2024 Volume 64 Issue 2 Pages 303-315
This study was set to fundamentally investigate the characteristics of austenite reversion occurring in maraging steels additive-manufactured by laser powder bed fusion (L-PBF). The maraging steel samples manufactured under different L-PBF process conditions (laser power P and scan speed v) were subjected to heat treatments at 550°C for various durations, compared with the results of the austenitized and water-quenched sample with fully martensite structure. The L-PBF manufactured samples exhibited the martensite structure (including localized austenite (γ) phases) containing submicron-sized cellular structures. Enriched alloy elements were detected along the cell boundaries, whereas such cellar structure was not found in the water-quenched sample. The localized alloy elements can be rationalized by the continuous variations in the γ-phase composition in solidification during the L-PBF process. The precipitation of nanoscale intermetallic phases and the following austenitic reversion occurred in all of the experimental samples. The L-PBF manufactured samples exhibited faster kinetics of the precipitation and austenite reversion than the water-quenched sample at elevated temperatures. The kinetics changed depending on the L-PBF process condition. The enriched Ni element (for stabilizing γ phase) localized at cell boundaries would play a role in the nucleation site for the formation of γ phase at 550°C, resulting in enhanced austenite reversion in the L-PBF manufactured samples. The variation in the reaction kinetics depending on the L-PBF condition would be due to the varied thermal profiles of the manufactured samples by consecutive scanning laser irradiation operated under different P and v values.
In recent years, additive manufacturing technologies have made extensive progress, enabling the fabrication of complex three-dimensional shapes that are impossible using conventional processes.1,2,3,4) The additive manufacturing process is often referred to as the 3D printing process.3) One of the most popular additive manufacturing processes for metals and alloys is the laser powder bed fusion (L-PBF).1,2,3,4,5) The alloy powder layer is bedded inside a processing chamber in an inert gas atmosphere, and the powder layer was locally melted by scanning laser irradiation. Repeating these processes enables the manufacturing of metal components. The L-PBF process enables the manufacturing of complex-shaped parts of steel and is expected to be applied not only to lightweight structural parts but also to fabricate molds for press-forming and injection processes. The L-PBF process6,7,8) for steel powders has been reported for SUS316L,9,10) SUS304L,10) and maraging steel,11,12) and it has been recently reported for various ferrous alloys including duplex stainless steels,13) 17-4PH steel14) (SUS630 equivalent), tool steels15) and high-strength steels.16)
With a view to realizing the fabrication of arbitrarily-shaped molds of maraging steel using the L-PBF process, we have systematically investigated the effects of L-PBF process conditions (laser power P and laser scanning speed v) on the relative density of maraging steel samples for optimizing L-PBF laser conditions for the fabrication of fully dense samples.17) As a result, it was found that the arrangement of P and v based on the deposited energy density,18) which takes into account the thermal diffusion length in solid parts, is more effective for densification of the maraging steel samples (rather than the commonly used volumetric energy density19)). It was confirmed that the approach was vailed to different alloy systems20,21) as well. The search for optimum conditions in the L-PBF process has identified a range of P and v for manufacturing maraging steel samples with relative densities above 99%.17) Preliminary investigations of microstructures of the maraging steel samples manufactured under different conditions (P and v) revealed the presence of fine retained austenite (γ) phases inside the martensitic microstructure in all of the L-PBF manufactured samples,22) which was consistent with other research reports.23,24) It was also found that the volume fraction of the residual γ phase varied depending on laser parameters of P and v, suggesting that the L-PBF processing condition might have a significant effect on the austenite reversion by subsequent heat treatments. Controlling the austenite reversion by initial L-PBF processing parameters would improve the strength and ductility of L-PBF manufactured maraging steel specimens.25) There have been many reports on the precipitation of fine intermetallic phases11,26,27) and changes in mechanical properties associated with post heat-treatments11,12,25,28) in maraging steel specimens manufactured by the L-PBF process, whereas few studies29,30) have systematically investigated microstructural changes associated with the austenite reversion at elevated temperatures. Conde et al.30) investigated the kinetics of austenite reversion (α→γ) in the L-PBF manufactured maraging steel samples by in-situ synchrotron radiation X-ray diffraction analyses for the variations in constituent phases at different temperatures ranging from 530°C to 670°C. However, detailed microstructural changes at elevated temperatures still remain unclear, and the effect of L-PBF processing conditions on the austenite reversion behavior has not been reported.
In this study, in order to understand the effect of L-PBF processing conditions (P, v) on the austenite reversion behavior, we systematically investigated microstructural changes in isothermal heat treatments of maraging steel samples containing fine retained γ phases with different volume fractions controlled by the L-PBF processing conditions. The results were compared with those of the maraging steel sample fully austenitized and subsequently water-quenched. Based on these results, the characteristics of the austenite reversion behavior of the L-PBF manufactured maraging steel samples were discussed.
In this study, maraging steel (18Ni300) alloy powder prepared by gas atomization (supplied by 3D SYSTEMS) was used. The alloy powder particles appear spherical22) and have a particle size distribution of approximately 1–30 μm. Table 1 shows the nominal composition of the experimental alloy powder and the compositions of the powder and the L-PBF manufactured samples (measured by inductively coupled plasma atomic emission spectrometry and infrared absorption). The alloy composition of the L-PBF sample is almost the same as that of the alloy powder, whereas it contains about 0.1 mass% oxygen. A ProX 200 (3D SYSTEMS) was used to fabricate the maraging steel. In this study, the laser scanning speeds (v) of 0.4–3.0 m/s and laser powers (P) of 45–255 W were applied under the fixed other processing parameters: beam focal spot size of approximately 100 μm, powder stacking thickness of 30 μm, and laser hatch distance of 50 μm.17) A hexagonal laser scanning pattern with a size of 40 mm was used, and stacked powder-bed layers were performed by rotating 90 degrees on each layer.22) The manufactured samples exhibited a cube with a size of approximately 15 × 15 × 15 mm3. The Archimedes method was used to measure the relative density of the manufactured samples. The relative densities of the samples manufactured under various P and v values are shown in Fig. 1(a). Figure 1(b) shows the preliminary measurement results (bubble chart) of the area fraction of the austenite (γ) phase measured by electron backscatter diffraction (EBSD) analysis for the samples with relative densities above 98%. Samples fabricated under high P and low v values tend to show a higher area fraction of the γ phase. In this study, we used two samples fabricated under two different conditions (P: 230 W, v: 1.67 m·s−1 and P: 213 W, v: 2.08 m·s−1) in which the distribution of the retained γ phase is expected to be different. The two samples are referred to as high and low energy-density samples in terms of volumetric energy density.18) Both samples were held at 550°C for 1–100 h and then water quenched. It has been reported that the volume fraction of the γ phase decreases significantly (due to the occurrence of martensitic transformation in cooling) when the maraging steel is cooled after holding at high temperatures above 650°C.30) In this study, the holding temperature was set at 550°C, where martensitic transformation does not occur in cooling, in order to systematically investigate the microstructural changes during the isothermal holding at an elevated temperature. As a comparison, the manufactured samples were held at 1000°C (corresponding to a γ single-phase region) for 1 h and then water-quenched. The quenched samples were subjected to the same heat treatment at 550°C.
| Ni | Co | Mo | Ti | Al | C | O | ||
|---|---|---|---|---|---|---|---|---|
| Nominal | 17–19 | 8.5–9.5 | 4.5–5.2 | 0.6–0.8 | 0.05–0.15 | ≤0.03 | – | |
| Measured | powder | 18.2 | 9.5 | 4.9 | 1.0 | 0.07 | 0.01 | 0.13 |
| built | 18.2 | 9.1 | 5.1 | 0.8 | 0.06 | 0.01 | 0.11 | |
These samples were mechanically polished and then finished with colloidal silica. The sample surface was etched with a 4% Nital solution for approximately 60 s. The sample surfaces were observed using an optical microscope and a scanning electron microscope (SEM: JEOL Ltd., JSM-IT510). Crystal orientations were measured by EBSD analyses using a field-emission type scanning electron microscope (JSM-7001F, JEOL Ltd.). An acceleration voltage of 20 kV and an electron beam scanning width of 0.3 mm were applied for the EBSD measurements. A X-ray diffractometer (ULTIMA IV, Rigaku Corporation) with a Cu tube (wavelength: 0.154 nm) was used for the phase identification of the samples. The measurement conditions were an X-ray power of 40 kV and a step width of 0.02°. The manufactured and subsequently heat-treated samples were cut into thin sheets and mechanically polished into disks with a thickness of about 0.1 mm. The disks were prepared for transmission electron microscope (TEM) observation by twin-jet electropolishing using a 10% perchloric acid-ethanol mixture. Microstructural observations were performed using a TEM (JEOL JEM-2100F/HK, JEOL Ltd.) at an acceleration voltage of 200 kV to capture bright-field images and scanning transmission electron microscope (STEM)-dark field images, and elemental analysis was performed using energy dispersive X-ray analysis (EDS). The Vickers hardness of each sample was measured using a microhardness tester (HMV-G30, Shimadzu Corporation) with a test load of 9.8 N and a loading time of 15 s. Differential scanning calorimetry (DSC) measurements (DSC6300, SII Nanotechnology Co., Ltd.) were performed using 3×3×2 mm3 rectangular samples. Measurements were performed under the following conditions: standard sample Al2O3 powder, Ar atmosphere, and heating rate of 0.67°C·s−1.
Figure 2 presents SEM images and phase maps and crystallographic orientation distribution map obtained by EBSD analyses for the maraging steel samples fabricated under different L-PBF conditions (P: 230 W, v: 1.67 m·s−1 and P: 213 W, v: 2.08 m·s−1) and the water-quenched sample. All images are observed from a cross-section parallel to the building direction. The L-PBF processed samples exhibited a melt-pool structure (Figs. 2(a), 2(d)). The melt pool depth was measured to be several tens μm. The microstructural morphology corresponded to the shape of a melt pool in which the powder-bed layers were locally melted and rapidly solidified in the scanning laser irradiation. Such a morphology was not observed, whereas martensitic structure was found in the water-quenched sample (Fig. 2(g)). In the L-PBF sample fabricated under a high energy density condition (P: 230 W, v: 1.67 m·s−1), fine γ phase was distributed in the α-phase matrix and tends to be localized at the melt pool boundaries (Fig. 2(b)). In the L-PBF sample fabricated under a low energy density condition (P: 213 W, v: 2.08 m·s−1), the γ phase was scarcely observed, and the area fraction measured by EBSD analysis was less than 1% (Fig. 2(e)). This result was found in the water-quenched sample (Fig. 2(h)). The crystal orientation distribution maps present the typical orientation distribution of the lath martensitic structure (Figs. 2(c), 2(f), 2(i)). The distribution of high-angle boundaries along the building direction (BD) (Figs. 2(c), 2(f)) corresponds to the morphologies of elongated γ grains before martensitic transformation (prior austenite grain boundaries22)). The morphology was not observed in the water-quenched sample (Fig. 2(i)), and the shape of the prior austenite grain boundaries suggests the formation of equiaxed γ grains at an elevated temperature of 1000°C. Note that the martensitic structure exhibited the Kurdjumov-Sachs orientation relationship22) ((111)γ//(011)α, [-101]γ//[-1-11]α).
Figure 3 shows a high-magnification SEM image of the L-PBF manufactured sample (P: 230 W, v: 1.67 m·s−1). The L-PBF manufactured sample exhibits an elongated cellular structure (cell spacing below 0.5 μm), most of which tends to be elongated in the building direction (BD) (Fig. 3(a)). The elongated direction of the cellular structure appeared to change at the melt pool boundaries, whereas some parts exhibited the same elongated direction across the melt pool boundaries. The cellular morphologies around the melt pool boundaries were almost the same as that inside the melt pools (Fig. 3(b)), whereas different phases (presumably corresponding to the fine retained γ phase) appeared locally due to etching using the Nital solution (as indicated by arrowheads in figures).
Figure 4 shows the DSC thermal profiles of the L-PBF processed samples and the water-quenched sample. In all samples, exothermic reactions were detected at temperatures of approximately 490°C and 610°C, and an endothermic reaction was detected at approximately 700°C. These reactions are considered to correspond to the two types of precipitation (exothermic reactions) and α→γ transformation (endothermic reactions) in heating. Therefore, the precipitation of intermetallic compound phases (presumably corresponding to the Ni3Ti and Fe7Mo6 phases by thermodynamic calculations22)) could occur prior to the α→γ transformation. The peak temperatures of each exothermic and endothermic reaction detected in the L-PBF samples were lower than those in the water-quenched sample, indicating that the precipitation sequence and the austenite reversion in the L-PBF samples occurred faster than those in the water-quenched sample. This result is in good agreement with the previous study.30) The peak temperatures of the L-PBF sample fabricated under a high energy density condition (P: 230 W, v: 1.67 m·s−1) were lower than those fabricated under a low energy density condition (P: 213 W, v: 2.08 m·s−1), suggesting that the kinetics of the precipitations and the austenite reversion during the heat-treatment would be controlled by L-PBF processing conditions.
Figure 5 summarizes the XRD profiles of the samples after the isothermal heat treatments at 550°C for different times (followed by water-quench). Not only diffraction peaks derived from the α phase but also minute diffraction peaks of the γ phase were detected from the L-PBF sample fabricated under the high energy density condition (P: 230 W, v: 1.67 m·s−1) (Fig. 5(a)). On the other hand, no diffraction of the γ phase was detected from the water-quenched sample and the L-PBF sample fabricated the low energy density condition (P: 213 W, v: 2.08 m·s−1) (Figs. 5(b), 5(c)). This is in good agreement with the results of the EBSD analyses (Fig. 2). In both samples, the diffraction intensities of the γ phase become higher at a longer time in the isothermal heat treatment, indicating that the α→γ transformation occurred during the holding at 550°C. There was no clear diffraction of the intermetallic compound phases (Ni3Ti or Fe7Mo6) corresponding to the precipitate phases. This would be due to low volume fractions of the precipitate phase. The XRD profiles obtained from each sample were used to quantify the γ phase using the reference intensity ratio method.31) The reference intensity ratio (RIR) values of the γ phase for each sample were summarized as a function of holding time. The result is shown in Fig. 6(a). In all samples, the RIR values of the γ phase increased with increasing the holding time. The RIR values of the L-PBF sample fabricated under the high energy density condition (P: 230 W, v: 1.67 m·s−1) increased from about 30% to about 80%, and then become almost saturated. This can be inferred from the fact that the volume fraction of the γ phase would reach an equilibrium state after 10 hours of holding at 550°C. The RIR value of the L-PBF sample fabricated under a low energy density condition (P: 213 W, v: 2.08 m·s−1) increases continuously, reaching about 80% after 100 hours in isothermal holding. On the other hand, the increase in the RIR value of the water-quenched sample appeared slower than that of the L-PBF manufactured sample and is approximately 25% even after 100 hours in isothermal holding. This indicates that the α→γ transformation of the L-PBF manufactured samples is faster than that of the water-quenched sample and is more pronounced under a high energy density condition. This is in good agreement with the result of the DSC measurements (Fig. 4).
Figure 6(b) shows the change in Vickers hardness of each sample with the holding time. The L-PBF manufactured samples exhibited higher hardness than the water-quenched sample, and this tendency is more pronounced for the L-PBF sample fabricated using a higher energy density condition. This is likely due to the fine precipitates formed by multiple reheating32,33) through the scanning laser irradiation process in the L-PBF process, and it is inferred that the higher energy density would promote the formation of the precipitated phase. The hardness of all samples increased to about 600 HV after 1 hour and decreased with increasing the holding time. Such age-hardening may be attributed to the precipitation of intermetallic phases inside the α-phase.
Figure 7 depicts the SEM images showing microstructures of the samples after the isothermal heat treatments at 550°C for different times (followed by water-quench). After 1 hour of holding, many elongated forms (bright contrast) with widths of approximately 1 μm were observed in the microstructures of the L-PBF sample fabricated under the high energy density condition (Fig. 7(a)). This tendency appeared more pronounced after the isothermal heat treatments for longer periods, and the samples that were heat-treated for more than 10 hours exhibited layered contrasts in whole microstructures (Figs. 7(b), 7(c)). This result was also found in the L-PBF sample fabricated under the low energy density condition (Figs. 7(d)–7(f)). After 1 hour of holding (Fig. 7(g)), the water-quenched sample showed a microstructural morphology corresponding to the initial martensitic structure, whereas the morphology changes significantly after 10 hours of holding (Fig. 7(h)).
Figures 8 and 9 present the phase distribution and crystallographic orientation maps obtained by EBSD analyses for the L-PBF samples and the water-quenched sample, respectively, after the isothermal heat treatments at 550°C for different periods. The area fraction of the γ phase in the L-PBF sample fabricated under the high energy density condition increased to 81% after 1 hour of holding, and fine α phases with a few micrometers in size appeared dispersed (Fig. 8(a)). The area fraction of the γ phase was measured to be almost 100% after more than 10 hours (Figs. 8(b), 8(c)). The transformed γ phase exhibited an elongated grain morphology with the <001> direction along the building direction (BD) (Figs. 9(a)–9(c)). This corresponds to the preferential growth direction of the γ phase (fcc structure) in solidification. Such a crystallographic feature was observed in other fcc metals as well.34) Therefore, the microstructure of the reversed austenite can be considered to reflect the characteristics of the solidified γ phase (microstructure before martensitic transformation in the L-PBF process). This feature was observed in the microstructure transformed from α to γ phases in conventional steels.35) Similar microstructural changes were also observed in the L-PBF sample fabricated under the low-energy density condition (Figs. 9(d)–9(f)), whereas the increase in the area fraction of the γ phase with holding time appeared slower, compared with the sample fabricated under the high energy density condition (Figs. 8(d)–8(f)). The change in the area fraction of the γ phase in the water-quenched sample was slower than in both L-PBF manufactured samples (Figs. 8(g)–8(i)). In the samples heat-treated for 10 and 100 hours, microstructures of the γ phase consist of relatively equiaxed grains (Figs. 9(h), 9(i)), which reflected the microstructure in the γ single-phase region at 1000°C. The above trend of the changes in the γ phase with the isothermal heat treatment corresponded well with the results of DSC (Fig. 4) and XRD (Fig. 6(a)) measurements.
Figure 10 shows SEM images and the corresponding EBSD analysis result for the L-PBF sample fabricated under a high energy density condition (P: 230 W, v: 1.67 m·s−1) and subsequently heat-treated for 1 hour. The relatively dark-contrast region (contamination part) corresponded to the EBSD analyzed area (Fig. 10(a)). In the high-magnification SEM image (Fig. 10(b)), many elongated forms (with a width below 1 μm) with different contrasts contrast were observed. From the crystal orientation distribution map and phase distribution map obtained by the EBSD analysis for the same region (Figs. 10(c), 10(d)), it is assumed that the elongated forms with a width of less than 1 μm would correspond to the α phase and the surrounding areas would be the γ phase (Fig. 10(b)). However, the EBSD-analyzed area fraction of the α phase tends to be underestimated because the area of the α phase in the SEM image would not be always identified as the α phase by the EBSD analysis (Fig. 10(d)). This is presumably due to numerous fine precipitates and high-density dislocations in the α-matrix phase (as discussed later), resulting in unclear EBSD patterns detected from the α phase compared to the γ phase. The present EBSD analyses can extract the changes in crystallographic features in the isothermal heat treatment at an elevated temperature, nevertheless the measured γ-phase fraction would be unreliable for quantification. Therefore, systematic EBSD analyses with smaller scanning steps would be necessary for precise quantification of the phase fractions by the EBSD analysis.
Figure 11 displays TEM bright-field images and a selected area electron diffraction pattern of the maraging steel samples and those subsequently heat-treated at 550°C. The L-PBF manufactured samples fabricated under both conditions (P: 230 W, v: 1.67 m·s−1 and P: 213 W, v: 2.08 m·s−1) exhibited a cellular microstructure with a width of less than 0.5 mm and containing high-density dislocations (Figs. 11(a), 11(b)). The microstructural morphologies corresponded to the cellular microstructure observed by SEM (Fig. 3). No clear diffraction derived from the γ phase was detected in the electron diffraction patterns captured from the local areas at cell boundaries. The retained γ phase was not observed in the limited areas in the TEM samples. It is noteworthy that the cellular microstructure was not observed in the water-quenched sample, as well as in the SEM image (Fig. 2). The water-quenched sample exhibited an elongated lath structure containing a high density of dislocations (Fig. 11(c)). In the L-PBF sample subjected to heat treatment for 1 hour, a number of the γ phases were observed along the cellular boundaries in the α matrix (Figs. 11(d), 11(e)). In addition, numerous fine needle-shaped precipitates were observed in the α matrix. An electron diffraction pattern (Fig. 11(e)) taken from the area (surrounded by broken lines) in the TEM image (Fig. 11(d)) presented not only diffraction spots corresponding to the (200) plane of the γ phase but also various diffraction spots derived from the η-Ni3Ti (D024 structure) phase.36) The phase identifications are consistent with the result of DSC measurements (Fig. 4), which is indicative of the occurrence of precipitation prior to the α→γ transformation. As indicated by the arrows in Fig. 11(f), the precipitated phases became coarsened after 100 h of holding. The γ phase grew and its volume fraction appeared to increase (Fig. 11(f)). The γ phase and the fine precipitates were also observed in the water-quenched sample and subsequently heat-treated for 10 and 100 h.
Figure 12 shows high-angle annular dark-field (HAADF)-STEM images and EDS elemental distribution maps of the L-PBF manufactured sample (P: 230 W, v: 1.67 m·s−1) and the water-quenched sample. Enrichments of Ni and Co alloying elements were detected along the cell boundaries in the L-PBF manufactured sample (Fig. 12(a)). Such a trend toward the enrichment of Ti and Al elements was observed, whereas these elements were enriched in particle-like morphologies inside the cellular structure. These enriched locations often coincided with the oxygen (O)-enriched regions, indicating that Ti and Al oxide particles with a size of several tens of nm were finely distributed. The O-free regions with Ti or Al enrichments can be considered to correspond to the precipitated Ni3(Ti, Al) phase. Such cellular microstructure and enrichment of alloying elements were also found in the L-PBF sample fabricated under a low energy density condition (P: 213 W, v: 2.08 m·s−1). On the other hand, alloying elements such as Ni and Co were uniformly distributed in the water-quenched sample (Fig. 12(b)). In addition, many fine oxide particles and precipitates were observed in the water-quenched sample as well. The relatively coarsened Ti-containing precipitates (Ni3Ti phase) are presumed to be the precipitates that existed in the L-PBF sample and remained after the holding at 1000°C.
Figure 13 shows the HAADF-STEM image and the corresponding EDS elemental distribution map of the L-PBF manufactured sample (P: 230 W, v: 1.67 m·s−1) that was subjected to heat treatment for 1 hour. An enrichment of Ni element was clearly found in the γ phases elongated along the cell boundaries within the α phase. Relatively high Co concentration was detected in the α phase. Not only fine oxides but also Ti-contained precipitates with relatively large sizes were observed. This result indicates the growth of the Ni3Ti phase during the isothermal heat treatment, which corresponds to the exothermic reactions of precipitation detected by the DSC measurement (Fig. 4). The tendency of Ti elements to be concentrated in the cell boundaries was also observed in the heat-treated samples.
This study investigated the microstructural evolution of maraging steels additive-manufactured using different L-PBF processing conditions (P: 230 W, v: 1.67 m·s−1 and P: 213 W, v: 2.08 m·s−1) during the isothermal heat treatments at 550°C, compared with the results of the water-quenched samples with the same composition. It was found that precipitation of intermetallic phases and subsequent α→γ transformation (austenite reversion) occurred in all samples and that the precipitation and austenite reversion occurred faster in the L-PBF manufactured samples than in the water-quenched sample. The onset of precipitation and phase transformation in the L-PBF sample fabricated under the high energy density condition (P: 230 W, v: 1.67 m·s−1) tended to be faster than that under the low energy density condition (P: 213 W, v: 2.08 m·s−1). In addition, fine cellular structures were observed in the interior of the α-phase matrix in the L-PBF manufactured samples, and concentrations of various alloying elements, such as Ni, were detected on the cell boundaries. The feature was not observed in the water-quenched sample. The cellular morphologies with inhomogeneous distribution of alloying elements can be considered to be formed through the local melting and following rapid solidification process by scanning laser irradiations in the L-PBF process. Such a cellular structure was often observed in Al alloys34) and Ni-based alloys37) manufactured by the L-PBF process.
In this study, to investigate the reaction pathway in the solidification of maraging steels, thermodynamic calculations for the alloy composition and solidification simulations under Scheil condition38) (assuming that complete mixing of a liquid phase, no diffusion in solid phases, and local equilibrium at the solid/liquid interface) were conducted based on the commercial thermodynamic database PanIron.39) Based on the results of the ICP-AES analysis (Table 1), the alloy composition of the maraging steel was determined to be a Fe–Ni–Co–Mo–Ti–Al system. The calculated results are summarized in Fig. 14. Figure 14(a) presents a cross-sectional phase diagram (with a composition range of Fe-xNi-9Co-5Mo-1Ti-0.1Al (mass%)) indicating an alloy composition of the experimental maraging steel. The alloy composition exhibited a γ single-phase region in an approximate temperature range from 1400 to 800°C. Below 800°C, the intermetallic phases containing Mo and Ti elements were in equilibrium with the γ phase. The calculated solidification sequence under Scheil condition (Fig. 14(b)) shows that the γ phase crystallizes from the liquid phase over a wide temperature range from 1050 to 1450°. The crystallization of the γ phase accounted for more than 95% in the solid phase through the solidification. Figures 14(c), 14(d) present the compositional changes of the γ phase and the liquid phase with solidification (increase in the solid-phase fraction). The concentrations of Ni and Ti elements in both γ and liquid phases increased with increasing the solid-phase fraction and appeared more pronounced above 80% in the fraction of solid phases. This trend is also observed for Mo and Al elements in the γ phase (Fig. 14(c)) but is not as pronounced as for Ni and Ti. These results indicate that high concentrations of Ni and Ti elements could be enriched in the γ phase near the finally solidified zones. This corresponds well to the enrichment of alloying elements localized along the cell boundaries (in the cellular structure) corresponding to the finally solidified zones (Fig. 12). Such cellular microstructure was observed in the L-PBF manufactured austenitic steels. It has been reported that the elongated direction of cellular microstructure followed the <001> orientation corresponding to the preferential growth direction in solidification.10) It can be therefore considered that the fine cellular microstructure in the L-PBF manufactured maraging steel (Fig. 4) could be the rapidly solidified microstructure of the γ phase resulting from the L-PBF process. In the present study, not only Ni and Ti but also Co elements were found to be enriched in the cell boundaries (Fig. 12). This fact cannot be rationalized by the calculated compositional change of the γ phase in solidification based on the local equilibrium at the solid-liquid interface (Fig. 14), since it has been reported that L-PBF processed metals and alloys often have supersaturated solid solutions containing highly concentrated solute elements beyond the solubility limits of metal-matrixs.40,41) This fact indicates that a local equilibrium could not be at the solid-liquid interface in solidification during the L-PBF process. It is assumed that the Co element might interact with Ni, Ti, and other elements in the non-equilibrium solidification process, resulting in the elemental distribution near the solid-liquid interface, but the details remained unclear. To elucidate the detailed mechanism experimentally, it is necessary to systematically investigate the inhomogeneity of the solidification microstructure and the concentration distribution of alloying elements in maraging steels solidified at different cooling rates.
When the L-PBF manufactured maraging steel sample with a fine cellular microstructure was subjected to an isothermal heat treatment at 550°C, the formation of an elongated γ phase on the cell boundaries was observed (Fig. 11(d)). The austenite reversion (formation of the γ phase) would be associated with the highly concentrated Ni element (Fig. 13). From these results, it is inferred that the formation of the γ phase could be due to phase transformation accompanied by the diffusion of Ni element (for stabilizing the austenite). The vicinity of the cell boundaries enriched in Ni element could play a role of preferential nucleation sites for the γ phase. These inferences are consistent with the slow α→γ transformation of the water-quenched sample with uniformly distributed alloy elements. In this study, the calculated phase diagrams for the alloy composition around the heat-treatment temperature were prepared to investigate phase equilibria at an elevated temperature of 550°C. The results are summarized in Fig. 15. Figure 15(a) shows a calculated isothermal section at 550°C of the Fe–Ni–Co ternary system for Fe-18Ni-9Co (mass%) corresponding to a major component of the experimental maraging steel. It is clear that Ni and Co elements can be partitioned into the γ and α phases, respectively. Figures 15(b), 15(c) present the variations of the equilibrium compositions of the α and γ phases with temperature in the multi-element system including all alloying elements. Based on the thermodynamic calculations, the partition coefficients (kγ/α) of the γ phase relative to the α phase for each element at 550°C were evaluated to be kNiγ/α = 8.3, kCoγ/α = 0.32, kTiγ/α = 17.5, and kMoγ/α = 0.18. Thus, even in the alloy composition of the experimental maraging steel, the Ni elements could stabilize the γ phase and be partitioned into the γ phase at 550°C. This was in good agreement with the experimental result of relatively high concentrations of Co in the α phase (Fig. 13). On the other hand, the Ti element (for stabilizing the γ phase) tended to concentrate on the cell boundaries even after the isothermal heat treatment at 550°C (Fig. 13), suggesting that Ti would slightly contribute to the growth of the γ phase. This is presumably because Ti would contribute to the formation and growth of precipitated phases such as the Ni3Ti phase. The aforementioned results can provide a conclusion that Ni element enriched in the cell boundaries (in the cellular structure) formed in the L-PBF manufactured maraging steel could play a significant role in nucleation sites for the γ phase at elevated temperatures, and promote the austenite reversion.
The cellular microstructure enriched with alloying elements such as Ni (Fig. 12(a)) was observed regardless of the L-PBF processing conditions. This result indicates a slight effect of L-PBF processing conditions on the cellular microstructure of the γ phase formed in rapid solidification. On the other hand, the austenite reversion behavior changes depending on the L-PBF processing conditions (Figs. 6(a), 8). The L-PBF sample fabricated with a high energy density (P: 230 W, v: 1.67 m·s−1) contained more retained γ phase than one fabricated with a low energy density (P: 213 W, v: 2.08 m·s−1) (Fig. 2) and show faster kinetics of α→γ transformation (Fig. 4). This is presumably due to the varied thermal profiles32,33) of the powder layer by continuous scanning laser irradiation process depending on the L-PBF processing condition. It is inferred that scanning laser irradiation under the high energy density condition could locally heat the sample and promotes the formation (or growth) of the γ phase inside the solidified microstructures (containing the retained γ phase). As a result, the distribution of retained γ phase in the L-PBF manufactured sample (Fig. 1(b)) would change depending on the L-PBF processing conditions. However, the detailed mechanism of the change in the distribution of the retained γ phase by the L-PBF processing condition, in particular, the preferential formation of the γ phase near the melt-pool boundaries (Fig. 2(b)), is currently unknown. SEM/EDS analyses for the vicinity of the melt-pool boundaries were performed, nevertheless, no enrichment of alloy elements was detected. In order to control the α→γ transformation utilizing the L-PBF process conditions, precise monitoring of the local thermal histories associated with continuous scanning laser irradiation and further analyses for the elemental distribution across the melt pool boundaries are required.
In this study, we investigated the microstructural variations in maraging steel samples additive-manufactured using different conditions of the L-PBF process (P: 230 W, v: 1.67 m·s−1 and P: 213 W, v: 2.08 m·s−1) by isothermal heat treatment at 550°C and compared the results with those of the austenitized and water-quenched sample with the same composition. The experimental results were discussed utilizing thermodynamic calculations, and the following conclusions were drawn.
(1) The L-PBF manufactured samples exhibited a fine cellular structure inside the martensitic microstructure. Enrichment of alloying elements was detected near the cell boundaries. The enrichments of alloy elements would be caused by local compositional changes in the γ phase due to the rapid solidification process in the L-PBF process. Such an inhomogeneous elemental distribution in the martensitic microstructure was not observed in the water-quenched sample.
(2) Precipitation of intermetallic phases followed by α→γ transformation (austenite reversion) occurred in both L-PBF manufactured and water-quenched samples. Precipitation and austenite reversion occur faster in the L-PBF manufactured samples than in the water-quenched sample, and their kinetics varied depending on the L-PBF processing condition. Ni element was concentrated in the cell boundaries formed in the L-PBF manufactured samples. The localized enrichment of Ni could play a role in nucleation sites for the γ phase at high temperatures and promote the austenite reversion. In addition, it is inferred that the variations in the thermal profile caused by continuous scanning laser irradiation under the different L-PBF processing conditions might contribute to the formation behavior of the retained γ phase in the solidified microstructure of the L-PBF manufactured samples.
The support of “Knowledge Hub Aichi” Aichi Prefectural Government (Japan) and JFE 21st Century Foundation (Japan) was gratefully acknowledged.
