Introduction

Additive manufacturing (AM) encompasses a diverse group of fabrication techniques based on a layer-by-layer accumulation process, enabling the rapid manufacturing of components for a wide range of applications. The ability to directly deposit material to manufacture complex parts enables more rapid manufacturing than those fabricated traditionally through casting or forging followed by subtractive manufacturing. As an example, AM could be used to facilitate the manufacturing of more complex cooling strategies in nuclear reactors, extending component lifetimes.1 AM can also be used for reduced lead time manufacturing of replacement parts to reduce power plant down time and improve power grid robustness. Features like high deposition rates, low equipment cost and relatively inexpensive feedstock materials offered by wire arc direct energy deposition (WA-DED) make this AM process particularly attractive for power generation applications.2

Austenitic stainless steels are commonly used in the fabrication of power generation components because of a favorable combination of properties such strength, ductility, corrosion resistance and creep resistance at elevated temperatures. Type 316L is often selected over other austenitic stainless steels such as 304 because of greater resistance to pitting and stress corrosion cracking.3 Type 316H is a high-carbon version of 316L and possesses higher strength and creep resistance in elevated temperature service due to the precipitation of M23C6 carbides under service temperatures between 538 and 816 ºC.4,5 Type 316LSi is a higher silicon version of 316L designed for improved fluidity of the melt pool during welding.6,7 The feedstock 16-8-2 is particularly different from the other 316L variants because it was designed to avoid the formation of the embrittling σ phase during long-term elevated temperature applications by lowering chromium and molybdenum content, making it ideal for welding 316H and other 300 series high carbon stainless steels.8 16-8-2 is commonly used for structural applications related to petrochemical and power generation industries.9

Although austenitic stainless steels are commonly used for power generation applications, the properties that make them attractive are known to evolve during high temperature exposure. Ayers showed that the average hardness of welded 316L stainless steels pipes increases after aging for 1000 h at 400°C.10 Abe et al. found that the austenite phase remains at roughly the same hardness during aging, but the ferrite phase increases in hardness during aging of 316L welds from 500 to 10,000 h at 335°C.11 Allahyari et al. evaluated the tensile ductility of 316L, reporting significant property deterioration during aging at 750 and 850 °C for up to 100 h caused by the transformation of δ-ferrite to the brittle intermetallic σ phase.12 Studies conducted on the higher carbon variant of 316L, 316H, found that the formation of several different precipitate phases M23C6, M6C, Laves and ferrite occurs during long-term thermal aging at temperatures between 640 to 650 °C for times between 28,000 and 60,000 h.13 The National Institute of Material Science (NIMS) found that phases such as χ and σ can also form in 316H microstructure depending on the combination of aging time and temperature.14 The σ phase is well known for being a hard and brittle Cr-rich phase. The formation of σ in austenitic stainless steels leads to Cr depletion in the metallic matrix, which deteriorates the material’s tensile ductility, toughness and corrosion resistance.15 The formation of deleterious phases during service must be considered for elevated temperature applications; however, most investigations about microstructure evolution during thermal aging have been conducted on base metal alloys of 316L and 316H, while there is considerably less information in the literature regarding the wide variety of welding feedstock materials.

The formation of embrittling phases such as σ can occur during the WA-DED process necessitating post build heat treatments. Chen et al. showed that the thermal cycling during the WA-DED process leads to σ formation at interfaces between δ ferrite and austenite at dendrite cores.16 Post-build heat treatments at 1100 and 1200°C for 1 h were shown to dissolve the σ phase leading to decreases in strength but increases in ductility compared to the as-built condition.17 Although there is a very well-developed body of literature regarding embrittling phase formation in conventionally processed austenitic stainless steels, very little work has been done to understand the stability of WA-DED processed stainless steels for long times at elevated temperatures. Moreover, little is known regarding how alloys specifically designed to be welding feedstock materials, such as 316LSi and 16-8-2, respond to thermal aging after WA-DED processing.

The objective of this work is to advance the understanding of how austenitic stainless steel welding feedstocks processed with WA-DED are affected by long-term exposure to elevated temperatures. We explore the microstructure and tensile property evolution of the common welding feedstocks 316L, 316LSi, 316H and 16-8-2 processed with gas metal arc welding (GMAW)-based WA-DED during aging at 650 °C for 1000 h. Variations in deposition strategy, straight and weaved deposition passes, were evaluated to determine whether the resulting differences in heat input affected the WA-DED material’s response to thermal aging. The results presented in this study are expected to provide insight into the suitability of commercially available austenitic stainless steel feedstocks for long-term high-temperature service.

Methods and Materials

Materials and Deposition Parameters

Four GMAW wire feedstocks were investigated in this work: 316L, 316H, 316LSi and 16-8-2. 316L and 316H were evaluated as a baseline because wrought versions of these materials are extensively used in high-temperature applications. No wrought versions of 316LSi and 16-8-2 exist, but these materials are common welding feedstocks for austenitic stainless steels and are therefore likely candidates for WA-DED. Lincoln Electric Blue Max Mig 316L18 and Blue Max Mig 316LSi19 were used for 316L and 316LSi, respectively. Feedstocks of 316H and 16-8-2 were sourced from Weldtool20 and Euroweld,21 respectively. A Fronius TPS 400i GMAW heat source using cold metal transfer (CMT), a controlled short circuit metal transfer mode, was used to make WA-DED builds. In all cases, the feedstock wire was 1.14 mm (0.045 in) diameter wire fed at 83 mm/s (195 in/min). A mixed Ar + 5% CO2 shielding gas was used at a flow rate of 11.8 l per min (25 cubic feet per hour). During deposition, the CMT deposition mode resulted in an average potential of 12.6 V and average current of 140 A with a travel speed of 4 mm/s (9.5 in/min). Two different deposition strategies were evaluated to create the WA-DED builds. A four straight pass per layer welding strategy with 7-mm-wide weld beads and a bead spacing of 4 mm was used to manufacture rectangular wall-shaped builds, measuring approximately 254 mm long, 101.6 mm tall and 19 mm thick. Additionally, a three weaved pass per layer welding strategy with 10-mm-wide weld beads and a bead spacing of 7.5 mm was also used to generate wall-shaped builds except the thickness was approximately 25 mm thick. The two build strategies were used to study the effect of heat input on the microstructure and mechanical properties of the builds. Schematics of the build strategies are shown in Fig. S-1 in the Supplementary Material (refer to online supplementary material). Chemical compositions of the WA-DED builds were measured by optical emission spectroscopy of the 316L, 316H, 316LSi and 16-8-2 feedstocks, shown in Table I. Nominal compositions of the wire feedstock materials are provided in Table S-1 in the online supplementary material.

Table I Compositions of WA-DED builds measured by optical emission spectroscopy and inert gas fusion* (C, N and O only)
Full size table

Thermodynamic Predictions

Thermo-Calc© version 2022a using the TCFE12 database was used to estimate phase stability as a function of temperature. The composition measurements taken on the builds shown in Table I were used for the simulations, except only the compositions of Fe, C, Cr, Cu, Mn, Mo, Ni, Si, Co and N elements were included. No restrictions were placed on the phases included in the simulations. Predicted phase amounts at 650°C were used to interpret changes in the microstructure of the WA-DED at 650°C.

Heat Treatments

Builds from each alloy were heat treated according to the American Society for Testing and Materials (ASTM) specification of 1 h at 1040°C22 followed by water quenching to homogenize the microstructure, aiming to dissolve any σ phase that might have been formed during the deposition process. Samples of each build were kept in the solution annealed condition for microstructure characterization and mechanical testing. Other samples of the builds were heat treated (aged) at 650°C for 1000 h to assess the influences of thermal exposures relevant to power generation applications.23 Small sections of each build, cut perpendicular to the deposition travel direction, were removed periodically during the 1000-h aging treatment to evaluate hardness. All heat treatments were conducted in an air furnace and water quenched after the prescribed time.

Hardness Testing

Hardness testing in the solution annealed condition and during aging was performed on samples sectioned from builds as described in the heat treatment section to get a sense of mechanical property evolution. Samples were mounted and metallographically polished samples and then hardness tested on a LECO AMH55 Automatic Hardness Testing System with a LM110AT indenter unit with a 500-g load. At least 20 hardness measurements were taken for each aging time (0, 100, 500, 1000 h) at 650°C. Only samples from the straight pass deposition strategy were hardness tested.

Microstructure Characterization

The samples taken during thermal aging were used for microstructure characterization. Samples were mounted and metallographically prepared prior to scanning electron microscopy (SEM), electron backscatter diffraction (EBSD) and X-ray diffraction (XRD) to study microstructure changes from thermal aging. Samples were polished down to 1-μm diamond and then vibratory polished with 0.05-μm diameter colloidal silica for 24 h. SEM imaging and EBSD were performed on a JEOL JSM-7000F field emission SEM with an accelerating voltage of 20 kV. Nearest neighbor pattern average reindexing (NPAR) was used to improve the EBSD phase indexing near grain boundaries. For lower magnification, EBSD scans (400 ×) to evaluate changes in phases present from aging, a 0.5 μm step size were used. To measure grain size, EBSD scans were performed at the lowest magnification possible (100 ×) to capture as many grains as possible. At least 53 grains were evaluated per condition, and the average grain size weighted by area is reported. A set of samples was etched using 2 ml nitric acid, 4 ml glycerin and 4 ml of HCl (glyceregia) at room temperature to reveal secondary phases for EBSD. Higher magnification EBSD scans (5000 ×) were performed on etched samples with a 0.05-μm step size to characterize grain boundary precipitates. XRD was performed using a Panalytical X’pert pro X-ray diffractometer with a Cu K-α source operated at 45 kV and 20 mA on samples sectioned perpendicular to the deposition travel direction. XRD scans were conducted with a step size of 0.008° and dwell time of 1120 s per step for a scan range between 40 and 55 2θ, resulting in an average scan time of approximately 5 h, while the same step size and a dwell time of 370 s per step were used to conduct scans in the range of 40–100 2θ, resulting in an average scan time of approximately 6 h. Fractography was conducted on failed tensile samples using a FEI Helios 600i SEM operated at 20 kV to determine whether any changes in fracture mechanisms accompanied changes in mechanical properties observed after aging.

Tensile Testing

Tensile testing was performed on solution-annealed and 1000-h aged samples at room temperature according to ASTM E8/E8M-2224 and at 650 °C according to ASTM E21-20.25 Three replicate samples were tested for each combination of heat treatment condition (solution annealed vs. aged), test temperature (room temperature vs. 650°C), build strategy (4 straight passes vs. 3 weaved passes) and feedstock material (316L, 316LSi, 316H and 16-8-2). Round dog-bone tensile specimens with a diameter of 6.35 mm (0.25 in) and 25.4 mm (1 in) parallel length were oriented horizontally or perpendicular to the build direction.

Results

Thermodynamic Predictions of Phase Stability

Figure 1 shows the results of thermodynamic predictions for WA-DED builds of 316L, 316LSi, 316H and 16-8-2 as a function of temperature. All four feedstocks are expected to experience primary ferrite solidification with a solid-state transformation to austenite, as shown in Fig. 1a-c. At lower temperatures, each feedstock is predicted to exhibit variations in the amounts of other phases that form. The phase amounts of η, Laves, σ, M23C6 and ferrite at 650°C are tabulated in Table II. As expected based on the alloy design intents, larger amounts of M23C6 are predicted at 650 °C in 316H and 16-8-2 (1 and 0.9 vol%, respectively) compared to 316L and 316LSi (0.5 and 0.6 vol%, respectively). Similarly, 16-8-2, which is designed to have minimal σ phase, is predicted to have no σ at 650°C, whereas the other alloys are predicted to have 14 vol%. Generally, 316L, 316LSi and 316H are predicted to have greater amounts of other secondary phases such as Laves and η compared to 16-8-2. Interestingly, 16-8-2 is predicted to have approximately 9 vol% ferrite at 650°C, whereas the other feedstocks have ferrite solvus temperatures ≤ 600°C (Fig. 1) and thus are predicted to have no ferrite. Considering the total number of phases associated with embrittlement, namely η, Laves and σ, 316LSi is predicted to have the largest amount (~ 18 vol%), and 16-8-2 is expected to have the smallest amount (~1 vol%) at 650°C. Accordingly, detailed characterization is presented on these two materials because 316LSi would be expected to exhibit the greatest degree of embrittlement and 16-8-2 would be expected to exhibit the least embrittlement from secondary phase formation during aging at 650°C.

Fig. 1
figure 1

© predictions of phase amounts present as a function of temperature for (a) 316L, (b) 316H, (c) 316LSi and (d) 16-8-2

Thermo-Calc

Full size image
Table II Predicted phase amounts (vol%) at 650°C in WA-DED builds
Full size table

Hardness Testing During Aging at 650°C

Hardness measurements were taken on samples from each feedstock material from the four straight pass build strategy during aging at 650°C, and the evolution of hardness as a function of time is shown in Fig. 2. Samples of 316H and 316LSi both show an increase in hardness upon the first 100 h of aging followed by a decrease in hardness for the remainder of the 1000 h aging period. It is possible that the formation of M23C6, Laves, σ and η may be responsible for the apparent age-hardening response in these materials. 316L shows very subtle changes in hardness during aging at 650°C for 1000 h and an overall lower hardness compared to the other materials despite having roughly similar amounts of M23C6, η, Laves and σ compared to 316LSi and 316H. Unlike the other feedstock materials, 16-8-2 shows a steady increase in hardness over the 1000 h aging period at 650°C, despite having lower predicted amounts of η, Laves and σ compared to the other feedstocks. The difference in age-hardening behavior of 16-8-2 compared to the other feedstocks indicates that minor secondary phases may not be responsible for the changes in strength at 650°C. It is possible that the formation of ferrite, as predicted in Table II, may be contributing to the steady increase in hardness of 16-8-2 during aging at 650°C. Although it was not evaluated, it is expected that samples of the various alloys with the three weaving pass deposition strategy would show similar hardness responses relative to each other during aging.

Fig. 2
figure 2

Hardness as a function of aging time at 650°C for WA-DED builds made with 316L, 316LSi, 316H and 16-8-2. The symbols are mean values, and the bars are 95% confidence limits using pooled standard deviations.

Full size image

Effects of Aging at 650°C on Tensile Properties

Solution-annealed and 1000-h aged samples of the evaluated feedstocks, with straight pass and weaved pass build strategies were tested at room temperature and 650 °C. Overall, minor changes in tensile properties were observed after aging, and no statistically significant effect of building strategy was found as shown in Fig. 3. Samples of 16-8-2 showed a statistically significant increase in ultimate tensile strength (UTS) of almost 100 MPa after aging when tested at room temperature, as shown in Fig. 3e. However, no other strength property of 16-8-2 showed statistically significant changes as a result of aging. Samples of 316LSi tested at room temperature showed a smaller yet statistically significant increase in UTS of approximately 5%, as shown in Fig. 3f. No other tensile property of 316LSi showed statistically significant changes as a result of aging. No significant changes in strength properties were observed in 316L or 316H at room temperature or at 650 °C because of aging (Fig. 3c, d, g and h). Aged samples of 16-8-2 showed significant decreases in total elongation compared to solution-annealed samples when tested at room temperature as shown in Fig. 3m. None of the other conditions showed statistically significant changes in ductility, at room temperature or at 650 °C as a result of aging (Fig. 3n, o and p).

Fig. 3
figure 3

UTS and YS vs. condition (solution annealed (SA) and Aged) for samples built with two different deposition strategies (straight path and weaving path) tested at 650°C for (a) 16-8-2, (b) 316LSi, (c) 316H and (d) 316L and at room temperature for (e) 16-8-2, (f) 316LSi, (g) 316H and (h) 316L. Total elongation vs. condition of the same samples for (i) 16-8-2, (j) 316LSi, (k) 316H and (l) 316L and tested at room temperature for (m) 16-8-2, (n) 316LSi, (o) 316H and (p) 316L. The symbols are mean values, and the error bars are 95% confidence limits calculated with pooled standard deviations

Full size image

Example engineering stress-strain curves of solution-annealed and aged 16-8-2 tested at room temperature are shown in Fig. 4. After aging, the strain-hardening rate is considerably higher leading to a higher UTS. The decrease in total elongation appears to be primarily related to a decrease in uniform elongation rather than a decrease in post-uniform elongation. This may suggest that microstructural features that serve to affect strain hardening, rather than process induced defects such as pores, are responsible for increases in UTS and decreases in ductility after aging at 650 °C. Interestingly, the composition of 16-8-2 is designed specifically to avoid embrittlement caused by σ phase formation during high temperature exposure, yet it is the only feedstock that shows an statistically significant decrease in ductility after aging. Further microstructural characterization is needed to understand the origin of the observed reduction in tensile elongation. Engineering stress-strain curves for all tested conditions are shown in the Supplemental material (Fig. S-2). Notably, the overall shape of the engineering stress-strain curves exhibited by 16-8-2 samples tested at room temperature is quite different from all of the other materials. Considerably higher strain-hardening rates are observed in the 16-8-2 samples during tensile deformation at room temperature.

Fig. 4
figure 4

Engineering stress-strain curves of 16-8-2 tested at room temperature (RT) demonstrating the difference in strain-hardening rate, UTS and total elongation between SA and aged conditions

Full size image

Microstructure Changes from Aging at 650°C

Grain Size

EBSD was used to evaluate possible changes in grain size caused by the straight pass and weaving build strategies as well as any changes from aging. Grain maps from each material and build strategy in the solution annealed condition as well as the straight path strategy after aging are shown in the Supplemental materials Fig. S-7. A summary of average grain sizes is shown in Table III. In general, the weaving strategy resulted in somewhat larger grain sizes compared to the straight path strategy, as expected. However, the differences in grain size observed did not result in any significant changes to mechanical properties as summarized in Fig. 3. No significant changes in grain size were observed as a result of aging for any of the evaluated feedstocks except for 16-8-2. Samples of 16-8-2 show a finer grain size after aging, which corresponds to the higher UTS and lower ductility observed in Fig. 3. Notably, the grain size reported by EBSD is based on a minimum boundary misorientation, and phase boundaries are treated as grain boundaries.

Table III Grain size (µm) of WA-DED builds measured by EBSD (mean \(\pm \) 95% confidence limits)
Full size table

Bulk Phase Content

Samples were evaluated using XRD to determine the phases present in the microstructure in the SA condition and after aging at 650°C for 1000 h. Figure 5a shows that in the SA and aged conditions, both ferrite and austenite are present in the microstructure of 16-8-2. Figure 5b shows the same diffractograms as Fig. 5a except over a narrower range of angles to highlight the peaks of minor phases. Peaks consistent with M23C6 and σ can be observed in both the solution-annealed and aged conditions. Figure 5c shows diffractograms of 316LSi before and after aging also indicating that ferrite and austenite are present in the SA and aged conditions. M23C6 and σ are also observed before and after aging in 316LSi, as shown in Fig 5d. A peak that can be attributed to M7C3 is observed in the solution-annealed condition but not in the aged condition for 316LSi. X-ray diffraction before and after aging for 316L and 316H is shown in Fig. S-3 in the Supplemental Materials. Peaks attributable to M23C6 and σ are observed before and after aging in both 316L and 316H. The relative intensities between the various peaks of the austenite and ferrite phases vary considerably between the solution-annealed and aged samples in Fig. 5 and S-3 indicating that each of the samples has different components of crystallographic texture. Therefore, definitive analysis of phase fractions in all the diffractograms was not performed. However, the XRD results do indicate that σ and M23C6 are present in the microstructures for the WA-DED samples.

Fig. 5
figure 5

X-ray diffraction patterns before and after aging for (a) 16-8-2 showing majority phases, (b) 16-8-2 showing minor phases, (c) 316LSi showing majority phases and (d) 316LSi showing minor phases

Full size image

Grain Boundary Precipitates

To better understand microstructural changes during aging, etched samples were evaluated using SEM. Secondary electron (SE) micrographs of 16-8-2 before and after aging are shown in Fig. 6a and b, respectively. In the SA condition (Fig. 6a), particles are observed dispersed throughout the microstructure, but very few particles are observed near etched features that suggest boundaries. After aging, almost complete decoration of grain boundaries with particles is observed in Fig. 6b. Similarly, particles are observed throughout the microstructure of 316LSi in the SA condition shown in Fig 6c. After aging, particles are observed along boundaries as shown in Fig. 6d. Although the SEM micrographs in Fig. 6 do not identify the phases, the results clearly indicate that new phases are forming at grain boundaries as a result of thermal aging at 650°C. A localized characterization technique focused on boundaries may provide insight into microstructural changes during the aging process.

Fig. 6
figure 6

SEM SE micrographs of (a) 16-8-2 after solution annealing, (b) 16-8-2 aged at 650°C for 1000 h, (c) 316LSi after solution annealing and (d) 316LSi aged at 650°C for 1000 h

Full size image

To better characterize the local change in the microstructure at the grain and grain boundary level after aging, EBSD was performed on samples in the SA and aged conditions. EBSD phase analysis on 16-8-2 in the SA condition detected σ and ferrite contents of 0.2 and 2.6 vol%, respectively, as shown in Fig. 7b and c. The phase fractions change to roughly 0.8 and 37 vol%, respectively, after aging for 1000 h (Fig. 7e and f). The increase in ferrite in 16-8-2 upon aging is apparent throughout the microstructure. However, the points indexed as σ are difficult to see because of the low amount, their location primarily at grain boundaries and the coarser step size of the lower magnification EBSD scan. Higher magnification EBSD analysis on etched samples revealed that the grain boundary particles observed in Fig. 6b are possibly σ and M23C6 carbides as shown in Fig. 7h and i. The local amount of σ phase near a grain boundary in Fig. 7h is considerably higher than the equilibrium prediction of 0% shown in Table II. Furthermore, the amounts of ferrite detected locally using EBSD after aging in Fig. 7e and f (36 vol%) are considerably higher than expected from Thermo-Calc predictions (9 vol%) as shown in Table II. It is possible that due to the localized nature of EBSD analysis, the amount of ferrite measured is not representative of the bulk phase content. It is also possible that local compositions near boundaries do not represent bulk equilibrium conditions. In any case, the amount of ferrite in the microstructure of 16-8-2 increases as a result of aging at 650°C for 1000 h. Comparing the IPF maps before and after aging (Fig. 7a and d) shows how ferrite formation leads to the refinement in effective grain size shown in Table III. It is likely that the formation of ferrite as a result of aging is related to the increases in strain-hardening rate, increases in UTS and decreases in ductility observed after aging in Fig. 3.

Fig. 7
figure 7

EBSD results of 16-8-2 (a) inverse pole figure (IPF) and image quality (IQ) map of the solution annealed (SA) condition, (b) phase + IQ map of SA condition, (c) SA condition phase analysis, (d) aged condition at 650°C for 1000-h IPF + IQ map, (e) aged condition at 650°C for 1000-h phase + IQ map, (f) aged condition at 650°C for 1000-h phase analysis, (g) aged condition at 650°C for 1000-h higher magnification IPF + IQ map, (h) aged condition at 650°C for 1000-h higher magnification phase + IQ map, (i) aged condition at 650°C for 1000-h phase analysis. The IPF + IQ maps are oriented to the build direction shown in the bottom right corner of the figure

Full size image

EBSD analysis of 316LSi before and after aging is shown in Fig. 8. Phase maps detected local amounts of σ phase and ferrite of approximately 0.4 and 0.8 vol%, respectively, in the solution-annealed condition, as shown in Fig. 8b and c. The local σ content increased to approximately 2.6 vol% and ferrite decreased to roughly 0.4 vol% after aging for 1000 h (Fig. 8e and f). Similar to the result for 16-8-2 shown in Fig. 7, the regions identified as σ are difficult to see in the phase maps for 316LSi shown in Fig. 8e and f because of the low phase amount and location at boundaries. A higher magnification EBSD scan on an etched sample focusing on grain boundaries in 316LSi after aging is shown in Fig. 8h, where σ and a small amount of M23C6 can be seen at grain boundaries, thus providing some insight into the identity of the particles observed in the SEM micrographs in Fig. 6d. The measured local amounts of σ phase from EBSD are generally smaller than the amount predicted at 650 °C (14 vol% shown in Table II), but the increase in σ with aging is consistent with the predicted change in phase amounts toward equilibrium. Similarly, the decrease in ferrite content in 316LSi with aging is consistent with an approach toward the equilibrium value of 0% ferrite at equilibrium at 650°C as shown in Table II.

Fig. 8
figure 8

EBSD results of 316LSi (a) SA condition IPF + IQ map, (b) SA condition phase + IQ map, (c) SA condition phase analysis, (d) aged condition at 650°C for 1000-h IPF + IQ map, (e) aged condition at 650°C for 1000-h phase + IQ map, (f) aged condition at 650°C for 1000-h phase analysis, (g) aged condition at 650°C for 1000-h higher magnification IPF + IQ map, (h) aged condition at 650°C for 1000-h higher magnification phase + IQ map, (i) aged condition at 650°C for 1000-h phase analysis. The IPF + IQ maps are oriented to the build direction shown in the bottom right corner of the figure

Full size image

EBSD results for 316L and 316H are shown in Fig. S-4 and S-5, respectively, in the Supplemental Materials. Similar results are observed in 316L and 316H compared to 316LSi. After aging, M23C6 and σ are observed at boundaries, with 316H showing a greater amount of M23C6 compared to 316L and 316LSi. Generally, 16-8-2, 316LSi, 316L and 316L all show elongated austenite grains parallel to the build direction in the solution-annealed condition as well as the aged condition. Negligible changes in austenite grain size are observed after aging, but precipitation of M23C6 and σ are observed to various degrees on the four alloys. Although Laves and η were not explicitly identified in XRD or EBSD, it is possible these phases are present but are too fine and low in amount to be detected. Notwithstanding the minor variations in M23C6 and σ among the four alloys, 16-8-2 shows the greatest change in microstructure after aging at 650°C. Considerable amounts of acicular shaped ferrite form during the 1000-h aging period at 650°C as can be seen in Fig. 7b and e.

It is recognized that EBSD should be used with caution to quantify the number of small particles at boundaries. For Fig. 7 h, 8 h, S-4e and S-5e, 10 s of adjacent pixels were confidently indexed as σ and M23C6, providing reasonable evidence that these phases are present at boundaries after aging. Although EBSD cannot be considered a definitive phase identification method, the combination of corroborating results from XRD for bulk analysis and EBSD for local analysis does give reasonable evidence that σ and M23C6 are forming during high temperature exposure at 650°C.

Fractography

Clear microstructural evidence was observed indicating that ferrite formation as a result of aging is associated with the increase in strain-hardening rate, increase in UTS and decrease in ductility in 16-8-2. Fractography was performed on the 16-8-2 tensile samples tested at room temperature in the SA and aged conditions. Fractography was also performed on the SA and aged 316LSi samples to evaluate the fracture mechanisms in a condition that showed negligible changes in ductility after aging. A summary of the fractographic results is shown in Fig. 9. Fracture surface analysis revealed that micro-void coalescence is the dominant fracture mechanism in all solution-annealed and aged samples in 16-8-2 and 316LSi as shown in Fig. 9c, f, i and l. Subtle changes were observed in both 16-8-2 and 316LSi samples after aging. All aged samples tested at room temperature displayed secondary cracks perpendicular to the primary fracture surface as shown in Fig. 9d, e, f, j, k and l. As Fig. 9f and l shows, micro-void coalescence is present adjacent to the secondary cracks, but it is not clear if the secondary cracks themselves have micro-voids on their fracture surfaces. Similar fractographic results were observed in 316H samples. However, 316L samples in the SA and aged conditions tested at room temperature displayed no secondary cracks as shown in Fig. S-6 in the Supplemental materials. These findings suggest that the secondary cracking alone cannot account for the large decrease in ductility in 16-8-2 and the minor (and statistically insignificant) decreases in ductility in the other conditions. The formation of ferrite as a result of the aging process is likely the dominant contributor to the decrease in ductility observed after aging in 16-8-2.

Fig. 9
figure 9

SEM fractography of failed tensile samples; (a), (b) and (c) 16-8-2 solution annealed samples tested at room temperature; (d), (e) and (f) 16-8-2 aged at 650°C for 1000 h tested at room temperature; (g), (h) and (i): 316LSi solution-annealed samples tested at room temperature; (j), (k) and, (l) 316LSi aged at 650°C for 1000 h tested at room temperature

Full size image

The presence of secondary cracks after aging indicates a reduction of transverse ductility that may suggest further embrittlement may develop with increased aging time at 650°C. Because tensile samples were oriented horizontally in each of the builds (perpendicular to the build direction), the elongated grains in the microstructure were oriented perpendicular to the tensile axis of the samples. This may suggest that the secondary cracking is along the columnar grain boundaries that are decorated with secondary phases. It is certainly feasible that if more grain boundary precipitation occurs compared to what was observed after 1000 h (Fig. 6), greater reductions in tensile ductility may result.

Discussion

Comparison of WA-DED Aging Behavior to Wrought Stainless Steels

The various aspects of aging behavior observed in this work are generally in alignment with those reported for wrought material or welds in the literature. The hardness of WA-DED 316L, 316LSi and 316H samples showed an initial increase and then a decrease during aging at 650°C for 1000 h as shown in Fig. 2. Similar results were observed in gas tungsten arc welds and heat-affected zones in 316L aged at 400°C for up to 5000 h.9 The fact that both gas tungsten arc welds and the heat-affected zones experienced the aging-hardening response suggests that the 316L composition is unstable regarding secondary phase formation during elevated temperature exposure. Certainly, microstructural features in welds or WA-DED microstructures such as retained skeletal δ-ferrite/austenite interfaces can provide nucleation sites for σ phase and thus accelerate precipitation kinetics,16 but austenite grain boundaries can also serve as nucleation sites as shown in Fig. 6. It is likely that the welds contained more skeletal ferrite because no solution heat treatment was discussed compared to the WA-DED builds presented here. The greater concentration of σ nucleation sites from more skeletal ferrite could possible accelerate σ formation, but more in-depth analysis would need to be performed to confirm this.

The tensile properties of WA-DED 316L, 316LSi and 316H showed very minor changes after aging at 650°C for 1000 h. Byun et al. aged wrought 316L and 304L at temperatures between 290 – 400°C for 1500 h and observed negligible changes in tensile properties or Charpy impact toughness behavior.26 The aging treatments employed by Byun et al. are at lower temperatures and longer times than employed in this work, but they provide a reference for the aging behavior of wrought material. On the other hand, Allahyari et al. aged gas tungsten arc welds of 316L between 550 – 850 °C for up to 100 h and found a similar aging response where the strength increased then decreased with increasing time.12 The increases in strength corresponded to decreases in ductility and Charpy impact energy and correlated to the transformation of retained δ-ferrite to σ phase. The aging treatments performed by Allahyari et al. can be considered more extreme than the conditions employed here based on the almost complete transformation of δ-ferrite to σ phase in the dendrite cores of the 316L welds. However, the results from Allahyari et al. do suggest that the aging response observed here is likely related to the formation of secondary phases during aging as suggested. Although WA-DED 316L, 316LSi and 316H showed very minor changes in tensile properties in the aging conditions employed in this work, further aging could result in additional decreases in tensile ductility and toughness provided sufficient time is given for more secondary phase (e.g.σ) formation.

Phase Stability of 16-8-2

The formation of ferrite after elevated temperature exposure of 16-8-2 welds has been observed previously, in alignment with the results reported here. Fink et al. found that gas metal arc welds using a 16-8-2 filler on 304H showed increases in the ferrite number (related to the phase amount of ferrite) evaluated using magnetic measurements after aging at 705 °C for up to 170 h.27 It was suggested that the increase in ferrite number was related to martensite formation during cooling from the aging temperature. Reductions in Charpy impact energy were reported and corresponded to increases in ferrite content during aging. Swanepoel et al. also observed increases in ferrite contents in gas metal arc welds using 16-8-2 filler to weld 304H base metal after aging at 750 °C for up to 3500 h.28 It was also suggested that the increase in ferrite content was due to martensite formation during cooling from the aging treatment, although the authors noted that predictions of martensite start temperature for 16-8-2 would not predict martensite formation in this alloy. It was suggested that carbide precipitation during aging locally changes the austenite composition such that the martensite start temperature is elevated and martensite can form upon cooling to room temperature. Similar to the results from Fink et al., decreases in Charpy impact energy were reported with increasing aging time and increasing ferrite content.28 The results presented here can neither explicitly confirm nor dispute the suggestion that the increase in ferrite content after aging is a result of martensite formation during cooling. The morphology of ferrite observed after aging in WA-DED 16-8-2 shown in Fig. 7d and e is acicular, which would be expected from martensite. However, some amount of ferrite (~ 9 vol%) is predicted to be stable at the aging temperature of 650 °C, so ferrite could also have formed isothermally during aging. An in situ technique that can detect ferrite formation during aging and/or during cooling, such as dilatometry, would be needed to identify when ferrite forms in the WA-DED 16-8-2 samples. The fact that 16-8-2 shows considerably higher strain hardening during tensile deformation at room temperature compared to all of the other alloys investigated could point to strain-induced martensite formation as discussed by Swanepoel et al., but further microstructure characterization would be needed to confirm this possibility.

Neither Swanepoel et al. nor Fink et al. reported the formation of σ during aging, whereas low-intensity XRD peaks that can be attributed to σ as well as EBSD phase analysis indexing σ at boundaries in the aged WA-DED 16-8-2 samples are reported here. Early work by Leitnaker suggested that a critical content of carbon and nitrogen is required to prevent the formation of σ from δ-ferrite in 16-8-2 during aging, although no σ was observed in the aging experiments. Rather, M23C6 and/or \(\chi \) was observed in 16-8-2 depending on variations in composition, aging temperature between 649 and 732 °C and time between 500 and 10,000 h.29 It is possible that the composition locally at boundaries in the WA-DED 16-8-2 samples in this work allows for σ formation. It is also possible that the particles observed at boundaries in the WA-DED 16-8-2 samples reported here are \(\chi \) or another related phase. Compositional similarities with between \(\chi \) and σ (both phases are Fe, Cr and Mo based)30 and the general low volume fractions and small sizes of the grain boundary precipitates make definitive characterization difficult. More sensitive characterization techniques than employed in this work (such as transmission electron microscopy) would be needed to definitively identify the grain boundary precipitates.

Outlook for WA-DED Stainless Steels Used for Long Term High Temperature Service

The results of this work suggest that for austenitic stainless steels such as 316L, 316LSi and 16-8-2, the use of single-pass or weaving bead build strategies does not significantly effect changes in mechanical properties that occur over a 1000-h exposure at 650 °C. That is not to say that differences in behavior would not be observed upon longer term aging when closer-to-equilibrium amounts of σ form. However, very subtle differences in grain size were observed between single pass and weaved beads, which may indicate a wide range of build strategies provide sufficient mechanical performance for high temperature service.

Thermodynamic predictions suggest that up to 14 vol% of σ phase should form at 650 °C given sufficient time. Although no statistically significant changes in tensile properties were observed in 316L, 316LSi and 316H that all formed σ at grain boundaries after aging, it is possible the longer term aging may lead to measurable decreases in tensile ductility, as suggested by the literature regarding aging of 316L welds. Although toughness was not evaluated in this work, an evaluation of toughness would likely be more sensitive to the changes in grain boundary precipitation and resulting decreases in transverse ductility observed in fractography compared to quasi-static tensile testing presented here.

The hints of possible impending embrittlement observed after aging at 650 °C for 1000 h indicates that more work is needed to evaluate the long-term microstructure and mechanical property stability of commercial austenitic stainless steel wire feedstocks used in WA-DED for high-temperature structural applications, particularly focusing on longer aging times. Evaluating mechanical properties such as impact toughness, in addition to tensile testing, should be employed to better detect the onset of embrittlement. What is certainly clear from the results presented here is that the formation of ferrite (or possibly martensite) in 16-8-2, an alloy that was designed to avoid embrittlement from σ phase formation, causes statistically significant decreases in tensile ductility at room temperature after aging. Future feedstock selection or alloy design efforts for WA-DED austenitic stainless steel feedstocks used in power generation applications must consider the stability of the microstructure regarding phases typically associated with embrittlement (σ, η, Laves) as well as ferrite and/or martensite.

Conclusions

Commercial austenitic stainless steel wire feedstocks of 316L, 316LSi, 316H and 16-8-2 were used to create WA-DED builds, then solution annealed and subjected to aging at 650 °C for up to 1000 h. Tensile testing was conducted to evaluate the effects of aging on mechanical behavior evaluated at room temperature and 650 °C. Microstructure characterization and fractography were conducted to understand changes in microstructure and fracture mechanisms that occurred during the aging process to interpret variations in tensile properties observed. Based on the results, the following conclusions can be made:

  • WA-DED builds made from 316L, 316LSi and 316H showed negligible changes in strength or ductility as a result of aging at 650 °C for up to 1000 h.

  • There was not a statistically significant effect of build strategy (4 straight passes vs. 3 weaved passes per layer) on tensile properties, evaluated at room temperature and 650 °C, in the solution-annealed or aged condition.

  • Samples of 16-8-2 show statistically significant increases in ultimate tensile strength and decreases in total elongation after aging when tested at room temperature. No statistically significant changes were observed after aging when testing was conducted at 650 °C.

  • 316L, 316LSi, 316H and 16-8-2 show evidence of grain boundary precipitation of phases that are consistent with M23C6 and σ, although the amounts of each phase were not quantified.

  • 16-8-2 forms significant amounts of ferrite or martensite as a result of aging at 650 °C. Although some ferrite is stable in 16-8-2 at 650 °C, it is not clear from this work whether ferrite formed during aging and/or martensite formed during cooling from the aging temperature.

  • Fractography shows that all feedstock materials develop secondary cracks on the fracture surface after aging and tensile testing at room temperature. The presence of secondary cracks indicates reductions in transverse ductility that may be associated with grain boundary precipitation, but it cannot explain the decrease in room temperature tensile ductility in 16-8-2.

  • The formation of ferrite or martensite in 16-8-2 from aging at 650 °C for 1000 h is likely a major contributor to the increase in strain-hardening rate, increase in ultimate tensile strength and decrease in ductility observed during tensile testing at room temperature.

  • Future alloy selection or design efforts for high temperature structural applications must consider the formation of phases typically associated with embrittlement, such as Laves, ηand σ, as well as the formation of ferrite and/or martensite to ensure microstructure and mechanical property stability during long times at elevated temperatures.