Modulated Structure Formation in Dislocation Cells in 316L Stainless Steel Fabricated by Laser Powder Bed Fusion

Fei Sun; Toshio Ogawa; Yoshitaka Adachi; Kazuhisa Sato; Shunya Takagi; Goro Miyamoto; Asuka Suzuki; Akinori Yamanaka; Nobuo Nakada; Takuya Ishimoto; Takayoshi Nakano; Yuichiro Koizumi

doi:10.2320/matertrans.MT-ME2022004

Abstract

Metal additive manufacturing enables producing complex geometric structures with high accuracy and breaks the design constraints of traditional manufacturing methods. Laser powder bed fusion, a typical additive manufacturing process, presents a challenge in experimentally understanding the nano-scaled microstructure-process relationship regarding the wide range of process parameters. In this study, we aim to reveal the novel nanoscale structural features by advanced scanning transmission electron microscopy to clarify the formation mechanisms in 316L stainless steel by laser powder bed fusion. Here we show that the slender columnar grains were confined to the centreline of the melt pool along the build direction, and the columnar cell structure at the side branching of the melt pool grew along orthogonal directions to follow drastic changes in thermal gradient across adjacent melt pools. Novel nano-scaled modulated structures have been observed in the dislocation cells parallel to the laser scan direction, which were mainly caused by the elastic strain involving the thermal gradient inside the melt pool and across adjacent melt pools as well as the effective strain field in the dislocation cell interiors. An in-depth understanding of microstructure developments is worthy of fabricating high-performance materials by controlling the additive manufacturing process.

1. Introduction

For the past decade, laser powder bed fusion (LPBF) has been attracting great research attention among the existing additive manufacturing (AM) techniques for metallic materials due to the possibility of near-net-shape parts with complex geometries.¹⁾ The LPBF involves a layer-by-layer iterative process utilizing a high-power laser to selectively melt and fuse the pre-laid powders with thermal gradients in the range of 5–20 K/µm and cooling rates in the range of 1–40 K/µs.²⁾ The LPBF presents complex solidification microstructures with unique features compared to cast and wrought counterparts. The LPBF allows the alteration of many process parameters, including laser power, scanning speed, scan strategy, layer thickness, powder size, and so on. These available processing parameters and their wide range of options or values make it possible to fabricate materials involving complicated defects and microstructures.³⁾

Austenitic 316L stainless steel (316L SS) has a wide range of industrial applications through well-known manufacturing techniques consisting mainly of casting, forging, and extrusion. These techniques have limited degrees of freedom, and the manufactured metal parts with complex shape geometries require post-processing, such as machining, which is accompanied by a considerable waste of materials and time. The LPBF-processed 316L SS has shown a superior combination of high strength and ductility that is not typically observed in conventionally-processed 316L SS. The LPBF-processed unique microstructures can overcome the trade-off between tensile strength and ductility regarding the fine-grained microstructure, additional small-scale compositional variations, the formation of a large number of melt pools with a size of tens of micrometers containing columnar grains where regular dislocation cell structures are embedded, and the presence of an unintended dispersion of nano-inclusions in the sub-grain structures. Although some investigations have been contributed to revealing the process-structure-property interrelationship for LPBF-processed and heat-treated 316L SS, it is still not fully understood. More specifically, an in-depth understanding of this threefold interrelationship at multiple length scales remains among the prominent scientific challenges in metal AM.¹^,⁴^–¹⁴⁾

Cellular structures decorated with elemental segregation at the cellular boundaries have been reported in many fcc LPBF-processed alloys. The most concern about the cellular structures involves the following three aspects: (1) the formation mechanism regarding elemental segregation and dislocation cells, and its features, such as the morphology, size, growth direction, and thermal stability, (2) the effect on the novel sub-micro cellular structure on the beneficial or detrimental trends of the mechanical and corrosion aspects of LPBF-processed alloys, and (3) the possibilities for designing advanced metal AM materials by controlling the cellular structure.⁶^,¹⁵^–²⁶⁾ However, the fine microstructure in the dislocation cells has been rarely reported. An in-depth understanding of microstructure development is required to fabricate high-quality AM products.

The present work aims to contribute to the further understanding the microstructure-process relationship of as-built 316L SS by LPBF. To this end, delicate structures that appeared in the dislocation cell interiors as a novel phenomenon were evaluated by the advanced characterization technique on the nanoscale. Further analyses about its underlying mechanism were performed responsibly for the microstructure development in connection with the LPBF process.

2. Experimental Procedure

The nominal composition of the 316L SS powder was 18Cr–14Ni–2.5Mo–0.03C (wt.%), and the powder size was under 53 µm. LPBF fabrication was conducted using a 3D printer (EOSM290, EOS GmbH, Germany) equipped with a Yb-fiber laser by the “X-scan strategy”, i.e., the laser beam was scanned bidirectionally along the X-axis without rotation. The X-scan strategy has been clarified to be beneficial in producing a single crystalline-like texture.²⁷⁾ A basic set of the LPBF process parameters was selected for generating the studied specimen: laser power (P) 250 W, laser scan speed (v) 1000 mm/s, laser hatching distance (h) 80 µm, and layer thickness (t) 40 µm. All the analyses in this study were performed in the as-built specimen without any heat treatment. The samples were cut from the cross-sections containing the build direction (BD), laser scan direction (SD), and transverse direction (TD) and then grinded and polished using an automatic polishing machine (PRESI, Mecatech 250 SPI). The microstructure was characterized after etching in a solution of nitric and hydrochloric acid (HNO₃:HCl:H₂O = 1:10:10) by scanning electron microscopy (SEM, JSM-7001F) and electron backscatter diffraction (EBSD) technique. The image quality map (IQ), inverse pole figure (IPF) orientation map, and kernel average misorientation (KAM) map regarding the local grain misorientation were derived from EBSD data. Transmission electron microscope (TEM) specimens were prepared using a focused ion beam (FIB) instrument (Scios2 Dual Beam, Thermo Fisher Scientific, Hillsboro, OR, USA) with Ga ions. TEM observations were carried out on JEM-2100PUS (JEOL Ltd., Tokyo, Japan) and JEM-ARM200F (JEOL Ltd., Tokyo, Japan). Compositional analysis was performed in high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) mode using energy-dispersive X-ray spectrometry (EDS) attached to the 200 kV JEOL JEM-ARM200F.

3. Results

3.1 Columnar structure and cellular boundary

Figure 1 shows the 3D visualization of the as-built microstructure of LPBF-processed 316L SS based on the SEM images observed along BD, SD, and TD. In the BD-TD cross-section, some melt pool boundaries were highlighted in yellow, which can be observed clearly. At the central part of the melt pool, the dark contrast region was distributed along BD and periodically along TD. In the SD-TD plane, the laser scan traces were displayed obviously with a certain hatch distance. The crystal orientations of the SD-TD and SD-BD planes caused by the X-scan strategy have already been analyzed by the EBSD measurement in the previous study.¹⁴⁾

Fig. 1

3D visualization of the as-built microstructure of LPBF 316L SS by SEM observation. BD, SD, and TD correspond to build, laser scan, and transverse directions, respectively. Some melt pool boundaries are outlined in yellow.

Figure 2 shows the crystallographic textures of the TD-BD (YZ) cross-section plane of the as-built 316L SS specimen. Figure 2(a) shows the IQ map with the misorientations of the grain boundaries, creating valuable visualizations of the solidification microstructure. Generally, the IQ map could provide superior grain boundary and strain contrast. In comparison, the overlapped misorientation map provides good consistent distribution with the dark contrast region regarding the large-angle and low-angle grain boundaries. The large-angle grain boundaries with misorientation angles above 15° were mainly distributed at the central part of the melt pool along BD, and the low-angle grain boundaries with misorientation angles ranging from 2°–15° were primarily distributed at the side-branching along orthogonal directions in response to follow the local heat flux.²⁸⁾ Figure 2(b) shows the KAM map in the whole TD-BD (YZ) plane displaying misorientation, which can reflect dislocation density caused by plastic deformation. It is evident that the central part of the melt pool along BD has severe grain misorientation with a high level of dislocation densities and local misorientations on the order of 0°–5°. Figure 2(c) shows the IPF image of the TD-BD (YZ) cross-section plane with the main texture {001} orientation due to the primary red color in the SD direction and the slender parts with other complex orientations due to the various colors (e.g., yellow, green, blue, etc.) indicating specific orientations compared to the matrix. The unique slender microstructure at the central part of the melt pool has been referred to as crystallographic lamellar microstructure (CLM).¹⁴⁾ The CLM along BD with [110] orientation periodically coincides with the laser hatching distance (80 µm). Figure 2(d) shows the zoom-in IPF map of the local region marked by a black rectangle in (c). Columnar grains can be observed clearly in the CLM at the central part of melt pools. The long black arrow indicates the growth direction of the columnar grains along BD. At both sides of the slender central parts, the side-branching indicated by the short black arrows with slight misorientations in SD can be observed along an incline of approximately ±45° to the BD. According to the crystal orientation maps shown in the lower part, the CLM part and all the side-branching have {001} orientations.

Fig. 2

EBSD maps of the BD-TD (YZ) cross-section plane of LPBF 316L SS. (a) IQ map with the misorientations of the grain boundaries, (b) KAM map, (c) IPF where red, green, and blue colors indicate {001}, {101}, and {111} orientations, (d) zoom-in IPF map of the black rectangle in (c) and the lower crystal orientation maps of the corresponding central and side branching parts of the melt pool.

Figure 3(a) shows the SEM image of the solidification microstructure of the melt pool containing the melt pool boundary, the central CLM part, and the side-branching columnar cells, which grew along [010] and [100] crystallographic directions, inclined about 45° with respect to the BD. Figure 3(b) shows the corresponding IPF indicating the columnar grains with various crystallographic orientations in the CLM at the central part of melt pools. The columnar cells at the side branching have a similar {001} crystallographic orientation. There is no obvious misorientation across the melt pool boundary (MPB) marked by black curves. Figure 3(c) shows the bright field (BF) TEM image obtained along the [100] zone axis of the columnar cell region on the TD-BD (YZ) plane containing the columnar cells and cellular boundaries with a high density of entangled dislocations. There still exists some extent of dislocation inside the columnar cells. The dislocation networks consisting of the cellular boundaries and the columnar cell interiors in 316L SS have been widely reported in the literature.⁶^,⁷^,¹⁶⁾

Fig. 3

Solidification microstructure in a multi-layer built of SUS 316L by LPBF along BD. (a) SEM secondary image of melt pool boundary (MPB) and the columnar cells, (b) corresponding IPF of (a), (c) TEM BF image of the columnar cells along the [001] zone axis on the BD-TD (YZ) plane with a high density of dislocations as the cellular boundaries.

3.2 Fine structure in dislocation cell interiors

Figure 4 shows the HAADF-STEM (Z-contrast) image of the dislocation cell structure on the SD-TD (XY) plane, taken with the incident beam parallel to the [110] direction. The bright contrast in the HAADF-STEM image indicates strong elemental segregation at the dislocation cell boundaries. The corresponding EDS maps confirm the Mn, Cr, Mo, Si segregation and Fe depletion at the cell boundaries. Some particles with the dark contrast in the HAADF-STEM image indicate that they were composed of some elements lighter than Fe. The EDS maps show that O, Mn, and Si were enriched in those particles. Similar oxide inclusions have commonly been reported in AM austenitic stainless steels with tens of nanometers and a few microns.⁵^,⁷^,²⁹^,³⁰⁾

Fig. 4

HAADF-STEM image of the dislocation cell structure on the SD-TD (XY) plane, taken with the incident beam parallel to the [110] direction and corresponding EDS maps of C, O, Mn, Si, Cr, Mo, Fe, and Ni.

Figure 5(a)–(d) show the TEM bright field (BF) images of the dislocation cell structure on the SD-TD (XY) plane along the electron incidence direction of [110] and two beam diffraction conditions with imaging g vectors $[\bar{1}13]$, [002], and $[1\bar{1}1]$, respectively. Interestingly, modulated contrast can be observed along [001] direction parallel to SD according to the indexed diffraction pattern in the dislocation cell interior and through the cell boundaries, as shown in Fig. 2(c). The modulated structure was invisible under a specific two-beam condition with imaging g vector [002]. Figure 5(e) shows the HAADF-STEM image of the corresponding dislocation cell containing the modulated structure. The contrast of the HAADF-STEM images is associated with the atomic number of the elements in the specimen (Z contrast) and the variations in the crystal structure (i.e., diffraction conditions). Figure 5(f) and (g) show the high magnification BF-STEM and HAADF-STEM images with obvious contrast. Figure 5(h) shows the TEM BF image at the micrometer scale, indicating the modulated structure keeps the same orientation with a long-range distance through the dislocation cell boundaries and grain boundaries.

Fig. 5

TEM BF images of the dislocation cell structure on the SD-TD (XY) plane along various beam directions: (a) [110] on-axis, (b)–(d) two-beam conditions with imaging g vectors $[\bar{1}13]$, [002], and $[1\bar{1}1]$, respectively, (e) HAADF-STEM image of the dislocation cell structure and the obvious modulation contrast with imaging g vector $[1\bar{1}1]$, (f) and (g) high magnification BF-STEM and HAADF-STEM images of the modulated structure with imaging g vector $[2\bar{2}0]$, (h) long-range of the modulated structure across dislocation cell structures and grain boundary with imaging g vector $[1\bar{1}1]$.

Figure 6(a) shows the HAADF-STEM image of the modulated structure in the dislocation cell on the SD-TD (XY) plane with obvious contrast. Figure 6(b) shows the zoom-in local region containing the EDS line profile position marked by a yellow rectangle. The quantitative EDS line profiles along the yellow arrow direction show slight elemental fluctuations. Combined with the accurate positions and the elemental fluctuations, the bright contrast regions marked by dotted lines involve a little enrichment of Fe and a depletion of Cr and Mo. In contrast, the dark contrast regions marked by dashed lines slightly involve a depletion of Fe and an enrichment of Cr and Mo.

Fig. 6

(a) HAADF-STEM image of the modulated structure in the dislocation cell structure on the SD-TD (XY) plane, (b) enlarged image (upper) of the local region in (a) indicating the EDS line profile position (yellow rectangle) and the quantitative EDS line profiles along the yellow arrow direction (lower).

Figure 7(a) shows the high-resolution TEM (HRTEM) image of the local region, indicating the bright/dark contrast close to the modulated structure interface in the dislocation cell interior on the SD-TD (XY) plane, taken with the incident beam parallel to the [110] direction, and the inserted corresponding fast Fourier transform (FFT) pattern. Figure 7(b) shows the inverse FFT (IFFT) image of the white square region acquired by masking the fcc matrix reflections in the inset FFT pattern. Applying four sets of filter masks on the $2\bar{2}0$, 002, $1\bar{1}1$, $\bar{1}11$ spots, the relative Bragg filtered lattice fringe images of the white square region in Fig. 7(a) are given in Fig. 7(c)–(f), respectively. In Fig. 7(c)–(f), the IFFT images show the extra half-planes of atoms at each dislocation core, indicating edge dislocations. The blue lines show the stacking faults induced by the two opposite extra half-planes of atoms between the two partial dislocations. In addition, lattice distortion can be observed in the local region using FFT and IFFT for further analysis since it is difficult to observe directly by HRTEM. Figure 7(g)–(i) show the elastic strain maps of the whole high-resolution TEM image shown in Fig. 7(a), which were measured by the geometrical phase analysis software.³¹⁾ The strain maps in a left-to-right direction indicate the horizontal (ε_xx), vertical (ε_yy) and shear (ε_xy) strain maps, respectively. The two strong yet noncollinear reflections in the Fourier transform of the HRTEM image are masked with red circles in the FFT pattern in Fig. 7(a). The intensity scale bar represents the strain range from −10% (compressive) to 10% (tensile). In Fig. 7(h), the compressive and tensile vertical strains in the white square were larger than those along the horizontal direction shown in Fig. 7(g) and the shear strain shown in Fig. 7(i).

Fig. 7

(a) High-resolution TEM image of the local modulated structure in the dislocation cell interior on the SD-TD (XY) plane, taken with the incident beam parallel to the [110] direction and the inserted corresponding FFT pattern, (b) IFFT image of the write square region acquired by masking the matrix reflections in the inset FFT pattern, (c)–(f) IFFT lattice fringe images of the write square region acquired by applying 4 sets of filter masks on the $2\bar{2}0$, 002, $1\bar{1}1$, $\bar{1}11$ spots, respectively. The blue lines indicate the stacking faults. Corresponding maps of (g) horizontal (ε_xx), (h) vertical (ε_yy) and (i) shear (ε_xy) strain. The intensity scale bar represents the strain range from −10% (compressive) to 10% (tensile).

4. Discussion

The super large temperature gradients induced by the rapid heating and cooling processes can generate high levels of internal stress and thermodynamically metastable state, which inevitably cause complex microstructure. The origin of dislocation structures in AM 316L SS has been proposed by controlling the effect of thermal stress through geometric constraints on the fabricated samples.³²⁾ Dislocation structures in AM 316L SS can originate due to thermal distortions during printing, primarily dictated by constraints surrounding the melt pool and thermal cycling, but cannot be originated by the constitutional stresses due to micro-segregation, coherency strains due to precipitation networks, and misorientations between dendrites. Regarding micro-segregation in the dislocation cell, the mechanism has been proposed that chemical micro-segregation occurs to dislocation walls after the dislocation cell formation due to the thermal stresses.³³⁾ The dendritic nature of the cell networks suggests that chemical micro-segregation occurs during solidification. In this study, the modulated structure has been preliminarily observed as a novel phenomenon in the dislocation cell structure along [001] direction according to the HAADF-STEM image and EDS mapping. Suppose that the modulated structures were dislocations. Generally, a classic approach is to use g · b contrast, wherein dislocations cause little diffraction contrast when the diffraction vector g is perpendicular to the Burgers vector b, i.e., when the invisibility criterion g · b = 0 is met. The Burgers vector is determined from the two g-vectors that satisfy the invisibility criterion. It is still impossible to identify the dislocation Burgers vectors based on the two beam imaging conditions, as shown in Fig. 5(b)–(d). To further analyze the modulated structure, the HRTEM image was taken to observe the atomic structure, as shown in Fig. 7(a). Usually, the dark contrast in TEM images involves defects or strain, such as dislocations, lattice distortion, vacancies, etc. Dislocations can cause severe local distortions of the surrounding crystals, cause local strain, and promote the gathering of vacancies to lower the energy. The FFT and IFFT analyzed HRTEM image in Fig. 7(c)–(f) show the lattice fringes by applying the filter masks on the diffraction spot, which can show the lattice distortion and the extra half-plane of atoms in case of dislocation exists. In addition, the strain maps in Fig. 7(g)–(i) provide the strain distribution behavior with the variation from −10% to 10%, exhibiting the strain gradient according to the colored bar of the intensity scale. In particularly, the strains along the vertical direction existed much stronger than those along the horizontal direction. According to the FFT pattern, the horizontal direction is almost parallel to the [001] direction. It means that the modulated structures formed along the [001] direction were mainly caused by the strain arising from $[1\bar{1}0]$ direction on a two-dimensional level, which can be considered parallel to the SD-TD (XY) plane. This strain analyze results brings into correspondence with the invisible of the modulated structure under two beam condition with imaging g vector [002] shown in Fig. 5(c). In addition, more TEM specimens have been made perpendicular to the SD-TD (XY) plane and parallel to the SD-BD (XZ) plane. The same modulated structures have been observed exactly owning the similar distribution direction. Figure 8 shows the schematic of the formation behavior of the modulated structure concerning the macro-scale of the melt pool. The modulated structure exists along [001] direction parallel to the SD on the SD-TD (XY) and SD-BD (XZ) planes. The formation mechanisms may be related to the thermal gradient inside the melt pool and across adjacent melt pools, and also by the microstructure features in the dislocation cells, involving the effective strain field in the interiors and the accumulated strain by dislocation accommodation at the cell boundaries during the heating-cooling cycles in the melt pool. Further extensive analysis of the modulated structure in the dislocation cell interior is under processing with various LPBF process parameters, which is valuable for fully understanding the microstructure-process relationship.

Fig. 8

Schematic diagram of the formation behavior of the modulated structure concerning the macro-scale of the melt pool.

5. Conclusion

In this study, the nanoscale microstructural features were investigated in the as-built 316L SS by LPBF. The following findings were obtained.

(1) The slender columnar grains were formed in the centreline of the melt pool along the build direction, and the columnar cell structure with high-density dislocations as cellular boundaries grew along orthogonal directions at the side branching of the melt pool to follow drastic changes in thermal gradients across adjacent melt pools.
(2) The formation of the modulated structures along [001] direction parallel to the SD on the SD-TD (XY) and SD-BD (XZ) planes was mainly caused by the strain arising from its vertical direction, which may relate to the thermal gradient inside the melt pool and across adjacent melt pools and the effective strain field in the dislocation cell interiors, and the accumulated strain by dislocation accommodation at the cell boundaries during the heating-cooling cycles in the melt pool.

Acknowledgments

This study was supported by Grant-in-Aid for Transformative Research Areas “Creation of Materials by Super Thermal Field” Research (21H05194 and 21H05196). All authors sincerely acknowledge this support. The corresponding author (F.S.) also gratefully acknowledges the partial financial support provided by the Ministry of Education, Culture, Sports, Science and Technology through a Grant-in-Aid for Early-Career Scientists 20K15057 (2020–2022).

REFERENCES

Corresponding author

Register with J-STAGE for free!