for hard X-ray scattering to study the static and dynamic structure behaviors of crystalline materials using a 9-circle diffractometer coupled with a heater, which can heat the two BDFe powders up to 900 °C.
Diffraction profiles were recorded during continuous heating at room temperature, then from 50 °C up to 900 °C with an increment of 50 °C and they were analyzed using the General Structure Analysis Software (GSAS)-II, which is an open source software for determination of crystal structures from different phases and lattice sizes 77 . After fitting the diffraction patterns obtained at different temperatures, the changes of the lattice parameters were calculated to yield the CTE. The average linear CTE was calculated using Eq. (2) 78 .
where L 0 is the initial lattice constant at room temperature T o . ΔL is the change in the lattice constant at a specific temperature and ΔT is the change in the temperature.
For thermal cycling test, the two BDFe powders were heated from RT up to 900 °C and cooled down to RT in the 1st cycle. The temperature was heated up to 700 °C and decreased to RT in the 2nd cycle, followed by an increase of temperature up to 500 °C and then cooled to RT in the 3rd cycle. The heating and cooling rates in the thermal cycling test were 10 °C/min. Diffraction patterns were recorded every 100 °C increment during heating and cooling. X-ray nano-diffraction. X-ray nano diffraction (XND) at TPS 21A BL of the NSRRC was exploited to map the X-ray fluorescence intensity of each alloying element with a spatial resolution of 90 nm for the examination of chemical homogeneity in the two BDFe samples. A mapping area of 50 × 50 µm 2 with a step size of 1.0 µm was recorded and each mapping consisted of 1 × 1 pixels. More details of the related protocols are archived 21,79 . Differential scanning calorimetry. To examine the thermal behavior subjected to heating and cooling, the two BDFe powders were measured by differential scanning calorimetry. The experiments were conducted by the continuous heating up to 1000 °C from RT with a heating rate of 10 °C/min, followed by the continuous cooling to RT with a cooling rate of 10 °C/min.

Data availability
The data will be made available on request.  www.nature.com/scientificreports/