2024 Volume 64 Issue 2 Pages 474-481
The hydrogen desorption behaviors of the SUS304 and SUS316L austenitic stainless steels during deformation at ambient temperature were investigated using a tensile test machine in a vacuum chamber equipped with a mass spectrometer. Obvious hydrogen desorption was detected only in the SUS304 steel, which exhibited a distinct martensitic transformation. Because the hydrogen desorption rate in SUS304 decreased when deformation stopped, a significant factor causing transformation-induced hydrogen desorption was an increase in martensite fraction during plastic deformation. Furthermore, hydrogen promoted both martensitic transformations to ε and to α′, which assists the hydrogen desorption. These results indicate the presence of synergistic interactions between the hydrogen uptake/diffusion and martensitic transformation. In contrast, SUS316L steel showed no martensitic transformation and exhibited hydrogen-assisted deformation twinning. No significant increase in hydrogen desorption was observed during plastic deformation. This result indicates that deformation twinning has no effect on hydrogen diffusion/desorption.
Austenitic stainless steels have been widely used in structural materials applied in hydrogen energy systems and nuclear power plants owing to their high ductility and corrosion resistance.1,2,3) However, hydrogen embrittlement (HE) has become a problem, resulting in the reduction of mechanical properties when materials are exposed to a hydrogen atmosphere. Therefore, hydrogen uptake and its subsequent diffusion have often been discussed in hydrogen-related damage research.
The susceptibility to hydrogen embrittlement in austenitic steels drastically increases when deformation-induced martensitic phase transformation from the austenite (γ) phase to hexagonal close-packed (ε) phase4,5) or body-centered cubic (α′) phase occurs.6,7) In particular, the α′ phase has a higher hydrogen diffusivity compared to that in the γ phase, which results in localization of hydrogen at stress concentration sites or microstructural interface/boundary and causes associated cracking.8) Furthermore, because hydrogen solubility in α′ phase is lower than that of γ phase, the α′ phase transformed from γ phase through diffusionless mechanisms shows a supersaturation condition of hydrogen, enhancing hydrogen activity for further hydrogen localization.9,10) Therefore, it is currently required to evaluate hydrogen activity during plasticity evolution involving deformation-induced martensitic transformation. In this context, deformation twinning, which can be formed by similar mechanisms in austenitic stainless steels, is another key microstructural evolution for understanding hydrogen embrittlement behavior. The formation of deformation twins is also known to assist hydrogen-related cracking in high-Mn austenitic steels,11) but the effect of deformation twins on hydrogen embrittlement is much smaller than that of deformation-induced martensitic transformations. To tune the deformation microstructure for hydrogen resistance, the effects of deformation-induced martensite and deformation twins on hydrogen behavior have been regarded as important issues.
However, the relationship between hydrogen embrittlement and martensitic transformation to ε and α′ or deformation twinning is complicated. The complexity results from four hydrogen effects: (i) suppression of thermally induced martensite,12) (ii) stress evolution resulting in surface martensite formation during hydrogen uptake,13) (iii) promotion of deformation-induced γ-ε martensitic transformation,14,15,16,17) (iv) suppression of deformation-induced martensitic transformation to α′,14,18,19) and (v) assistance in deformation twinning.20,21,22,23,24,25) The multiple synergistic effects of hydrogen and the microstructure make quantitative evaluation difficult. Furthermore, when electrochemically hydrogen-charged, the specimen generally exhibits a heterogeneous hydrogen distribution. Therefore, the microstructural evolution behavior is dependent on the depth from the specimen surface, which requires an evaluation of the gradient of the hydrogen effects on the microstructural evolution from the surface. Hence, to evaluate the effects of the deformation microstructure on the hydrogen activity, the optimal selection of materials and hydrogen charging conditions, quantification of the local microstructure for each condition, and in situ detection of hydrogen action during deformation are required for specific specimens with identical deformation conditions.
The in situ detection of the hydrogen action involved in plastic deformation has been recognized as a difficult issue. Therefore, we monitored the hydrogen desorption behavior. In a previous study, Hojo et al.26) succeeded in detecting hydrogen desorption during deformation-induced martensitic transformation using a tensile test machine in a vacuum chamber equipped with a mass spectrometer. In this study, an in situ detection technique for hydrogen desorption was applied to austenitic stainless steels to understand the complex phenomenon of hydrogen action associated with deformation. Furthermore, deformation-induced martensitic transformation and deformation twinning were correlated with hydrogen desorption, which were analyzed by electron backscatter diffraction (EBSD) measurements coupled with controlled hydrogen charging.
In this study, SUS304 and SUS316L stainless steels were selected because they exhibit deformation-induced martensitic transformations and deformation twinning without thermally induced martensite. The SUS304 and SUS316L plates were solution-treated at 1100°C for 20 s and at 1080°C for 20 s, respectively. The samples were then quenched with water. The solution-treated plates were confirmed to show fully austenitic microstructure with equiaxed grains, as shown in Fig. 1. Their chemical composition is presented in Table 1. The specimens for the observations were firstly mechanically polished by SiC waterproof papers, and further polished with 9 and 3 μm diamond slurries and colloidal silica with a particle size of 60 nm. The microstructure was examined by scanning electron microscopy (SEM) at an acceleration voltage of 10 kV.
| C | Si | Mn | P | S | Ni | Cr | Mo | |
|---|---|---|---|---|---|---|---|---|
| 304 | 0.06 | 0.39 | 1.11 | 0.33 | 0.03 | 8.01 | 18 | 0.26 |
| 316L | 0.022 | 0.69 | 1.01 | 0.03 | 0.001 | 12.12 | 17.28 | 2.05 |
Tensile specimens with the geometries shown in Fig. 2 were prepared using spark machining. The gauge portion had a length of 8 mm, width of 3 mm, and thickness of 1 mm. The specimens were mechanically ground and electrolytically polished at 40 V for 120 s to obtain smooth surfaces with thickness of 0.9 mm. Cathodic hydrogen charging to the specimen was performed in an electrolyte of 0.1 N NaOH aqueous solution containing 3 g/L NH4SCN with a current density of −50 A/m2 at 80°C for 96 h. A Pt wire was used as the counter electrode for hydrogen charging. No martensitic transformation was confirmed to occur during hydrogen charging at 80°C both in SUS304 and SUS316L. Tensile tests were performed at an initial strain rate of 5×10−4 s−1 at room temperature in air. Diffusible hydrogen contents were determined by thermal desorption spectroscopy (TDS) operated at a heating rate of 100°C/h from ambient temperature to 800°C. The diffusible hydrogen was defined as hydrogen that diffuses at room temperature, of which the content was obtained by measuring cumulative desorbed hydrogen from ambient temperature to 500°C. The grip sections of the tensile-fractured specimens were used for TDS measurements. The diffusible hydrogen contents of the SUS304 and SUS316L specimens were measured to be 128 and 66 mass ppm, respectively. Assuming that the holes of the grip sections can be ignored, the diffusible hydrogen contents, cM, can be converted to surface hydrogen contents, c0, by using the following equation.27)
| (1) |
where w is the specimen thickness, D diffusion coefficient, and t the hydrogen charing time, respectively. The diffusion coefficients of SUS304 and SUS316L were set to 2 × 10−14 28,29) and 7 × 10−15 m2/s,30) respectively. The estimated surface hydrogen contents of SUS304 and SUS316L were 616 and 537 mass ppm, respectively.
To detect hydrogen desorption during tensile deformation, specimens with the same geometry as those in Fig. 2 were placed on the tensile test machine in a vacuum chamber equipped with a quadrupole mass spectrometer. The specimens were set to the tensile machine at ambient pressure and temperature. Subsequently, the chamber was evacuated to below 10−5 Pa. The tensile test was started after the hydrogen pressure measured by the mass spectrometer became relatively stable. The tensile tests were performed at a crosshead speed of 0.3 mm/min and stopped at a crosshead displacement of approximately 3 mm for all tests. After the deformation experiment, a cross section of the gauge part of the specimen was analyzed using SEM coupled with EBSD measurements. The specimens for the EBSD measurements were prepared by mechanical polishing using the procedure described above. EBSD measurements were conducted at an acceleration voltage of 20 kV with a beam step size of 0.2 μm.
Figure 3 shows the engineering stress-strain curves of SUS304 and SUS316L with and without hydrogen charging. As previously reported for austenitic stainless steels,31,32,33) hydrogen charging increases the yield strength of SUS304 and SUS316L steels. Hydrogen charging for SUS304 caused the transition from ductile (Figs. 4(a) and 4(a′)) to quasi-cleavage fractures (Figs. 4(b) and 4(b′)), which resulted in mechanical degradation. The fracture surface of the hydrogen-uncharged SUS316L also showed ductile feature, which is covered with dimples (Figs. 5(a) and 5(a′)). When charged with hydrogen, the SUS316L specimen showed a small portion of brittle feature at the fracture surface corner (Fig. 5(b′)). The thicknesses of the embrittled region on the fracture surfaces of hydrogen-charged SUS304 and SUS316L specimens were measured to be 114 and 39 μm, respectively. In addition, necking in both the hydrogen-charged SUS304 and SUS316L specimens did not occur due to the premature fractures, according to the comparisons of the fracture surfaces with and without hydrogen charging (Figs. 4 and 5). The tensile properties are listed in Table 2.
| Specimen | 0.2% proof stress (MPa) | Tensile strength (MPa) | Elongation (%) | Reduction in area (%) |
|---|---|---|---|---|
| SUS304 without H | 280 | 701 | 79 | 81 |
| SUS304 with H | 316 | 553 | 31 | 47 |
| SUS316L without H | 266 | 537 | 80 | 84 |
| SUS316L with H | 281 | 559 | 73 | 72 |
Figure 6 shows the hydrogen desorption curves and corresponding stress-time curves obtained from tensile tests at a constant cross-head speed in a vacuum environment. The tensile tests were stopped at a crosshead displacement of approximately 3 mm to compare the microstructures at nearly identical strains and within a range of uniform elongation. After the tests, the reductions in the thickness values at the center of the specimen width of the gauge sections of the hydrogen-uncharged SUS304 and 316L and hydrogen-charged SUS304 and 316L were 11.1%, 11.6%, 10.7%, and 11.4%, respectively. Because the shear strains and the normal strain along the width direction are regarded as zero at the center of the specimen width, the reduction in thickness values can be used as the normal strain along the tensile direction.34) The hydrogen-uncharged specimens of SUS304 and SUS316L showed no significant increase in hydrogen pressure. The hydrogen pressure measured in the hydrogen-uncharged specimens was regarded as that from the background hydrogen. When hydrogen was introduced for the SUS304 specimen, the hydrogen pressure was stable during elastic deformation, and subsequently increased after the initiation of plastic deformation. The hydrogen pressure continuously increased until the end of the deformation. The increase in hydrogen pressure indicated an increase in the hydrogen desorption rate with strain. Furthermore, the hydrogen pressure suddenly decreased when the test was stopped but was still higher than that before the deformation. In contrast, the hydrogen-charged SUS316L showed no significant change in the hydrogen pressure, even after the onset of plastic deformation.
Figures 7 and 8 show the phase maps obtained from the transverse direction (TD) of the gauge part of the specimens after the experiment shown in Fig. 6. SUS304 showed deformation-induced ε-martensite and α′-martensite (Fig. 7(a)). Because the surface was under a plane stress condition, the deformation constraint was smaller than that in the interior of the specimen. Thus, the martensite fractions of ε-martensite and α′-martensite in the near-surface region were slightly higher than those in the deeper region of the specimen (Fig. 9(a)). Hydrogen charging promoted the deformation-induced formation of both ε-martensite and α′-martensite (Fig. 7(b)) and caused surface cracking in the α′-martensite region (Fig. 7(b′)). The ε-martensite and α′-martensite fractions in the hydrogen-charged SUS304 specimen were over 15% and 30% in the near-surface region, respectively. These fractions were three times higher than those of the hydrogen-uncharged specimen, as shown in Fig. 9(a). Notably, there is a drastic decrease in the martensite fraction with increasing distance from the surface. The martensite fractions decreased to the same level as for the hydrogen-uncharged specimen when the distance from the surface was over 100 μm. Accordingly, surface crack propagation stopped in the austenite region. Here note that stress concentration at the surface cracks may have assisted the martensitic transformations. However, the upper left region of Fig. 8(b) showed no cracking and showed the promotion of the martensitic transformation. In addition, the region at a depth of around 70 μm showed no cracks because of the lower local hydrogen content. This region also showed a slight promotion of the martensitic transformations by hydrogen, as shown in Fig. 9(a). These results indicate that the promotion of the martensitic transformations is the intrinsic effect of hydrogen in the present experimental condition.
In contrast, the hydrogen-uncharged SUS316L (Fig. 8(a)) showed no martensitic transformation; instead, deformation twins were observed (Fig. 8(b)). The hydrogen charging of SUS316L (Fig. 8(c)) did not induce a martensitic transformation and increased the number density of deformation twin plates (Fig. 8(d)). Figure 10 shows the relationship between the number density of the twin plates and the distance from the surface of the SUS316L specimens. The number density of deformation twins in the hydrogen-uncharged SUS316L was stable around 2.8×10−2 μm−2 against distance from the surface. The number density of deformation twins in the hydrogen-charged SUS316L was twice as high as that in the hydrogen-uncharged specimen, particularly near the surface. The number density of twin plates decreased until about 3.6×10−2 μm−2 when the distance from the surface reached 60 μm.
Similar to the previous studies,35,36,37) the present SUS304 also showed serious hydrogen embrittlement associated with the formation of α′-martensite due to its lower solubility and higher diffusivity of hydrogen compared to those of austenite.38) To prove the significant effect of the lower hydrogen solubility and the associated situation of hydrogen supersaturation in the present condition, we note the sudden decrease in hydrogen desorption rate at the end of the deformation. Specifically, the hydrogen desorption rate was approximately halved after the deformation was stopped. Because the effect of higher hydrogen diffusivity remains even after the interruption of deformation, it cannot result in a “sudden” decrease in the hydrogen desorption rate. On the other hand, continuous plastic deformation involving a martensitic transformation is required to maintain the effect of hydrogen supersaturation. That is, the sudden decrease in the hydrogen desorption rate to half its value after the deformation ended indicates that the effect of hydrogen supersaturation on hydrogen activity is comparable to the effect of hydrogen diffusivity.
It is also noteworthy that hydrogen promoted formations of the two types of martensite: ε-martensite and α′-martensite (Fig. 9(a)). In SUS304 and similar austenitic stainless steels, the martensitic transformation occurs through the sequence of γ→ε→α′.39) In this context, Hosoya et al.40) reported that hydrogen promotes deformation-induced γ→ε martensitic transformation owing to a reduction in stacking fault energy but suppresses ε→α′ martensitic transformation when tensile deformation was provided under in situ hydrogen charging. However, the present result is contradictory to this report, that is, hydrogen promoted the formation of α′-martensite as well. To explain the contradiction, we note (1) the deformation anisotropy of the hexagonal close-packed structure of the ε martensite, (2) the martensite morphology, and (3) the difference between in situ hydrogen charging and pre-hydrogen charging. Since the non-basal slip of ε-martensite is difficult, ε-twinning or ε→α′ martensitic transformation is required to intersect ε-martensite plates for plastic deformation. Accordingly, the ε→α′ martensitic transformation occurs at the intersection of ε-martensite plates41) or the intersection of the ε-martensite plate and extended dislocation slip path.42) Therefore, when a considerable number of ε-martensite plates forms, e.g., the situation for the hydrogen-charged SUS304 in the present study, needle-like α′-martensite, which is observed as dot or square shape as two-dimensional morphology as partially observed in Fig. 7, appears in the early transformation stage to satisfy the geometrical requirement. In other words, the increase in α′ martensite fraction by hydrogen would be triggered by the promotion of γ-ε martensitic transformation and subsequent geometrically necessary shear. However, when in situ hydrogen charging is performed, further growth of α′-martensite does not occur due to the suppression effect of hydrogen. In contrast, in case of pre-hydrogen charging, hydrogen can be diffused and desorbed from the ε/α′-martensite interface through the highway consisting of the needle-like α′-martensite. According to the (2Dt)1/2 criteria where D is diffusion coefficient and t time, the diffusion distance x with 600 s at 21°C in α′ martensite is 16 μm when D is set to 2.2 × 10−13 m2 s−1 43) (Here, hydrogen is assumed to be continuously provided from ε-martensite to α′-martensite.). Therefore, at least, the highway path of α′-martensite can be active within one grain layer from the surface even at the scale of several hundred seconds. In reality, the influence of the hydrogen supersaturation condition in α′-martensite at the moment of transformation near the ε/α′-martensite interface further promotes the hydrogen diffusion and desorption, which assists the ε→α′ martensite growth. In terms of experimental fact, significant hydrogen desorption was detected when α′-martensite formed as shown in Fig. 6(a). The large hydrogen desorption even after interrupting the deformation also supports the significance of the α′-martensite highway effect. Because the remarkable promotion effect of hydrogen on martensitic transformation was observed only near the surface, it is plausible that the hydrogen desorption through the α′-martensite highway occurred significantly. The hydrogen escape from ε-martensite allows further martensitic transformation from ε to α′. When this mechanism occurs, the promotion of γ→ε martensitic transformation by hydrogen directly contributes to increasing α′-martensite fraction.
In contrast, the hydrogen-charged SUS316L showed no mechanical degradation or martensitic transformation, as reported previously44) but exhibited a small portion of quasi-cleavage features around the fracture surface corner. The cause of quasi-cleavage cracking in austenitic steels and Ni alloys has been reported to be twin-boundary cracking.45,46) Because the hydrogen diffusivity in austenitic steels is low47,48) and the cracking resistance at twin boundaries is generally higher than that in martensite, the quasi-cleavage feature can be limited to the corner of the specimen. In general, SUS316L and other stable austenitic stainless steels do not show hydrogen-related brittle cracking at ambient temperature, unlike twinning-induced plasticity steels that show numerous deformation twins.49,50,51) However, when numerous twins form, such as in single-crystal specimens with a tensile direction of <111>, type 316L also shows hydrogen-induced twin boundary cracking.45) In the present case, hydrogen charging significantly promoted deformation twins, which allowed the induction of numerous deformation twins even in the middle stage of plastic deformation (Fig. 10). Therefore, hydrogen-induced quasi-cleavage fractures due to twin formation can occur locally.
The brittle features on the fracture surfaces of the SUS304 and SUS316L were observed from surfaces to the depths of 114 and 39 μm, respectively. The diffusion distances of hydrogen from the surfaces at the present hydrogen charging time were estimated by using the equation of x = (2Dt)1/2 where the D of type 304 and 316L at 80°C were 2 × 10−14 28,29) and 7 × 10−15 m2/s,30) respectively. Accordingly, the calculated hydrogen diffusion distances for the SUS304 and 316L specimens were 117 and 69 μm, respectively. The hydrogen diffusion distance for SUS304 matches the depth of the brittle feature zone, whereas that of SUS316L does not. Furthermore, at the depth of over 100 μm, there was no promotion effect of hydrogen on martensitic transformation (Fig. 9(a)). Although the promotion effect of hydrogen on martensitic transformation was observed at the depth of around 80 μm, the martensite fraction was one third of that at around 30 μm due to the presence of hydrogen concentration gradient from the specimen surface. That is, only slight amounts of hydrogen and martensite existed at the depth of around 100 μm. These results also indicate that martensite cracking occurs even at low hydrogen contents and martensite fractions. In addition, it is implied that the resistance to cracking at twin boundaries is higher than that in martensite, because only a small portion of twins in SUS316L were selected as the cracking site even in the near-suarface region where the number density of twin plates became double by the hydrogen uptake.
Some reasons why the twin boundary is less susceptible to hydrogen than the martensite could be found in the hydrogen desorption behavior shown in Fig. 6. Significant hydrogen desorption was observed when the martensitic transformation occurred during deformation in hydrogen-charged SUS304. In contrast, no significant increase in hydrogen pressure was observed in hydrogen-charged SUS316L. Because hydrogen charging promoted deformation twinning to double the number density, a large amount of hydrogen was present in the nucleation site or growth path. These results indicate that the motion of twinning dislocations and formation of twin boundaries do not assist hydrogen atom motion and desorption, perhaps due to the high velocity of twinning dislocations52,53,54,55) and the high coherency of Σ3 twin boundaries. The high dislocation velocity and boundary coherency would show no hydrogen transport by the twinning dislocation motion and a similar hydrogen solubility at the twin boundaries to that of the grain interior. The lack of dynamic interactions between hydrogen and the twin boundaries indicates that the effect of deformation twinning on hydrogen-related cracking is smaller than that of martensite, as long as the coherency of the twin boundaries remains during plastic deformation.
In this study, we investigated the hydrogen desorption behavior of SUS304 and SUS316L austenitic stainless steels with and without hydrogen pre-charging. The hydrogen desorption behavior correlated with the results of the quantitative characterization of the deformation microstructure evolution, that is, martensitic transformation and deformation twinning. The present study revealed the following findings:
(1) Remarkable plasticity-induced hydrogen desorption behavior was detected in the hydrogen-charged SUS304, which is associated with a deformation-induced martensitic transformation, as reported for other metastable austenitic steels. The hydrogen pressure started to increase after the initiation of plastic deformation and suddenly decreased when the deformation stopped. This result indicates that the effect of hydrogen supersaturation associated with the formation of α′-martensite is significant on hydrogen activity.
(2) Both γ→ε martensitic transformation and ε→α′ martensitic transformation were promoted by hydrogen in SUS304 austenitic stainless steel. Owing to the hydrogen concentration gradient from the specimen surface, the martensite fraction decreased monotonically with increasing depth from the specimen surface. According to a previous report, ε→α′ martensitic transformation was delayed by hydrogen charging when hydrogen charging was conducted under tensile deformation. This contradictory hydrogen effect on martensitic transformation may indicate requirements of hydrogen diffusion and desorption through the highway path of α′-martensite.
(3) SUS 316L showed no martensitic transformation but deformation twinning. The number density of the deformation twin plates was doubled by hydrogen charging, which may have caused a quasi-cleavage fracture at the corner of the specimen. However, brittleness was limited to the local position of the fracture surface, which indicates a high cracking resistance at the twin boundaries in austenitic stainless steels. A possible reason for the high cracking resistance at the twin boundaries is the negligible contribution of hydrogen activity. More specifically, the hydrogen desorption pressure did not increase significantly even after the onset of plastic deformation, although a significant amount of hydrogen, sufficient to promote twinning, was present at the twin nucleation site and/or growth path.
This work was supported by JSPS KAKENHI (JP19H00817).
