Preprint for: SEM-DIC comparative study of plastic deformation evolution in AISI 304L fabricated by two different methods

SEM-DIC comparative study of plastic deformation evolution in AISI 304L fabricated by two different methods Dalibor Pavelčík1,2, Kateřina Neumannová1,3, Vojtěch Bartošík1,2, Daniel Koutný4, Michal Jambor1 and Miroslav Šmíd1 1Institute of Physics of Materials, Czech Academy of Sciences, Žižkova 513/22, 602 00, Brno, Czech Republic 2Institute of Material Sciences and Engineering, Faculty of Mechanical Engineering, Brno University of Technology, Antonínská 548/1, 601 90, Brno, Czech Republic 3Department of Theoretical Physics and Astrophysics, Masaryk University, Kotlářská 2, CZ-611 37 Brno, Czech Republic 4Institute of Machine and Industrial Design, Faculty of Mechanical Engineering, Brno University of Technology, Technická 2896/2, Brno 616 69, Czech Republic Abstract With increasing demands on the materials performance, new production methods are pursued, such as additive manufacturing (AM), which are generally leading to complex internal structures. Therefore, various non-destructive in-situ methods, for example digital image correlation (DIC), have been developed for deeper understanding of the processing – microstructure – properties relationship. Within this work, interrupted tensile test experiments on two AISI 304L steel specimens produced by two different methods, conventional hot-rolled and laser powder bed fusion (LPBF) processed, were performed. The experiment consists of simultaneous observation of the plastic deformation evolution by SEM-DIC method. The evolution of crystallography was followed by EBSD from the same area. Additively, high resolution backscatter electron diffraction (HR EBSD) was performed on selected grains to follow stress distribution and local crystal lattice orientation evolution via Kernel average misorientation (KAM) map. Test interruptions for measurements were done at 1.5, 3, 4.5, and 10 % of total strain. The collected data enabled the identification of active slip systems, which enabled to obtain a comprehensive insight into the plastic deformation evolution. This approach was used for both test specimens to compare the response of differently produced structures of the very same material. Significant heterogeneity of plastic deformation based on the orientation of the crystal lattice can be observed within individual grains of the LPBFprocessed microstructure. In contrast, in the case of hot-rolled microstructure, the plastic deformation appears more homogeneously distributed. Keywords: Plastic deformation; in-situ tensile test; digital image correlation; slip systems 1. Introduction With the increasing demands for the performance of machine components and assemblies, there has been a significant development of both new materials and new manufacturing processes in recent years. Austenitic stainless steels, such as AISI 304 L, are widely used in many sectors such as the food and chemical industry, transport or architecture due to their combination of excellent mechanical properties, chemical resistance and stability. However, modern manufacturing methods and sophisticated materials can cause an increasing complexity in their internal microstructure. This progress is often difficult to capture with traditional material testing methods using postmortem specimen analysis. For this reason, new methods are also beginning to gain traction, capable of describing in more detail the relationship between the investigated microstructure and its mechanical properties. For example, Digital Image Correlation (DIC) method is able to observe the plastic deformation nature at an individual grains level. In this study, an interrupted tensile test experiment was performed on two AISI 304 L steel specimens produced by two different methods – hot-rolled and LPBF-produced. After reaching a certain value of total deformation, within each interruption of the test, the level of deformation in the loading axis was measured using the DIC method. At the same time, the development of the material microstructure was investigated using the HR-EBSD method. 2. Experimental details 2.1 Material The study was conducted on two variants of AISI 304L austenitic stainless steel, hotrolled and LPBF-processed. Chemical composition of both variants is shown in Table 1. The production details of hot-rolled sheet with subsequent heat-treatment were described in detail in the previous study (Šmíd et al., 2021). Similarly, the LPBF fabrication is described in more detail in the earlier study (Šmíd et al., 2023). The LPBF-processed 304L was built in the form of a horizontal dogbone-shaped block, which was subsequently cut horizontally by electrical discharge machining on tensile specimens. The hot-rolled specimen was sectioned from the sheet along a plane containing the rolling direction and the normal direction. The flat dogbone specimen geometry is shown in Figure 1a. Tab. 1. Chemical composition of investigated 304L steel (in wt%). C Mn S P Si Ni Cr N Hot-rolled 0.023 1.79 0.003 0.04 0.17 8.18 18.12 0.086 LPBFprocessed 0.03 1.4 0.004 0.027 1 8.18 19.41 0.15 Both specimens were manually ground using SiC abrasive papers with increasing grit size, followed by mechanical polishing with a series of diamond pastes with decreasing particle size. As the last preparation step, electrolytical polishing was carried out by a solution of 58.8 % of methanol, 35.3 % of ethylene glycol monobutyl ether and 5.9 % of perchloric acid under voltage of 35 V for 40 s at temperature of 15 °C. Subsequently, a dense and homogenously distributed pattern for DIC imaging was needed to be introduced on the specimen surface. For that purpose, an OPS polishing colloidal solution was used on both 304L variants. A small amount of the suspension was applied to a polishing cloth soaked in deionized water. Afterwards, the specimens were gently polished for 30 s at plate speed of 150 revolutions per minute. The specimens were then rinsed for 8 s and allowed to dry naturally. Successfully distributed SiO particles, creating a suitable DIC pattern, is shown in Figure 1b. 2.2 Methodology The tensile testing was performed using a ZWICK Z50 testing machine in strain control regime. For that purpose, a Multisens extensometer was used to enable reaching the test interruptions at desired total strain levels (1.5, 3, 4.5, and 10 %). The specimens were strained at displacement rate of 0.6 mm/min. After unloading, the subsequent analysis of deformation evolution was performed in the selected region by DIC and EBSD techniques. The tensile specimens were studied using a TESCAN LYRA 3 XMU field emission gun scanning electron microscope (SEM). DIC imaging was performed by an in-lens SE detector with electron beam emitted by voltage of 5 kV. For the hot-rolled specimen, a region consisting of 9 images, each with a field of view 100 µm aligned in a 3x3 grid with image overlap of 20 %, was acquired. Therefore, the total analysed DIC area was 260 µm wide. For the LPBF-processed specimen, a grid of 16 images in a 4x4 grid, each with an 80 µm field of view, using overlap of 20 %, was acquired. The total area of interest was 272 microns wide. All images were taken at a resolution of 4096x4096 pxs. EBSD measurement, using electron beam emitted at 20 kV voltage of undeformed and deformed states was done by an EBSD Symmetry detector (Oxford Instruments, UK) and AZtec software. DIC analysis was performed with the open-source Matlab add-on Ncorr (Blaber et al., 2015). The following parameters were used in the DIC analysis: subset radius of 20 pixels with spacing of 3 pixels and strain radius of 3 pixels. The DIC maps were plotted on the reference image and did not contain information about the grain boundaries. These were subsequently obtained by EBSD from the undeformed material and added to the finished maps (Di Gioacchino & Quinta da Fonseca, 2013) The calculation of elastic stresses was carried out by HR-EBSD method using CrossCourt4.5 software (BLGVantage). For that purpose, small areas on both specimens were analysed by EBSD detector with resolution of 1024x1024 pixels while recording all electron backscattered patterns for the subsequent analysis. Using electron beam held at 15 kV, the HR-EBSD analysis was carried out for test interruptions at 0, 1.5, 3, and 4.5 % of total strain while using reference points located approximately at similar sites of each grain. For this contribution, we selected only a single grain for each material variant. 3. Results 3.1 Underformed structure Inverse pole figure (IPF) maps, presented in projection along loading direction (axis X), of undeformed structures are shown in Figure 2a and 2b, presenting distinct difference in grain size. Hot-rolled 304L presents a typical microstructure after conventional hotrolling, featuring polyhedric grains with frequent annealing twins. A fine-grained microstructure was found in the LPBF structure, while the grain morphology is notably more complex. Based on previous detailed characterization of both microstructural variants (Šmíd et al., 2021; Šmíd et al., 2023) high-angle grain boundaries are the most frequent type. The difference can be found in twin interface frequency, which is higher in the hot-rolled variant. Both microstructures are nearly random in terms of crystallography, except for a mild (011) texture along the build direction, which is characteristic of LPBF microstructures. 3.2 Tensile testing and DIC/EBSD analysis The diagram comparing stress-strain curves of both tensile tests is shown in Figure 2c. The striking difference in strength was apparent and manifested by a yield strength increase from 236 MPa to 592 MPa from hot-rolled to LPBF-processed specimen, respectively. Despite such an increase in strength properties, the latter one still retained very good ductility. On the other hand, the slope of the curve of the plastic part indicates higher work-hardening in hot-rolled material. The diagram inset shows the initial part of the tensile tests with denoted test interruptions. Figures 3 and 4 depict IPF X and DIC axial strain maps of both microstructural variants obtained during tensile test interruptions at 4.5 and 10 % total strain. While the IPF maps change shape due to tensile loading, the DIC maps retain the same square shape as strain fields are plotted into the reference images acquired before loading. The results of both test interruptions present the strain localization evolution and noticeable changes in lattice orientations of individual grains, which were distinct especially in the hot-rolled microstructure due to significantly larger grains. A typical lattice rotation trend towards 〈001〉 and 〈011〉 directions could be seen in both microstructures. In the case of the hot-rolled specimen, a mild and more diffused Fig. 1. IPF X maps of undeformed structure - a) hot-rolled specimen and b) LPBF-processed specimen. c) Stress-strain curves of carried out tensile tests, the detail highlighted within the graph shows the individual test interruptions for DIC and EBSD acquisition. Fig. 2. DIC and IPF maps characterizing the investigated areas of the hot-rolled specimen - a) and b) - and the LPBF-processed specimen - c) and d) - at 4.5 % strain. a), c) IPF X maps with the DIC analysis area highlighted. b), d) DIC maps of the εxx deformation component. distribution of plastic deformation into large number of slip bands was observed, see Figures 3b and 4b. Individual grains deform in a slightly different way depending on their crystallographic orientation with respect to the loading direction. Intergranular interaction was observed rather rarely, most likely due to the random nature of the hotrolled microstructure crystallography. The apparent increase in frequency of active slip bands and strain localization intensity with an increase to 10 % strain was observed. In the case of the LPBF-processed specimen, Figures 3d and 4d, the situation was slightly different. The evolution of grain orientation was not clearly discernible from the IPF X map due to the fine nature of the microstructure. However, in the case of coarser grains, deformation twins nucleation was detected already at 4.5 % deformation by SEM-DIC analysis. The strain localization is more pronounced, since a lower number of active slip systems per grain were observed. Strain magnitude in individual slip bands was higher, intergranular interactions were more intensive, thus slip transfer into neighbouring grain was more frequent. This led to appearance of grain clusters with intensive slip activity surrounded by grains with low level of plasticity. Therefore, the overall character of the strain distribution was more heterogenous in comparison with hot-rolled microstructure. The strain distributions, shown in Figure 5, were integrated from the entire DIC areas for both investigated load steps. The fine-grained LPBF-processed microstructure provided larger data set in terms of grain number, which is reflected by smoother shape of the strain distribution. Contrary to that, coarse hot-rolled microstructure resulted in irregular shape of strain distribution. At 4.5 % strain, the strain distribution of hot-rolled microstructure Fig. 4. DIC and IPF maps characterizing the investigated areas of the hot-rolled specimen - a) and b) - and the LPBF-processed specimen - c) and d) - at 10 % strain. a), c) IPF X maps with the DIC analysis area highlighted. b), d) DIC maps of the εxx deformation component. was more compact than the distribution of LPBF-processed microstructure, which contained areas strained up to 10 %. The occurrence of such high strain values already at 4.5 % of strain indicated a strong localization of plastic deformation in the LPBF-processed material compared to the hotrolled one. This plastic deformation nature – small number of highly active slip bands – was clearly visible in the DIC map, see Figure 3. This trend is even more pronounced at 10 % strain, see Figure 4. Generally, the strain distribution peak was significantly broader and shifted to higher strain values. However, heterogenous character of strain localization is further strengthened, which was indicated by highly deformed areas and nucleated deformation twins. 3.3. High-resolution EBSD characterization of selected grains To make a comparison between the forms of plastic deformation that occurred in two microstructures studied, one grain was chosen from each specimen for HR-EBSD analysis which was performed using CrossCourt software (see Figures 6 and 7). These grains were extracted from the undeformed IPF X maps and the grain crystal lattice slip systems; their planes, directions and Schmid factor (SF) values were determined. Slip planes with the highest SF were taken in account and their traces are highlighted in IPF maps (see Fig. 6a, 7a). Within each experimental interruption at strain values of 1.5 %, 3 % and 4.5 %, KAM maps representing the crystal lattice local misorientations and Von Mises stress maps were determined for both grains. The HR-EBSD analysis of the hot-rolled grain is illustrated in Figure 6. Pursuant to the calculated SF values, it can be assumed that a pronounced dislocation slip activity was initially hindered due to low SFs of principal slip systems. KAM maps revealed a local crystal lattice misorientations increase at 1.5 % strain, presumably due to nucleation of dislocation pile-ups along grain boundaries, for instance in upper left segment. With further straining, the dislocation density in grain interior decreased, reflected by lower values of KAM maps, but also mild emergence of KAM values along system with the highest SF. The Von Mises stress map showed a gradual increase in stress values during loading with increased values at the sites of dislocation pile-ups. At 4.5 % strain, regions Fig. 5. Plots showing axial strain distributions comparison at selected test interruptions of LPBF-processed specimen (translucent barplot) and hot-rolled specimen (line-fit plot). of highly variable stress values within the grain occurred that may indicate the influence of surrounding grains. The HR-EBSD analysis of the LPBF-processed material is shown in Figure 7. This grain possessed a slip system with notably high SF, predetermining it as a favourable slip system. The KAM maps contained multiple vein-like high-valued misorientation areas already in the undeformed state. These lines represented subgrain boundaries, which are commonly observed in LPBF materials (Šmíd et al., 2023). The effect of subgrains is also notable in Von Mises stress maps, where several subgrain regions are characterized by slightly different stress levels. The dominance of slip activity along the favourable slip system is apparent already at 1.5 % strain and is increasing with further loading. Several not calculated pixels along the slip system (Fig. 7g) were present already at the initial test interruption, which expanded across the grain with further loading. This feature reflected Fig. 6. Characterization of the crystallography of a single grain of the hot-rolled specimen conducted at strain levels of 0 % (a, b, f), 1.5 % (c, g), 3 % (d, h) and 4.5 % (e, i). a) The undeformed crystal orientation of the grain shown as an IPF X image, with the slip systems and their Schmid factors annotated. b) - e) KAM maps of the evolving misorientation of the grain crystal lattice during loading. f) - i) Maps of evolving stress within the analysed grain during loading. Fig. 7. Characterization of the crystallography of a single grain of the LPBF-processed specimen conducted at strain levels of 0 % (a, b, f), 1.5 % (c, g), 3 % (d, h) and 4.5 % (e, i). a) The undeformed crystal orientation of the grain shown as an IPF X image, with the slip systems and their Schmid factors annotated. b) - e) KAM maps of the evolving misorientation of the grain crystal lattice during loading. f) - i) Maps of evolving stress within the analysed grain during loading. the process of nucleation and growth of deformation twins. Due to their high misorientation to parent grain, HR-EBSD automatically excluded these pixels from Von Mises stress calculations. Similarly, a particular subgrain area is absent in the maps at 3 and 4.5 % strain. 4. Discussion A tensile test diagram comparing both tests performed on the two specimens studied was presented in Figure 2. A comparison of the curves indicated the influence of the microstructure on the mechanical properties of a material with the same chemical composition. The LPBF-processed specimen, characterised by an order of magnitude smaller grain size, exhibits significantly higher strength characteristics in comparison to the hot-rolled specimen without major loss of plasticity. In general, grain refinement has a very positive effect on increasing mechanical properties but often leads to a significant reduction in materials plasticity (Fan et al., 2020). In the case of a conventional austenitic steel with FCC structure, here represented by a hotrolled specimen, plastic deformation takes place by the movement of Shockley partial dislocations through the crystal lattice, leading to the formation of slip bands (Kettunen & Kuokkala, 2006). In the case of LPBF-processed material, however, the situation is more complex. A further level of subdivision of the material structure into dislocation cells has been reported (Wang et al., 2018) within individual grains. This is a consequence of the substantial temperature gradient that exists during the AM process (Melia et al., 2019). These columnar cells, which are orders of magnitude smaller than a single grain, have been shown to strongly influence the movement of partial dislocations through the grain. As dislocations pass through the cell boundary, they undergo a process of splitting, while the leading partial traverses the cell, the trailing one halts at the cell boundary, thereby generating a stacking fault between these two partials. The subsequent movement of these partial dislocations can only occur in the event of an increase in external load. As documented earlier (Liu et al., 2018), the dislocation network hinders but does not halt the movement of the partial dislocations, thereby resulting in an enhancement of mechanical properties without a substantial loss of plasticity. Another phenomenon captured well by DIC and KAM maps is the process of localization of plastic deformation in LPBF-processed material. While the hot-rolled material manifests a typical activation of numerous slip bands studied by (Kettunen & Kuokkala, 2006), the LPBF-processed specimen shows a intensive localization of plastic deformation into a limited number of slip systems. This localization is caused presumably by the complex microstructure of the dislocation cells of each grain, which limit the slip bands formation. Therefore, a few but more active slip bands are formed in the LPBFprocessed materials. Furthermore, this phenomenon can be observed by comparing the histograms of the DIC strain values of the two samples, where a greater number of strain values higher than the macroscopic strain value of the sample can be seen. Moreover,