Sensitization of 316L Stainless Steel made by Laser Powder Bed Fusion Additive Manufacturing

Laser powder bed fusion (L-PBF) additively manufactured (AM) facilitates the production of complex geometry parts with minimal need for postprocessing. The layer-by-layer approach of manufacturing allows for geometrical freedom, enhanced performance, and reduced upfront production costs.¹  Realizing these advantages, industries such as nuclear, aerospace, and medical have adopted the technology and begun evaluating materials for specific applications. For noncritical applications, AM parts may be used with no detrimental effects; however, critical applications require extensive evaluation before implementation. Traditionally, manufactured metals have been studied for decades resulting in a significant knowledge base. Unlike their wrought counterparts, AM materials produce microstructures with heterogeneous and anisotropic features such as porosities and columnar grains with small, dendritic cellular structures forming within the grains.^1-3  Many of these features are caused by the extreme temperature gradient that forms because of the localized heating from the laser^1-3  and are heavily dependent on AM process parameters.^3-7 

Sensitization impacts alloys exposed to high temperatures for prolonged periods of time and corrosive environments. This phenomenon refers to the precipitation of Cr carbides along grain boundaries (GB) within the temperature range of 500°C to 800°C. As a result of the diffusion of Cr, neighboring GBs generally show Cr depletion thus making the material susceptible to intergranular corrosion (IGC).^2,8-10  The alloys with higher carbon content possess a great threat of sensitization.¹¹  As such, low carbon materials such as AISI 316L stainless steel (SS) (UNS S31603⁽¹⁾) have been developed to minimize the impact. 316L SS is an austenitic steel widely used for conditions where sensitization is concerning.¹²  In the temperature range of 500°C to 800°C, M₂₃C₆ tends to form along high-angle GBs (HAGB) (boundaries with misorientation angle >15°) with little to no carbides precipitating along coherent ∑3 twin boundaries within low-carbon austenitic SS.^11,13-15  The nucleation preference of HAGBs is due to the higher energy of HAGBs compared to the lower misorientation angle boundaries.

Of the evaluation methods for the degree of sensitization (DOS), two popular tests have emerged: double-loop electrochemical potentiokinetic reactivation (DL-EPR)¹⁶  and ASTM A262¹⁷  practice “A” (also called “ditching test” in this work). DL-EPR tests have been extensively used for wrought austenitic SS^18-21  with Rebak and Dean¹⁸  showing a high degree of repeatable and reproducible results among different testing laboratories. Although DL-EPR has been used for wrought materials, little work has been done on AM austenitic SS outside of the work of Macatangay, et al.,¹⁰  Laleh, et al.,⁹  and Man, et al.,²²  where conflicting conclusions were noted. Macatangay, et al.,¹⁰  found similar IGC resistance between AM and wrought austenitic SS while Laleh, et al.,⁹  found the AM material showed superior IGC resistance compared to its wrought counterpart due to the high percentage of twin boundaries. The ditching test has been used as a rapid screening test for many years^12,20,23  with Macatangay, et al.,^8,10  demonstrating its use for AM materials. The results showed that AM materials in the as-built condition are highly susceptible to corrosion attack by means of melt pool boundary rather than GB attack. The limited knowledge of DOS in AM materials and conflicting conclusions of previous work^8-10,22  require further investigation into the application and evaluation of the two testing methods for AM materials.

The objectives of this work are two-fold. The study is to provide further understanding into the use of different DOS evaluation techniques when applied to AM austenitic SS and evaluate the role of dislocation cellular structures on sensitization kinetics of AM 316L SS. Due to AM 316 SS (both high and low carbon variants) being used for intermediate or high-temperature nuclear environments, it is essential to ensure current testing methods are properly evaluating the material, otherwise, parts may be evaluated with false negatives or worse, false positives. Another objective of this work is to generate DOS data of both wrought and AM 316L and develop mechanistic understandings. Given the conflicting conclusions and lack of data, additional research is required to evaluate AM 316L for critical applications within the sensitization temperature range.

Commercially available 316L SS powder manufactured by nitrogen gas-atomization (Carpenter Powder Products) and wrought 316L SS (Rolled Alloys, Inc.) were used in this work. The chemical composition of the two materials used was specified in Table 1. The 316L SS powder with an average size of 15 µm to 45 μm was used to fabricate 1.0 cm × 1.0 cm × 1.0 cm specimens with a Concept Laser Mlab LaserCusing®^† system with the following process parameters: 90 W laser power, 600 mm/s scan velocity, 80 μm hatch distance, and 25 μm layer thickness. The process parameters were predetermined for density greater than 99.7%.¹⁵  The laser scan strategy consisted of a continuous laser with a 90° rotation between layers. The wrought material was received as 12.7 mm (0.5 in nominally) round stock in the solution annealed condition.

Several sensitization heat treatments were applied to both AM and wrought samples. Based on previous work,^8-11,22  sensitization heat treatment at 700°C was chosen. The AM samples were sensitized from the as-built condition at 700°C for five different times: 0 h, 1 h, 24 h, 50 h, and 100 h. In addition to these treatments, an as-built AM sample was heat treated at 1,038°C for 10 min to recover dislocation structures within the material and subsequently heat treated at 700°C for 24 h. This specific heat treatment was designed to evaluate the contribution of dislocation cellular structures to sensitization kinetics. The wrought material was sensitized at 700°C for 0 h and 24 h for comparison. All sensitization heat treatments were followed with a water quench to prevent further sensitization during cooling. A table with all the materials and heat treatments we are found in Table 2.

To prepare the samples for testing, the AM parts were cut using a Buehler Isomet 2000^† Precision Cutoff Saw transverse to the build direction. A 304L wire was spot welded on the back of each sample and fed through a glass tube. The glass tube acts to isolate the spot-welded wire from testing solutions and ensure no corrosion occurs on the wire. The samples were then cold mounted in Struers EpoFix^† with the epoxy covering the sample, spot welded wire and at least 0.5 in of the glass tube. Before testing, specimens were polished from 120 upto 1000 grit and cleaned in an acetone bath in an ultrasonic cleaner for 5 min. Following cleaning, the testing area was isolated using 3M^† 470 electroplaters tape with a 0.25 in diameter hole with sharp edges punched out. To achieve high-quality electrochemical tests, crevice corrosion must be avoided, so preliminary tests were performed to ensure little to no crevice corrosion occurred when developing sample preparation techniques.

Once the specimens outlined in Table 2 were prepared according to the above process, they were evaluated for their DOS by means of DL-EPR according to ISO 12732.¹⁶  The DL-EPR test was conducted in a 0.5 M H₂SO₄ + 0.01 M KSCN solution deaerated with N₂ gas. The test was conducted with a Gamry Reference 600+^† with a platinum counter electrode (CE) and saturated calomel electrode (SCE) reference electrode (RE). The CE was kept in place while the working electrode (WE) (mounted sensitized samples) was placed 0.5 in from the CE, ensuring the WE and CE were parallel. The luggin probe tip of the RE was placed as close to the WE as possible without breaking a direct current path to the CE. Each specimen was exposed to the solution for 30 min to obtain open-circuit potential (E_OCP) and to ensure a more consistent surface oxide layer. Once OCP was reached, a potentiodynamic test was performed in the anodic direction starting at −100 mV_EOCP to +300 mV_SCE at a scan rate of 1.67 mV/s. Once +300 mV_SCE was reached, the scan direction was reversed and finished at −100 mV_EOCP at a scan rate of 1.67 mV/s. DOS is reported as the maximum current in the reactivation loop (i_r) divided by the maximum current in the activation loop (i_a). If no reactivation peak occurs, the DOS is said to be zero.

Tests were performed according to ASTM A262¹⁷  practice “A” with both an oxalic acid and ammonium persulfate (APS) solution for comparison. All samples were prepared by the procedures outlined in Material Preparation section. The 10% oxalic acid and 10% APS test were conducted by etching the samples at 1 A/cm² for 1.5 min and 5 min, respectively. Following the tests, the samples were subjected to an acetone ultrasonic cleaning for 5 min. Grain boundary attack is classified as “step,” “dual,” or “ditch” according to the severity of IGC. Samples with “step” or “dual” structures pass; however, samples with “ditch” structures do not pass and require further quantitative testing. It is important to note that the ditching test is not able to fail a sample, only pass it.

Following DL-EPR and ditching testing, all samples were examined by optical microscopy using a Zeiss Axiovert 10^† for qualitative analysis of the grain boundary attack without further polishing. Electron backscattered diffraction (EBSD) was performed on a JEOL 7000 F^† scanning electron microscopy (SEM) where samples were mechanically polished up to 1 μm aluminum oxide suspension followed by vibratory polishing in 0.05 μm colloidal silica suspension to help remove residual surface strain. EBSD data processing and mapping were performed in MATLAB^† using MTex²⁴  and ATEX²⁵  was used for inverse pole figure mapping. SEM characterization was performed on a Quanta 650^† FEG using secondary electron (SE) detection. Disk specimens of the AM-100 condition were prepared for scanning transmission electron microscopy (STEM) using a twinjet polisher at 40°C and 10 V to 20 V. The electropolishing solution was made with 10 vol% perchloric acid and methanol. The energy-dispersive x-ray spectroscopy (EDS) maps were performed on a Thermo Scientific Talos F200X^† at 200 kV. Line scan data were generated using Velox^† Software.

Comparing the DOS values of AM-24 and W-24 samples in Figure 4, there seems to be little difference (on average 0.44% and 0.37%, respectively) given that these two samples are considered unsensitized (below 1%). This conclusion is very similar to the work of Macatangay, et al.,¹⁰  and contradictory to the reported work of Laleh, et al.,⁹  in which AM 316L SS exhibited significantly fewer DOS than wrought 316L SS. Such a difference in DOS can be explained by the negligible twin fraction present in this work’s AM 316L SS. Therefore, the process-to-process variation of parameters in L-PBF can lead to different corrosion responses.

To further understand the response to ditching, APS was also used as a ditching solution. Samples etched with APS showed significantly more attacks as seen in Figure 6. APS is recommended for molybdenum-containing austenitic stainless steels which may show difficulty revealing step structures. However, the as-built AM and nonsensitized wrought samples showed step structures using oxalic acid indicating there was no need to use APS for these samples. In Figure 6, one can see the significant amount of end grain pitting in the wrought samples. The end grain pits significantly hinder the evaluation of the GB attack. As such, APS may not be the best tool to evaluate IGC susceptibility in all molybdenum-containing austenitic stainless steels.

Investigation into the recovered and nonrecovered 24-h sensitization samples, AM-R-24 and AM-24 respectively, was performed to investigate the effects of dislocations on the sensitization kinetics of this material. After calculating the DOS for each specimen, the recovered and nonrecovered samples, on average, exhibited a DOS of 0.00438 and 0.00443, respectively, with the data points plotted in Figure 4. In Figure 6, the ditching test shows an HAGB attack that is quite comparable between the two heat-treat conditions in both oxalic acid and APS. With both conditions showing similar sensitization characteristics, the recovery heat treatment had little to no effect on sensitization kinetics. As such, the role of dislocation structures as diffusion channels for carbon and Cr seems to be minimal at best. This does not particularly align with the literature^38-40  where dislocations function as pipelines for carbon diffusion and significantly increase the diffusion at large depths. For wrought 316L, it has been shown that increasing dislocation density by cold work is effective at increasing DOS after low-temperature sensitization heat treatments (500°C for 264 h).⁴⁰  Thus, further investigation into dislocation effects on sensitization kinetics in AM 316L is warranted. It is well known that the diffusion of carbon is significantly faster than that of Cr, so one would expect the carbide precipitation kinetics to be rate limited by Cr. With the low-carbon variant of 316, there is little carbon available to form carbides, and as a result, a lower driving force for precipitation. AM 316L SS by L-PBF presented a higher amount of GBs than wrought coarse-grain 316L SS. A higher nucleation rate in AM 316L SS may accelerate carbon consumption. As carbides start forming, there is even less carbon in the solution that is available for carbide precipitation, leading to the majority of solutionized carbon being consumed. Because there was no significant difference in DOS after a recovery heat treatment, dislocation density appears to have minimal impact on carbide precipitation. Because dislocation density plays a large role in Cr diffusion, Cr diffusion should not be concluded as the rate limiting factor. While Cr diffusion may have been slightly retarded by the recovery heat treatment, we propose the rate limiting factor for carbide precipitation of AM 316L SS is the low carbon concentration.

Investigating the Cr level at GBs far away from carbides and the Mo particles, it is seen that, in Figure 9(b), some Cr depletion occurs (dropping from approximately 17 wt% to 14 wt%). Knowing that the AM-100 sample showed significant IGC susceptibility using both testing methods, on average, GBs away from particles and carbides are generally prone to IGC. We propose the Cr wt% that makes a material prone to IGC is closer to 14 wt%. By long-time heat treatment, desensitization can happen by the back-diffusion of Cr to the depleted region. However, in this study, desensitization was not seen at the sensitization times up to 100 h. Figure 9(c) shows an EDS scan of a Mo-rich precipitate. While these precipitates originally showed high S, Si, and P levels, plotting the atomic fraction of these elements with Mo revealed that the precipitates are Mo₃Si due to the Si fraction closely following the shape of the Mo scan. Mo₃Si has been observed in other works.¹⁵ 

Once the DL-EPR SEM images were taken, the sample was ground until no optical evidence of GB attack was present, further ground in accordance with Intergranular Attack Susceptibility (Ditching Test) section, and tested using oxalic acid ditching. The area of interest was found and SEM images were taken (Figure 10[b]). One can see that the grains do not exactly mirror the ones found in the DL-EPR test which is likely due to the material removal by grinding. As seen in Figure 10(b), the ditching test significantly attacks the GBs and induces pitting. Because the ditching test is a current-controlled test, the sample may have varying electrical potentials which could result in a potential being high enough to cause pitting. Interestingly, some melt-pool boundary attacks (blue lines in Figure 10[b]) can also be seen in the ditching test which is not present in the DL-EPR test. An important conclusion from this work is that both tests are attacking the GBs of the sensitized AM 316L SS, and the attack using the ditching test are more severe.

The AM-100 sample was on average considered non-sensitized by STEM work (Figure 9[a]) by the traditional 12 wt% Cr rule²⁰  but statistically showed some boundaries will be sensitized below 12 wt%. This fundamental, quantitative approach of determining sensitization provides great insight to assess other more time-conscious approaches such as the DL-EPR and ditching tests. An important distinction should be made that the DL-EPR test did not reveal melt pool boundaries in the AB and AM-1 conditions. Therefore, the DL-EPR test evaluates material for IGC regardless of it being AM or wrought. The quantitative results of the DL-EPR test (Figure 4) for the 700°C 100-h condition indicated the material as nearly sensitized which corresponds well to the STEM work performed where Cr depletion occurs and can statistically reduce GB Cr levels below 12 wt%. Any sample sensitized at 24 h or longer also qualitatively showed an IGC attack (Figure 5) which is contradictory to the quantitative results that evaluated the samples as nonsensitized. With IGC being seen after 24-h sensitization time, samples tested should be microstructurally evaluated in addition to calculating DOS. This does raise a question: if there is IGC even when the DOS is below 1%, should the material pass even if IGC is present? It is imperative to not jump to conclusions about the IGC susceptibility based on one point of data but to use as many characterization tools as possible to fully evaluate the material. On the positive side, the DL-EPR test evaluated the DOS as increasing with sensitization time which is consistent with the microstructural features. The combination of the benefits manifests the DL-EPR test as a prime candidate for IGC evaluation of AM and wrought austenitic SS, although ISO 12732¹⁶  considering a DOS of 1% to be sensitized may not provide a full picture of IGC susceptibility. As such, further work should be performed to develop a DL-EPR standard that takes the potentiodynamic response and microstructural features into account.

Sensitization evaluation using the ditching test of ASTM A262¹⁷  could present some complications when evaluating austenitic SS produced by AM. In this work, as-built samples showed heavy melt-pool boundary attacks with stepped GBs. The initial thought would be to fail the material, but further investigation into melt-pool attacks is required. Godec, et al.,⁴¹  found slight Cr and Mo depletion along melt pool boundaries (17.4 wt% to 16.6 wt% and 4.2 wt% to 3.1 wt%, respectively) in AM 316L SS in the as-built condition. The slight reduction in Cr along the melt pool boundary would explain the preferential etching phenomenon. Sommer, et al.,⁴²  argued that melt pool boundaries could be etched due to chemical segregation or changing solidification modes from a cellular to planar form. With this in mind, melt pool boundary etching should not be characterized as sensitization using the ditching test because the elemental segregation is not significant enough to promote IGC and carbide formation does not occur. Performing the ditching test does reveal stepped GBs in the AB condition, which is consistent with its wrought counterpart, however, the deep etching of melt pool boundaries may lead some to fail the sample unjustly. As the sensitization time increased to 24 h and beyond, the ditching test revealed the GBs as susceptible to IGC, and significant pitting was also observed. From this change in preferential attack location, one could conclude that sensitization treatment was significant enough to promote sensitization. Observing IGC in samples sensitized for 24 h and beyond is consistent with the DL-EPR test, although the GBs in the ditching test are qualitatively attacked more (Figures 5, 6, and 10). While an advantage of the ditching test is its ability to quickly screen a material, AM austenitic SS samples require more complex analysis of different microstructural features. ASTM A262¹⁷  would, therefore, need to be updated for AM austenitic SS to include melt pool attack in addition to the normal GB classification. Beacuse the ditching test does not only test the GBs, and, as a result, evaluators must use a more subjective evaluation method, the DL-EPR test seems to be a more promising testing method for AM austenitic SS for sensitization.

• Sensitization heat treatments at 700°C for various times up to 100 h were performed on AM and wrought 316L SS. In order to evaluate the DOS, the DL-EPR and ditching tests were performed on all samples. This work also investigated the role of dislocation cellular structures on the precipitation kinetics of Cr carbides and the Cr depletion along GBs after 100 h of sensitization in AM 316L SS.

• The DL-EPR can be used to detect IGC susceptibility in AM 316L SS. Melt-pool boundary and cellular structure were not attacked by a DL-EPR test, but selective surface oxide dissolution was present. The quantitative results of the DL-EPR test are a useful tool to evaluate DOS between samples, but post-test microstructural characterization should be performed to evaluate IGC attack. The DL-EPR test should be updated to provide a more nuanced determination of sensitization that includes microstructural evaluation rather than solely relying on the 1% rule. The DL-EPR test reactivation peak current does not rely only on IGC but also surface oxide dissolution which may lead to overestimating DOS.

• The ditching test can be used to detect IGC susceptibility in AM 316L SS as well. A significant melt-pool boundary attack was observed. The use of APS is generally used for Mo-containing steels, but the attack observed was much more significant compared to the oxalic acid. This could make post-test microstructural characterization challenging. The qualitative results of the ditching test provide a subjective qualification protocol. As the melt-pool boundaries do not have enough Cr depletion to cause IGC, the ditching test should be updated to inform evaluators that melt-pool boundaries should not be considered ditching structures.

• The minimum time for sensitization peaks to occur during a DL-EPR test is between 1 h and 24 h. This correlates well with the microstructural results of the DL-EPR and ditching tests. There is minimal difference in the DOS between AM and wrought 316L in this work.

• A recovery heat treatment of 1,038°C for 10 min was effective in promoting dislocation annihilation without promoting recrystallization. As a result, the recovered condition does not show a cellular structure. Recovery heat treatment had little to no effect on DOS and IGC resistance which is likely due to the low carbon level in AM 316L SS rather than Cr diffusion. As such, dislocation density likely has little effect on sensitization kinetics in AM 316L.

• Cr carbide precipitation and HAGB Cr depletion are present in AM 316L where the depletion reached 12.35 wt% on average after 100 h of sensitization at 700°C. GBs away from carbides exhibited 14 wt% Cr which were prone to IGC, thus the Cr wt% that makes a material prone to IGC is close to 14 wt% in contrast to 12 wt%.

The experimental work was primarily supported by The U.S Nuclear Regulatory Commission (NRC) under Grant No. NRC-313100-21-M-0047. TEM characterization was supported by Dr. Miao Song (University of Michigan) through the Idaho National Laboratory’s Laboratory Directed Research & Development (LDRD) Program under the U.S. Department of Energy (US DOE)’s Idaho Operations Office Contract No. DE-AC07-05ID14517. J. Snitzer also thanks the support from the University Nuclear Leadership Program (UNLP) Graduate Fellowship by the Office of Nuclear Energy, The U.S. Department of Energy. This material is based upon work supported under a Department of Energy, Office of Nuclear Energy, University Nuclear Leadership Program Graduate Fellowship.