748 K (475 °C) Embrittlement of Duplex Stainless Steel: Effect on Microstructure and Fracture Behavior

The surface of the specimens was prepared by grinding to 4000 grit, followed by fine polishing using 3-, 1-, 0.25-, and 0.1-µm diamond paste, finalized with an OP-S active oxide suspension polishing containing 40-nm colloidal silica particles (pH 9-10). Microstructure characterization and microhardness values were obtained from these surfaces. For high-resolution scanning electron microscopy (SEM) observations, an FEI Magellan high-resolution SEM was used in through-the-lens (TLD) imaging mode using 2 kV and 25 pA. For representation of a macroscopic view of the as-received duplex microstructure, one specimen was prepared for analysis using a Zeiss optical microscope. The specimen was ground and polished to 0.25 µm and chemically etched with Groesbeck reagent (100-mL water, 4-g NaOH, 4-g KMnO4; room temperature), which stains the ferrite.

Macrohardness measurements were carried out on a Vickers macrohardness device (Georg Reicherter Briviskop 187.5) with a load of 30 kg (HV30). For each sample, 10 hardness indentations were made and the arithmetical mean with standard deviation calculated. All samples had 600-grit SiC paper ground surface. Macrohardness testing was augmented by microhardness measurements to obtain information about the behavior of individual microstructure constituents with 748 K (475 °C) embrittlement treatment.

Tensile tests were performed to obtain mechanical properties of flat tensile specimens, with parameters determined as 0.2 pct offset yield stress, R _p0.2, and ultimate yield stress, R _m. A mill-annealed sheet 22Cr-5Ni DSS was used for all tensile tests with a composition of 22.44Cr, 5.75Ni, 3.32Mo, 1.41Mn, 0.42Si, 0.015C, 0.155N, 0.006Nb, 0.21Cu, 0.12Co, and Fe (bal.). Tensile tests were performed on an Instron 5569 tensile testing machine with an extensometer from MTS. The strain rate was 2 mm/min. Tensile specimens, with 50-mm gage length, 10-mm gage width, and a total length of 80 mm, were machined from 2-mm-thick grade 22Cr-5Ni DSS sheet. The fracture surfaces of the as-received and 20-, 50-, and 255-hour aged conditions were analyzed using the FEI Quanta 650 SEM. The fracture area and degree of embrittlement were quantified using the ImageJ image processing programme (https://​imagej.​nih.​gov/​ij/​).

The microstructure of the as-received condition in the rolling direction (RD) is shown in Figures 1 and 2 for the plate and sheet materials, respectively. Ferrite (δ-bcc) and austenite (γ-fcc) phases with typical ferrite/austenite grain boundaries can be seen, whereas the ferrite/ferrite and austenite/austenite boundaries are not visible. The phase ratios of austenite to ferrite (as-received) were previously reported as 56:44 and 54:46 for the plate and sheet materials, respectively.[35,36] The grain size of the plate material in the RD of ferrite was 10.4 ± 3.2 µm, while the grain size of austenite was 10.7 ± 3.6 µm. The average aspect ratio (horizontal-to-vertical mean length) was 2.5 for ferrite and 1.9 for austenite. The grain size of the sheet material in the RD of ferrite was 4.2 ± 2.3 µm, while the grain size of austenite was 2.2 ± 1.4 µm. The sheet material was expected to behave in a similar manner to the plate.

More detailed microstructure information of the austenite and ferrite was obtained through TEM analysis (Figure 3). In the as-received condition, austenite grains were characterized by the presence of stacking faults and some dislocations, consistent with a recrystallized microstructure. The ferrite contained dislocation loops, consistent with a deformed or recovered microstructure. This indicates the presence of residual strain from the thermomechanical process history of the as-received material. No crystallographic phases other than the austenite and ferrite were detected in the as-received material.

The sample aged for 5 hours had a mottled contrast in the ferrite, most likely associated with spinodal decomposition products, as shown in the high-angle annular dark-field (HAADF) image in Figure 4. The mottled contrast is related to compositional differences between Fe-rich α′ and Cr-enriched α″. This is in line with the work of Weng et al.[16] who reported spinodal structures of 1 to 2 nm in the ferrite of 22Cr-5Ni DSS after 2 hours of aging at 748 K (475 °C). No other phases were observed.

The microstructure of the specimen aged for 20 hours was characterized by the presence of dislocation structures consisting of numerous dislocation loops in the ferrite with dislocation forests on ferritic subgrain boundaries, all indicating high dislocation activity. These can be seen in the bright-field TEM images in Figure 5(a) and (b). These loops were still present after the aging process. Small discrete realms of strain contrast, possibly associated with precipitation incubates, were observed in the ferrite, which can be seen in Figure 5(b). The mottled contrast in the ferrite associated with spinodal decomposition products became more apparent in comparison to the specimen aged for 5 hours. This is an indication of enhancement of the amplitude of spinodal decomposition (i.e., distance of Cr-enriched α″ and Fe-rich α′ regions) in ferrite, which can be seen in Figure 5(c). This observation is in good agreement with the work of Weng et al.[16] who reported spinodal decomposition of 2- to 4-nm size after 16 hours of aging at 748 K (475 °C) of a 22Cr-5Ni DSS. In austenite, numerous stacking faults and a deformation structure were observed, as shown in Figure 5(d).

The microstructure after 50 hours of aging contained numerous R-phase precipitates, which could already be seen in the SEM, as shown in Figure 6. The sizes of these precipitates varied between 20 and 50 nm and were located within ferrite grains only. Further secondary phases seemed to coexist with the R phase along interphase boundaries.

Significant microstructure changes were evident after aging at 748 K (475 °C) for 255 hours. The microstructure containing decomposed ferrite with numerous precipitates is shown in Figure 7. The majority of these precipitates could be ascribed to R phase. R-phase precipitates showed disc or lenticular shape in sizes between 50 and 400 nm and nucleated at inter- and intragranular sites in ferrite grain interiors and at ferrite grain boundaries, as shown in Figure 8. Some ferrite regions were enriched in R precipitates, where high-angle grain boundaries (HAGBs) and low-angle grain boundaries (LAGBs) seemed to function as preferential nucleation sites. R-phase precipitates arrayed along slip planes and LAGBs can be seen in Figures 7 and 8. The formation of R was heterogeneous, and discrete precipitation occurred in certain ferrite regions. A precipitation density of 13 particles per square micrometer, on average, in the ferrite was calculated using Image J software from TEM images. Since precipitation was heterogeneous, the particle density can be significantly higher in regions containing more precipitates. No phase transformation products could be observed in the austenite. These observations are in good agreement with Nilsson and Liu,[19] Redjaimia et al.,[12,37] and Karlsson et al.[21] who observed R-phase precipitates at 823 K (550 °C), 873 K (600 °C), and 948 K (675 °C), respectively. They also found Mo- and Si-depleted zones at and adjacent to R precipitates.

Higher magnification TEM images of a region containing R-phase precipitates revealed stronger diffraction contrast than the surrounding ferrite matrix, as shown in the BF- and DF-TEM images in Figure 9. R precipitates seemed to be heavily faulted and had lenticular morphologies, in good agreement with previous studies.[12,19,21, 37] Spinodal decomposition in ferrite became more apparent in SEM analyses, as shown in Figure 10. This mottled structure is the intertwined, three-dimensionally interconnected Cr-rich α″ and Fe-rich α′ phase, which was confirmed by TEM diffraction analysis. The bright appearing features are Cr-Mo–enriched α″. They protrude out of the Fe-enriched α′ matrix. The spinodal arm width of α″ was determined to be 15 to 30 nm, and the ferrite seemed to have decomposed entirely.

The hardness test results are summarized in Figure 11. A slight increase in hardness with aging up to 50 hours was observed, with a significant rise in macrohardness in the specimen aged for 255 hours, clearly indicating the effect of 748 K (475 °C) embrittlement. The microhardness of the ferrite followed a similar trend, except for the 20-hour aged specimen, with maximum hardness observed after 255 hours. Phase transformation reactions occurring in ferrite resulted in a hardness increase. The hardness of austenite remained constant between a minimum of 263 HV0.01 and a maximum of 290 HV0.01. This indicates that no significant microstructure change occurred that could have influenced the hardness values in austenite. The hardness increase of the ferrite is due to the mismatch of the elastic moduli and lattice parameters between the components of the decomposed ferrite requiring more stress for dislocation mobility.

Bulk hardness delivered more global information, whereas microhardness gave local information of each microstructure constituent. The bulk hardness is a convolution of the hardness of both ferrite and austenite phases, whereas the microhardness data for each phase gives local information. Microhardness measurement, however, gives local hardness of the microstructure and is a convolution of grain orientation and precipitation heterogeneity, which causes different hardness values within the ferrite. It is evident that the scattering for austenite is less than that for ferrite, supporting this statement. It has become evident that the ferrite is affected by aging at 748 K (475 °C), whereas the austenite seemed to not show a significant change in local hardness. Precipitates mainly caused a hardness increase in the ferrite, which was more apparent after 255 hours of aging.

All tensile test results for the as-received and 20-, 50-, and 255-hour aged specimens are shown in Figure 12. The elongation to failure (A) halved after 20 hours aging from 50 ± 5 to 28 ± 6 pct, showing clear signs of embrittlement. The largest embrittlement was observed in the 255-hour aged specimen, with only 12 ± 4 pct of elongation to failure. The yield stress, R _p0.2, increased to 735 ± 35 MPa after 255 hours of aging. The ultimate tensile strength rose from 830 ± 20 to 1030 ± 20 MPa after 255 hours of aging. Precipitates and spinodal decomposition can impede dislocation movement and, hence, the onset of plasticity. This leads to an increase in yield strength, which is associated with a decrease in toughness. Aging at 748 K (475 °C) caused microstructure strengthening at the expense of increasing embrittlement.

The observed embrittlement was also manifested by a large drop in uniform and total elongation, evident in the stress-strain behavior of all aged samples. The 255-hour aged sample, for example, showed immediate failure upon reaching ultimate tensile stress, as shown in Figure 13. Both uniform and total elongations are a measure of material ductility and formability. Both uniform and total elongations were decreased sharply by aging time, with short-term aging showing a much steeper drop in ductility. This is evidence that short-term aging can already significantly deteriorate material ductility. Spinodal decomposition seems to have a larger impact on ductility than on hardness, since the hardness of both ferrite and austenite was not affected during short-term treatments at 748 K (475 °C) in DSS.

Spinodal decomposition and precipitation in the ferrite phases seems to be primarily responsible for most mechanical property degradation. This is because ferrite is the matrix phase in the duplex microstructure and, hence, the main load carrying element.

The fracture surfaces of all tensile specimens are shown in Figure 14. The cross-sectional area of the fracture surface increased from 5.2 to 9.1 mm² after 255 hours of aging, as summarized in Figure 13. This is related to a reduced ductility, resulting in earlier fracture with less necking. The largest necking was observed on the as-received condition, which gradually decreased with aging time, confirming embrittlement. There is a large loss of ductility associated with heat treatment at 748 K (475 °C), as shown in Figure 12.

The fracture surface of the as-received microstructure is given in Figure 15. Most necking was observed in the as-received condition, indicating the highest ductility and toughness among all heat-treated specimens. Microvoid coalescence was the main fracture mode, as visible from prominent cup-and-cone-shaped structures, indicating ductile fracture. These dimples were observed all over the fractured surface, indicating ductile rupture behavior of both ferrite and austenite grains. Most dimples were equiaxed, typical for strain acting normal to the fracture surface. Elongated dimples were also observed, particularly in the vicinity of the main crack (Figure 15(e) and (f)) and in regions showing signs of cleavage fracture (Figure 15(c) and (d)). Elongated dimples typically indicate shear deformation, clearly evidencing ductile fracture behavior. These observations are in good agreement with Pettersson et al.[38] and de Lacerda et al.[39] who also reported ductile fracture surface for DSS in the as-received condition (solution annealed). Easy slip activity from ferrite and austenite, and vice versa, is reported for as-received duplex microstructure.[11] Austenite accommodates most plastic strain due to its closely packed slip systems. However, the austenite is not continuous, so the ferrite can limit slip activity, which requires slip interchange between austenite and ferrite. Plastic deformation of ferrite usually results in deformation twins, which lead to high stored energy in the microstructure and to cleavage rupture.[16,38,40]

The fracture surface of the 20-hour aged specimen is shown in Figure 16. Ductile fracture by microvoid coalescence governed the fracture surface, with some regions, indicating quasi-cleavage. Most microvoids were nonequiaxed, and numerous cavities were seen decorating the fracture surface, indicating loss of constraint and delamination rupture along interphase boundaries. Loss of toughness and ductility associated with 748 K (475 °C) embrittlement was reported due to the inability of the ferrite to form deformation twins.[38,41] Spinodal decomposition in the ferrite retards slip activity requiring more energy for slip interchange between ferrite and austenite. Spinodal decomposition is reported to be associated with higher elastic strain energy due to different atomic sizes between iron and chromium atoms.[11] Hence, the reduced ability to plastically deform ferrite resulted in austenite being constrained, leading to less ductile fracture. Decoherence between ferrite and austenite grains led then to delamination and cleavage fracture.

The fracture surface of the 50-hour aged condition can be seen in Figure 17. More extensive cracking was observed with significantly less necking and less dimples, all indicating embrittlement. Cleavage fracture areas were observed, which indicated strain heterogeneity in the microstructure. R-phase precipitates and further secondary phases in the ferrite most likely led to enhanced slip retardation affecting macroscopic fracture appearance. Hence, slip incompatibility at the interphase boundaries may have resulted in nonuniform distribution of surface damage. Slip bands were seen in certain elongated dimpled regions of austenite grains, as shown in Figure 17(d), most likely indicating slip incompatibility at interphase boundaries, resulting in easier delamination rupture along ferrite-austenite grain boundaries. Large cavities (Figure 17(e)) and heterogeneous fractured regions (Figure 17(f)) were observed, indicating loss of strain homogeneity and local brittleness. Such sites were most likely associated with high strain fields, in which austenite and ferrite slip systems had low Schmid factors.

The most severe fracture was seen on the specimen aged for 255 hours, as shown in Figure 18. Extensive cracking was noticeable with little necking, indicating the highest 748 K (475 °C) embrittlement among all heat treatment conditions. Long and wide cracks were seen with fibrous appearance, defined as delamination fracture, indicating preferential rupture along interphase boundaries. Similar observations were reported by Pettersson et al.[38] and Pilhagen et al.[42] who investigated the effect of aging in the 748 K (475 °C) temperature window on the fracture behavior of DSS. They observed delamination cracking associated with enhanced slip incompatibility between ferrite and austenite in combination with cleavage fracture. However, the fracture mechanism was not explained. In our case, ferrite grains containing numerous slip bands are shown in Figure 18(c) and (d). Such features were seen on ferrite grains neighboring austenite grains and along the microstructure processing direction (RD) only. Cracks with wide crack openings were seen along interphase boundaries showing enhanced phase mismatch, as shown in Figure 18(e). In certain austenite regions, transgranular cleavage tongues were present (Figure 18(e)), indicating the loss of strain homogeneity and enhanced local embrittlement. This showed loss of ductility of the austenite, supporting the overall embrittlement phenomenon, which is in line with the work of Pettersson et al.[38] who observed similar cleavage tongues on 748 K (475 °C)—embrittled fractured 2507 super DSS. Cleavage tongues are signs of a decrease of the overall fracture energy.[38]

Marrow and Harris[43] studied the fracture behavior of 748 K (475 °C)–embrittled Zeron 100 super DSS and concluded that the brittle fracture of ferrite is twin nucleated. Thus, twinning is necessary for ferrite cleavage and crack initiation is controlled by the critical shear stress, which decreases with aging. However, when the critical shear stress is higher than the shear stress for twin or slip band nucleation and yield, fracture occurs with no or minor plastic deformation. There is strain and stress partitioning among both crystallographic phases in duplex steels.[44] Stress partitioning to the ferrite occurs by plastic deformation of the softer austenite, which in turn increases the stress in the ferrite.[43,44] The internal friction of ferrite is affected by the 748 K (475 °C) embrittlement heat treatment, whereas the friction stress of austenite is believed to be nonaffected by age hardening. Therefore, long-term aged microstructures were more affected due to the increased differences of the flow stresses of both ferrite and austenite. Marrow and Harris[43] argued that 748 K (475 °C) embrittlement is not accompanied by the precipitation of a brittle phase and neglected secondary brittle phase formation in the austenite. There is a high fraction and density of precipitates in the ferrite (Figure 7). The lattice parameters of the R phase are approximately seven times larger than those of the ferrite,[37] and the R phase is considered as the harder phase, so that dislocation cutting through the particles is not possible, for larger particles at least. Dislocation bowing around these phases would be the hardening mechanism in the ferrite. Therefore, the fracture mechanism of long-term embrittlement DSS, containing high fractions of precipitates, would be precipitation hardening controlled.

Fine-scale dimples in ferrite grains were also seen, which indicated rupture. These dimples were associated with the large number of precipitates in the ferrite causing microvoid coalescence and may have resulted in inter- and transgranular cleavage. There is a clear change of fracture mechanism noticeably associated with 748 K (475 °C) embrittlement. Austenite, in contrast, fractured in a ductile manner and seemed to accommodate most plastic deformation. Hence, ferrite alone can be made responsible for the 748 K (475 °C) embrittlement of 2205 DSS.

Analysis of the cross-sectional fracture surface of the as-received specimen showed extensive necking, with large local plastic deformation at the ruptured fracture surface regions. The fracture surface had a smooth appearance with both ferrite and austenite showing extensive deformation at the rupture area, shown in Figure 19(a) and (b). Both phases accommodated large plastic strains prior to rupture, indicating that ample cross-slip of dislocations among both phases was possible. Despite the extensive strain hardening at the necked area, a broad zone of bent and elongated austenite and ferrite grains was observed on the fracture surface, indicating ductility. In contrast, the fracture surface of the specimen aged for 255 hours had a zigzag pattern with very little necking. Almost the entire population of ferrite grains fractured at an angle of ~45 deg, with respect to the tensile direction, indicating brittle failure. This also indicated that almost no yielding occurred and that fracture occurred after little or no yielding. The ferrite seemed not to show the classic fracture shape, in which necking followed by a fibrous fracture occurs, whereas the austenite showed some necking, indicating ductile behavior. However, there was also a minor loss of ductility noticed in the austenite associated with 748 K (475 °C) embrittlement. The fracture of the ferrite was transgranular, showing a similar typical appearance to low-temperature cleavage rupture of nonembrittled steels. Thus, the ferrite was the phase mainly responsible for the brittle fracture.

The fracture area increased and the reduction in area decreased with increasing aging time. There seemed to be a linear relationship between fracture area and aging time at 748 K (475 °C) and likewise for the reduction in area (Figure 13). The reduction in area on the as-received material was 74 pct, whereas after 255 hours of aging, a final reduction in area of 54.5 pct was measured. The effect of aging on the reduction in area was less than on uniform and total elongation, which means that material deformation in cross-sectional areas was more pronounced than along the tensile direction.

Tensile properties are seemingly enhanced, and most plastic deformation was observed perpendicular to the loading direction, causing large contraction and perhaps even local compression of the ductile austenite grains. Forged DSSs typically have, due to their processing history, nonequiaxed grain structure with elongated or flattened grain shapes. Mechanical properties, accordingly, are dependent upon the microstructure processing orientation. Decomposition products of ferrite reduced the ability for uniaxial tensile deformation; therefore, plastic deformation concentrated on the austenite, acting perpendicular to the loading direction.

In summary, the fracture behavior of 748 K (475 °C)–embrittled 22Cr-5Ni DSS is controlled by precipitation hardening occurring in the ferrite phase, not only by spinodal decomposition. This is clearly noticeable from the hardness trend and fracture behavior, as shown previously. The phase mismatch between ferrite and austenite increases with age hardening; hence, their flow stress increases, causing delamination fracture.