Measuring the Residual Stress and Stress Corrosion Cracking Susceptibility of Additively Manufactured 316L by ASTM G36-94

Additively manufactured (AM) materials are of interest to a wide variety of fields such as aerospace and defense. A major barrier to qualification and commercial implementation is the potential unpredictability of replicated parts. This unpredictability in AM parts, specifically those made using laser-based powder fusion (LPBF) techniques, stems from microstructures, residual stresses, and processing defects unique to the process. In turn, these affect material properties such as corrosion resistance or environmental assisted cracking.^1-2  Conflicting reports exist as to whether the tensile residual stresses induced by LPBF stainless steels lead to an increase in susceptibility to local corrosion initiation or change in corrosion resistance.^3-5  The stress corrosion cracking (SCC) susceptibility of LPBF stainless steel in the as-built condition is generally reported to be increased compared to their wrought counterparts due to their inherent levels of tensile residual stresses.⁶ 

Wrought austenitic stainless steel products that have tensile residual stresses typically show signs of enhanced susceptibility to SCC when exposed to hot chloride environments.^7-9  Most of the residual stress/SCC studies on wrought austenitic stainless steels are performed on machined surfaces, by milling, grinding, or blasting, where the residual stresses are in the near surface region. Rhouma, et al., show grinding of a 316L (UNS S31603⁽¹⁾) bar product to induce tensile residual stresses (>400 MPa) and cause rapid initiation and growth of cracks in a 40% boiling MgCl₂ environment while a sand blasted surface had compressive residual stresses (∼ −175 MPa) and showed no signs of cracking.⁷  An approach to remove similarly susceptible surface layers was shown by Zhang, et al., for a milled 316 (UNS S31600) stainless steel part.⁹  Electropolishing was used to remove the milled surface layer, and its large tensile residual stresses, exposing the parts underlying annealed and stress relieved microstructure which never showed signs of SCC when immersed in boiling MgCl₂. This would suggest stainless steel parts with similar annealed microstructures, and near zero or compressive residual stresses, will be less susceptible to SCC in hot chloride environments.

Several studies have observed the beneficial effects of heat treatments on alleviating tensile residual stresses of LPBF 316L and their subsequent impact on SCC susceptibility. De Bruycker, et al., tested LPBF 316L with various stress relieving heat treatments in boiling magnesium chloride (MgCl₂).¹⁰  They found the parts subjected to 950°C possessed fewer cracks than those treated at 450°C and the as-built condition. Lou, et al., also tested heat-treated LPBF 316L for SCC but in high-temperature water.¹¹  Their results were similar to De Bruycker, reporting that a heat treatment of 650°C for 2 h showed much higher rates of SCC crack growth compared to material treated at 955°C for 4 h which was partially recrystallized. Dong, et al., found that a heat treatment of 750°C for 2 h would increase or decrease SCC of LPBF 316L exposed to MgCl₂ at 75°C and 70% RH depending on the build direction.¹²  They attributed the decrease in SCC to stress relief and the increase in susceptibility to the persisting non-equilibrium microstructure, obscuring the relationship between residual stress and SCC susceptibility in LPBF materials.

This paper examines the relationship between residual stress and SCC susceptibility as well as the effect of recovery and recrystallization on SCC susceptibility in LPBF 316L. The residual stress is controlled via two different methods: sample height and heat treatment. Finite element analysis (FEA) was used in addition to experimental residual stress measurements to gain a more complete understanding of the residual stresses in the as-built conditions. Residual stresses were experimentally measured in both the as-built and heat-treated conditions with hole drilling. Susceptibility to SCC was tested with boiling MgCl₂ experiments via ASTM G36-94.¹³ 

The following work was conducted on 316L stainless steel made by LPBF from powder provided by 3D Systems on a 3D Systems ProX DMP 200^†. The composition of the powder, as determined by NSL Analytical, can be seen in Table 1. Samples were printed as cubes 15 mm × 15 mm × 15 mm with the processing parameters seen in Table 2. Each build layer was rotated 90° from the previous layer. The hexagon scan strategy, which is the default for the 3D printer, can leave a hexagonal, patchwork-quilt like impression on the as-printed surface and has been included in Table 2 for completeness. All samples were removed from the build plate with wire electrical discharge machining (EDM) at various heights to produce samples that were nominally 15 mm, 10 mm, 8 mm, 6 mm, and 4 mm thick. A schematic is shown in Figure 1. The EDM process will only affect the cut surface, comparatively far from the top surface being exposed. The thinnest sample in this study was 4 mm thick while the recast layer from EDM has been shown by other researchers not to exceed ∼30 μm.^14-15  The density of the parts was 98.9% as determined by the Archimedes method and using 8.027 g/cm³ as a reference value for 316L.¹⁶  Removal at varying heights was guided by computational modeling, described later in this section, and performed to keep the microstructure consistent for the as-built samples while varying residual stress associated with sample size. In an attempt to reduce residual stress, a variety of heat treatments were imposed to the thickest (15 mm) and thinnest (4 mm) samples. The heat treatments occurred under vacuum at 600°C, 800°C, and 1,200°C for 1 h followed by a furnace cool.

To experimentally assess the residual stress of the LPBF 316L cubes (before and after heat treatments), a commercial Stresstech electronic speckle pattern interferometry (ESPI) hole drilling system was used. A class 3R green laser at 532 nm and a reference beam were combined at a digital camera creating a speckle pattern through interference, which measures the deformation of the part in the direction of the measurement sensitivity vector caused by the drilling process. A high-speed 0.8 mm diameter drill on a computer-controlled stage is used to incrementally drill a hole into the specimen at known depths so residual stress measurements can be made from the surface to a depth of 0.6 mm at 0.025 mm increments. For these measurements a thin white coat of paint was applied to the surface to create an improved laser reflection and superior image quality. Due to a limited number of samples, the only conditions that had their residual stresses measured were the 15 mm, 10 mm, and 4 mm tall as-built parts and two heat-treated parts, 4 mm tall at 600°C and 15 mm tall at 800°C.

The PrismS^† software was used to analyze the ESPI measurements with the following assumptions: linear elastic and isotropic material, infinite thickness (4× the diameter of the hole), and no plastic deformation caused by hole drilling.¹⁷  The stress at each depth was determined by the relaxation created by the drilled hole and observed by the interferometric fringe pattern analyzed with a least-squares fitting concept.¹⁸  The software then used a look-up table approach which discretizes the problem based on the ratio of the hole depth and diameter, the modulus of elasticity and Poisson’s ratio, and the in-plane stress states correcting for any rigid-body motion of the sample. The sigma xx (σ_xx) values of the surface normal to the build direction are reported for the specimen in this study because the stress measurement sensitivity is greatest in this direction due to the experimental setup.

In addition to predicting resulting residual stresses, computational simulations were also used to design the experimental test specimens used in this study. Residual stress predictions were performed in the Sierra^† finite element software suite.^19-20  This software suite has been previously used by researchers to model different aspects of the AM process.^21-24  Simulations demonstrated that by removing progressively larger layers of material, the residual stress magnitude on the top surface could be reasonably controlled. This layer removal technique was motivated by destructive residual stress measurement methods such as the slitting method and contour method, which depend on stress relief caused by material sectioning.^25-27 

Modeling the AM process at the scale of the physical process settings is currently computationally prohibitive. As a result, a reduced order agglomerated layer simulation approach was used where an entire layer, scaled up to 0.5 mm, was deposited and individual laser scans ignored. This approach disregarded stress variations due to scan strategy but has been shown to be in reasonable agreement with measured elastic strains in AM builds.²⁸  A finite element mesh of the part, shown in Figure 2 below, consisting of 30 agglomerated layers 0.5 mm tall, along with a baseplate 140 mm × 140 mm × 15 mm, was created. Elements corresponding to the 15 mm cube were initially inactive. In this study, a planar heat source was used to heat each agglomerated layer for 0.1 s, and elements were activated upon reaching melt temperature through an increase in thermal conductivity. Laser power was adjusted until full melting of each layer occurred. After heating for 0.1 s, the layer was allowed to cool with no heat input. Total cooling time between each agglomerated layer was chosen to match the total cooling time for the number of physical layers in a 0.5 mm height. The bottom surface of the baseplate was constrained to ambient temperature, and convection and radiation were not included. After all layers were deposited, the entire part was allowed to cool to ambient temperature. The temperature-dependent thermal properties, including thermal conductivity and specific heat, were taken from Hodge, et al.²⁹  Remaining simulation parameters are summarized in Table 3.

After the thermal simulation was completed, the thermal history was used in a mechanical simulation. As in the thermal simulation, cube elements were initially inactive and were activated upon reaching melt temperature. A temperature-dependent viscoplastic constitutive model calibrated to 300 series austenitic stainless steel was used in this study. The formulation and parameters of the model are described by Beghini, et al., and Stender, et al.^23-24  The bottom surface of the baseplate was fixed in all three dimensions. As mentioned previously, different levels of residual stress on the top surface of the cube were desired to study the effect of stress magnitude on SCC while keeping the as-built material microstructure relatively constant. To achieve these different levels of stress, sectioning of the part at different heights up the cube was modeled by deactivating the baseplate and different lower sections of the cube followed by an equilibration step. For example, if a 4 mm tall section of the cube we desired, the baseplate and 11 mm of the lower portion of the cube were deactivated, leaving a 4 mm × 15 mm × 15 mm section of the cube. The deactivation process causes a change in residual stress, with the greatest top surface stress reduction corresponding to the cuts with the highest degree of material removal. Sectioning near the baseplate, such as in the 15 mm cube results, showed a negligible effect on the top surface stress. Sectioning near the top of the cube, such as in the 4 mm cube results, showed a significant change in residual stress on the top surface.

Prior to boiling MgCl₂ exposures the cube samples were coated in 3M^† Scotchkote^† liquid epoxy 323 on all surfaces except the top build surface, normal to the build direction. This exposed surface was left in the as-built condition and was not subjected to any kind of surface finishing. Following ASTM G36-94, samples were submerged in saturated boiling MgCl₂ at 155°C using the setup seen in Figure 3. If the samples exhibited cracking after 24 h, the test was concluded. Samples that did not show evidence of cracking were submerged again, checking periodically (roughly every 72 h) and eventually removed after 350 h or more. All samples that did not show signs of cracking after 24 h immersion also did not show cracking after the 350 h exposure.

After immersion in boiling MgCl₂, optical images were taken of the sample surfaces with a Keyence VHX-7000^† digital microscope. Subsequently samples were cut in half across the diagonal of the exposed surface for cross-sectional analysis. The cross sections of the exposed samples were mounted in epoxy and polished with successively finer silicon carbide paper, followed by vibratory polishing with alumina suspension to a 0.3 μm finish. This surface was examined with scanning electron microscopy (SEM) performed with a Zeiss Supra 55-VP^† field emission SEM equipped with an Oxford Instruments^† Symmetry electron backscatter diffraction (EBSD) detector. SEM was performed at a working distance of 9 mm to 12 mm at 20 kV and EBSD was performed with a step size of 0.4 μm. Geometrically necessary dislocation (GND) density analysis was also performed on the EBSD data using the techniques described by Bellotti, et al.³¹ 

The residual stress, determined by hole drilling, of select cubic LPBF 316L samples sectioned at different heights from the baseplate as measured from the top surface (4 mm, 10 mm, and 15 mm) can be seen in Figures 4(a) through (c). This approach verifies that the residual stress can be controlled while maintaining the as-built microstructure allowing the effects of microstructure and residual stress on SCC susceptibility to be separated. As anticipated, the sample cut at the build plate (15 mm thick) exhibits the largest average tensile stresses, while the thinnest sample (4 mm thick) has the lowest. Note that the center hole for the 15 mm thick specimen is the median value, while the 10 mm and 4 mm center hole show this to be the minimum value, a response shown in the simulations below. After a 4 mm thick sample was heat treated at 600°C, the σ_xx showed no statistical variation from the as-built 4 mm thick sample. This lack of stress reduction could indicate that at the levels of residual stress present in the 4 mm thick samples, 600°C was not high enough to initiate substantial residual stress relief by either plasticity from temperature-induced yield strength reduction or creep. For larger samples possessing higher residual stress values, residual stress relief could potentially be more apparent. On the other hand, the σ_xx measurement for a 15 mm thick sample after 800°C heat treatment showed a 75% reduction in residual stress.

Similar trends were seen in the FEA results, with in-plane and stresses shown in Figure 5. Stress contours are shown on a plane 0.5 mm below the top surface, which is a depth similar to where the hole drilling results are measured. The predictions show that in the 15 mm sample, high-tensile stresses are widespread across the top surface. However, as the part height is decreased, the tensile stresses begin to concentrate closer to the edges of the specimen. The initially tensile stresses at the center of the top surface of the sample transition into compression for the thinner specimens. This transition was also observed in hole drilling results and is caused by the redistribution of residual stress as material is removed from the bottom of the cube. Deformations also occur due to the residual stresses relieved by material removal, which forms the operating principle behind relaxation-based measurement methods.¹⁷  Cracking appeared in this location, which will be discussed later. Of note, the residual stress predictions, on average, are lower in magnitude when compared to hole drilling. This discrepancy is believed to be related to the agglomeration approach and decreased resolution when compared to finer scale hole drilling results. The element size in this simulation was 0.5 mm, so capturing effects at lower length scales is not possible. Directional stresses that are dependent on individual laser passes have been measured,³²  and this effect would not be captured in our reduced order model approach. The authors are actively working to develop more efficient high-resolution modeling approaches to capture the fidelity necessary to predict this behavior in large scale parts.

After 24 h of exposure to boiling MgCl₂, the top surfaces of the as-built LPBF 316L specimens, seen in Figure 6, all showed cracking. Cracking is randomly distributed over the surface for the 15 mm sample in Figure 6(a) and is much more pronounced than that of the shorter samples, indicating more severe cracking. This is consistent with the residual stresses being higher and more evenly distributed on the face of the thicker sample as indicated by the experimental and simulation results. A wrought sample of 316L was also exposed to boiling MgCl₂ to provide a general comparison to the LPBF material. The wrought sample showed, in Figure 6(f), no cracking after >300 h of immersion. The area of exposure for the wrought sample was roughly half that of the as-built samples due to the dimensions of the wrought samples in our possession, which may have impacted its susceptibility to SCC. The goal of this work was not to compare the residual stresses inherent to the LPBF process to traditional wrought nor was it to examine the effects of residual stress on the SCC behavior of wrought material so hole drilling was not conducted on the wrought material.

The shorter samples (10 mm to 4 mm) exhibit cracks propagating in a seemingly specific orientation, shown in Figures 6(b) through (e), with more cracks being straight or slightly curved compared to the sporadic propagation of the 15 mm sample. Cracks through the center of the samples generally traverse perpendicular to the melt tracks of the final build layer. Cracks near the edge of the sample generally form a ring around the circumference of the sample. The location of this crack ring is consistent with the location of the highest stresses identified by the hole drilling measurements and simulations in Figures 4 and 5, respectively. The difference between the residual stress in this ring (+100 MPa to +150 MPa) and the center of the surface (0 to −50 MPa) is more pronounced for thinner samples.

In samples with heights from 4 mm to 10 mm, the cracking in the center region (lower relative residual stresses) of the sample face propagates primarily along the diagonal, perpendicular the melt tracks of the final build layer, similar to the schematic shown in Figure 7. Simulations suggest the macroscale residual stress states are significantly reduced near the center compared to the sample edges. Strantza, et al., showed there can be large fluctuations in residual stress state from layer to layer with peaks in stress near the top surface (the surface perpendicular to the build direction).³²  Others have shown that for individual melt tracks, on the microscale, the maximum tensile stresses present themselves parallel to the melt track orientation and could cause the cracks to travel perpendicular to the melt tracks.^33-35  These cracks are not perfectly perpendicular to the melt tracks but instead bow out toward the edges of the samples. This is likely due to the higher residual stress ring near the edges as suggested by simulations and by other studies that show a transition gradually from low to high stress as you move away from the center of the surface.^36-37  Despite the reduced residual stress near the center of the surface, the samples between 4 mm and 10 mm in height still exhibit cracking in this region, primarily perpendicular to the melt tracks. This indicates that the microscale residual stresses associated with the melt tracks are likely a contributing factor but at high enough residual stresses arising from the sample geometry, such as might be the case for the 15 mm sample, those contributions are obscured. From this point on the 15 mm, 10 mm, and 4 mm tall samples are exclusively investigated because 15 mm and 4 mm are the extreme cases for the as-built materials and the 10 mm sample is the tallest sample to show a drop in residual stress at the center of the exposed surface.

Cross sections of samples exposed to boiling MgCl₂ are shown in Figure 8, the arrows correspond to the cracks most easily seen in optical images. The deepest cracks predominantly appear nearer to the edges of the sample, as seen in Figure 8(b). This aligns well with the results from the simulation and the approximate location of the high stress ring. The 15 mm sample, seen in Figure 8(a), is the exception and shows deep cracks along the length of the sample, near the edge and in the center of the sample. This is a testament to the elevated residual stresses across the entire surface for the 15 mm thick sample. Regardless of the sample, the deepest cracks extend between a third and halfway through the height of the sample. The 15 mm sample, with the largest tensile residual stresses, has a maximum crack depth of ∼7.2 mm, the 10 mm sample has a maximum crack depth of ∼3.4 mm, and the 4 mm sample has a maximum crack depth of ∼1.4 mm.

Backscatter electron (BSE) images of a representative crack from the 15 mm and 4 mm cross-sectioned specimen after boiling MgCl₂ exposure are shown in Figures 8(c) and (d), respectively. These cross-sectioned samples were etched prior to imaging to reveal microstructural features such as cell and melt pool boundaries. For both specimens, the cracks propagated in a mixed mode fashion, intergranular and transgranular, with minimal sign of melt pool boundaries (MPBs) acting as preferential crack pathways. The resolution of these images was insufficient to determine if the Cr/Mo enriched cell boundaries acted as preferential pathways for cracks, but it did not appear as if that was the case. These results are in opposition to the work by Yazdanpanah, et al., who suggested crack propagation occurred along MPBs.⁴  Yazdanpanah’s crack path could have been a result of using an environment less aggressive than boiling MgCl₂ (3.5% NaCl). It is more likely there was no cracking and what Yazdanapah observed was etching. MPBs are known to be sites for preferential etching and recent work by Godec, et al., suggested MPBs are depleted in Cr/Mo which would cause this preferential attack.³⁸ 

The residual stresses may be the key to the crack depths as well as the initiation. One possibility is that the cracks penetrate the sample until the residual stresses become compressive.^37,39  Bayerlein, et al., conducted a neutron diffraction study on LPBF manufactured cuboids (1 cm × 4 cm × 4 cm) of a nickel-chromium alloy and varied the location of residual stress measurement along the build and horizontal directions.³⁷  They showed the surfaces of the cuboids have tensile residual stresses but the interior of the cuboids, halfway down the sample, had compressive residual stresses. They also observed that the change from tensile to compressive residual stress varied in the build direction with differing sample build height. This could explain why the maximum crack depth varied with sample height. Additionally, Bayerlein, et al., notes that stresses were largely symmetric within the build layer with higher stresses being located near the edges of the sample.³⁷  This corresponds well with the higher stress ring reported in the simulation results above.

The results of the hole drilling residual stress measurements for the heat-treated specimens (600°C, 800°C, and 1,200°C) are given above the optical image of the same sample size following 24 h immersion in boiling MgCl₂ in Figure 9. In this study, samples heat treated at 600°C for 1 h showed cracking across most of the surface, similar to the thinnest samples tested in the as-built condition (4 mm and 6 mm). Increasing the annealing temperature to 800°C, the residual stress drops to a negligible level and there was no observed cracking after 24 h of immersion in MgCl₂ or following >350 h of immersion. The highest annealing temperature tested at 1,200°C showed similar results to the 800°C sample with no cracking observed following 24 h and >350 h of immersion. These results were verified by optical images of the surfaces and optical/SEM images of the annealed samples in cross section.

De Bruycker, et al., reported that a heat treatment at 450°C for 2 h for AM 316L shows reduced SCC susceptibility compared to as-built and samples annealed at 950°C for 2 h, which show no signs of SCC.¹⁰  They attributed the change in SCC susceptibility to the reduction of tensile residual stresses. The work presented here suggests that the reduction of SCC susceptibility cannot be solely attributed to a reduction in the tensile residual stress. The 15 mm thick sample treated at 800°C showed no evidence of SCC after the boiling MgCl₂ but its residual stress levels are similar, but still slightly larger, than those seen in the 4 mm thick sample both in the as-built condition and treated at 600°C. Both the as-built and 600°C treated 4 mm sample experienced SCC after exposure to boiling MgCl₂. If the magnitude of the residual stress was the only controlling factor for SCC susceptibility in these samples, the 15 mm sample treated at 800°C should exhibit the same behavior as the 4 mm as-built sample and 4 mm sample treated at 600°C.

Heat treatments affect not only the residual stress but also the microstructure. To quantify the impact of the heat treatments on the as-built microstructure, SEM/EBSD was used and the geometrically necessary dislocation (GND) density were determined, shown in Figure 10. The GND density for the 600°C heat treatment showed minimal change in GND compared to the as-built material, shown in Figures 10(a), (b), (e), and (f), coinciding with no significant stress relief or reduction in SCC susceptibility. Additionally, the etched microstructure of the as-built and 600°C samples, shown in Figures 11(a) and (b) suggests minimal differences, with both showing preferential etching of the Cr/Mo enriched cell boundaries as well as melt pool boundaries.

The 800°C specimen, in Figures 10(c) and (g), exhibits a more noticeable reduction in GND density, but not significant enough to suggest recrystallization. The 800°C etched microstructure, in Figure 11(c), still shows signs of preferential attack at cell boundaries and some melt pool boundaries but a significant change from the as-built microstructure is the prominence of grain boundaries, an indication of recovery. This recovery is happening on such a small length scale that it is difficult to see in EBSD. Voisin, et al., performed heat treatments of LPBF 316L material at 600°C and 800°C and, with transmission electron microscopy, showed a reduction in dislocation density at the cell boundaries for the 800°C annealed sample and no change in dislocation density for the 600°C material. A similar behavior is expected for the material in this study.⁴⁰  For the fully annealed specimen, 1,200°C in Figures 10(d) and (h), the GND density is similar to that of a fully annealed wrought material. This result is corroborated by the etched microstructure in Figures 11(d) and (e), which shows large grains (larger than wrought) and annealing twins along with a random distribution of oxide inclusions. The band contrast image with grain boundary character overlay and accompanying inverse pole figure maps are shown in the Supplemental Material, Figure S1, providing additional microstructural information.

A critical annealing temperature for LPBF 316L to mitigate susceptibility to SCC in boiling MgCl₂ was somewhere in the range of 600 < T ≤ 800°C in this study, because 800°C showed a reduction in tensile residual stress to levels on the order of the thin as-built samples and microstructurally showed signs of recovery (annihilation of dislocations and growth of sub-grains). EBSD analysis suggests recovery and a reduction in GND density is an important step toward lowering SCC susceptibility. Ronneberg, et al., studied the impact heat treatments have on LPBF 316L material showing 800°C to be a critical temperature which causes the dislocation cell structure to begin to disappear and MPBs to become less pronounced, similar to that shown in Figure 11.⁴¹  Voisin, et al., observed similar behavior, going a step further to show annealing at temperatures >600°C led to a clear drop in tensile yield strength and increase in ductility with an accompanied reduction in local dislocation density at cell boundaries.⁴⁰ 

The suspected reason for a reduction in SCC susceptibility was two fold, the reduction in local dislocation density coupled with a reduction in tensile residual stresses. The large local dislocation density at cell boundaries could be acting as areas of high localized stress or a region where H-trapping is prominent, leading to crack initiation.^42-43  The tensile residual stresses likely contribute to crack initiation by acting as an external stress to the bulk along with a secondary effect of dictating the depth and severity of the cracks after initiation. However, the annealed specimen at 800°C is a unique scenario where tensile residual stresses persist, but the local dislocation density has been significantly reduced, suggesting the latter is more critical to SCC initiation than residual stress. That said, separating these crack initiation factors is incredibly difficult and we are unable to make any concrete conclusions at this point.

In this study, the relationship between residual stress and SCC susceptibility of LPBF 316L was examined using boiling MgCl₂ tests, ASTM G36-94, showing that samples with larger tensile residual stresses exhibited an increased susceptibility to SCC. Residual stress in the as-built condition was controlled by varying sample height and was relieved by heat-treatment. The key findings of this study can be summarized as follows:

• Reducing sample thickness and applying heat-treatments were both effective ways of controlling residual stress as seen by hole drilling experiments and FEA simulation.

• All thicknesses tested for the as-built condition showed SCC susceptibility after 24 h in boiling MgCl₂ (ASTM G36-94).

• The thickest as-built sample (15 mm) shows the most severe cracking compared to the other samples, with cracks distributed randomly over the sample face.

• Thinner as-built samples (≤10 mm) show cracking traversing the center of the sample face perpendicular to the melt tracks of the final build layer, because of the lower overall tensile residual stresses at the center, and cracking around the circumference of the sample in a ring identified in the simulations as having higher tensile residual stress than the center of the sample face.

• A heat treatment of 600°C for 1 h exhibited no significant reduction in tensile residual stresses, GND density, or SCC susceptibility compared to its as-built counterpart.

• Heat-treatments of 800°C and 1,200°C for 1 h eliminated SCC susceptibility. EBSD, GND analysis, and the previous work by Voisin, et al., suggest a reduction in dislocation density is a critical step in preventing SCC initiation in the 800°C sample as opposed to simply reducing the tensile residual stress (>55 MPa) which persisted in the 800°C sample.⁴⁰ 

• The heat treatments at 800°C and 1,200°C for 1 h eliminated SCC susceptibility, even for boiling MgCl₂ immersion tests exceeding 350 h.

We wish to thank C.A. Profazi, J.E. Yang, and L.J. Jauregui for materials preparation and characterization. This study was supported by the Aging and Lifetime program and the Joint Munitions Program between the Department of Defense and Department of Energy. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology and Engineering Solutions of Sandia LLC, a wholly owned subsidiary of Honeywell International Inc. for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525. This paper describes objective technical results and analysis. Any subjective views or opinions that might be expressed in the paper do not necessarily represent the views of the U.S. Department of Energy or the United States Government.