Ductility Loss of a Metastable Austenitic Stainless Steel and Its TIG Weldment Due to Hydrogen Embrittlement at Low Temperatures Considering the Effect of Pre-strain at 4 K

Abstract

The amount of martensite in austenitic stainless steels produced during plastic deformation at low temperatures is related to the reduction in hydrogen embrittlement resistance. A pre-strain at 4 K was employed in this work to produce strain-induced martensite (SIM) in the microstructure of SUS316L and its weldment to verify the changes in hydrogen embrittlement susceptibility through slow strain tensile (SSRT) tests in a high-pressure hydrogen environment. As the base metal specimens, the weld metal specimens, manufactured by gas tungsten arc welding (GTAW or TIG) were pre-strained at different levels (5%, 10%, and 15%) for comparison with the non-pre-strained condition. Analysis of the most degraded samples tested from −150°C to 0°C is conducted through fracture surface observations, lateral crack length measurement, and crack density. It was possible to indicate that the pre-strain effect induced earlier crack nucleation in comparison to the situation observed in the non-pre-strained material. The pre-existing martensite produced by the pre-strain at 4 K is responsible for earlier crack nucleation, leading to a loss in the hydrogen embrittlement resistance for the SSRT pre-strained base metal specimens.

1. Introduction

Metastable austenitic stainless steels (MASSs) can be used in numerous applications, mainly because of their formability, weldability, good corrosion resistance, good fracture toughness at low temperatures, and reasonable costs.^1,2) This type of steel has been used in cryogenic storage tanks and is an excellent candidate to be used for hydrogen storage systems. The hydrogen embrittlement (HE) resistance of MASSs and their weldments has been investigated throughout many research projects and companies.^{3,4,5,6,7,8,9,10,11,12,13,14,15,16)} When an austenitic stainless steel (ASS), with low austenite stability, is plastically deformed at low temperatures, transformation-induced plasticity (TRIP) takes place, and this phenomenon has been reported to cause impacts and changes in the mechanical properties and/or HE resistance.^17,18,19) A lower nickel content leads to a higher martensitic transformation during cold working, whereas ASS with a high nickel content has good austenite stability and does not experience TRIP even with cryogenic treatment.¹⁾ Although many studies have stated that the austenite stability may influence the HE resistance, the root cause for the existence of a temperature at which the degradation in the hydrogen environment is maximum for the ASS needs to be clarified since the increase in nickel content in ASS implies higher product costs.

For a large liquefied hydrogen storage tank that can be installed in Japan, a candidate material of this project is SUS316L, which is classified as MASSs since the deformation at low temperature leads to the formation strain-induced martensite (SIM). To ensure the safety and reliability of the structure, it is necessary to consider events and situations that can make the material susceptible to HE phenomenon. During operation, the inner tank material of the liquefied hydrogen storage tank is at −253°C. However, at this temperature, hydrogen embrittlement does not occur. By using the full tank, emptying the tank, or during the operational conditions due to boil-off, the inner tank material can experience increases in temperature, and the material performance at the temperature range of −150~0°C must be assessed regarding HE risks. Therefore, evaluation tests at this temperature range with the virgin material and plastically deformed were assessed to represent the effect of earthquakes and temperature changes in the inner material.

This current work investigates the material performance and HE mechanisms employing SSRT tests at various low temperatures with smooth specimens, made by austenitic stainless steel (SUS316L) and its weldment manufactured by the TIG process. The test temperatures were in the range between −150°C and 0°C, encompassing the region in which maximum hydrogen embrittlement occurs. Moreover, investigations of phase transformation showed that as the test temperature for MASSs decreases, the TRIP becomes more evident, and the martensite content increases more easily.²⁰⁾ Thus, pre-strain 4 K in a helium environment was considered to generate the highest amount of martensite for base metal and weld metal specimens. As their microstructures changed due to three different amounts of pre-strain 5%, 10%, 15%. Electron Back Scattered Diffraction (EBSD) and X-ray diffraction (XRD) analyses were conducted to verify it. Furthermore, the density of cracks induced by hydrogen is also investigated to verify the effect of pre-strain on the crack nucleation, as well as the relationship between an estimated crack nucleation with the ductility losses.

2. Methodology and Materials

Two types of specimens, one extracted from the base plate, and another extracted from the welded joint, were tested in the pure 10 MPa hydrogen environment and a reference environment, using slow strain rate tensile (SSRT) tests. Whereas the test in the hydrogen environment used a pressure chamber equipped with an auxiliary chiller to control the test temperature, the reference environment was composed of air and nitrogen spray, which was used to cool down the specimens at the desired temperature.

The SUS316L, Table 1, was used for both specimens, weld metal (WM) and base metal (BM), to manufacture smooth round specimens, as shown in Fig. 1. The TIG welding process was employed in a butt joint with a double V groove, using the commercial welding rod TG-S316L. A vertical position welding process was carried out to simulate actual flat bottom cylindrical tank fabrication. Additionally, the effect of plastic strain was considered by straining the specimen in helium at 4 K in three different levels (5%, 10%, and 15%), subsequently, the non-pre-strained and pre-strained specimens were tested under the same conditions.

Table 1. Chemical composition of the SUS316L steel plate and welding rod.

Chemical Composition of the SUS316L steel plate or base metal (mass%)
C	Si	Mn	P	S	Ni	Cr	Mo	Co
0.017	0.64	0.83	0.031	0.001	12.07	17.51	2.08	0.43

Chemical Composition of the wire TG-S316L (mass%)
C	Si	Mn	P	S	Ni	Cr	Mo	Cu
0.007	0.41	1.54	0.027	0.001	12.11	18.52	2.13	0.15
Nb	N	FN (Delong)		FN (Schaeffler)			FN (WRC)
0.01	0.01	10		7			9

Fig. 1. a) Specimen dimensions; b) joint configurations; c) cross-section from the welded joint, indicating the melting line; d) position from extracted part from the welded joint to manufacture the WM specimens. (Online version in color.)

In this current work, SSRT tests were conducted in two environments at five different temperatures, −150, −100, −75, −40, and 0°C. The slow strain rate tensile (SSRT) tests were executed following ASTM G129-21,²¹⁾ with machine displacement control to strain the specimen in the elastic region at 2.8×10⁻⁵/s.

Figure 2 illustrates the main methodology adopted in this work. After the SSRT tests, the lateral surface of the fractured specimens was observed using a digital microscope (VHX-8000; KEYENCE, Japan), and the hydrogen-induced cracks on the external surface were quantified. Although cracks can nucleate on the fillet radius surface of the specimen, the crack number and density measurements were carried out only for the reduced section to appropriately quantify the relationship between crack nucleation and plastic strain amount.

Fig. 2. Methodology employed to analyze the hydrogen embrittlement and pre-strain effect through SSRT tests. (Online version in color.)

Also, the fracture surfaces were carefully inspected with the aid of a JSM-IT300LV scanning electron microscope (JSM-IT300LV; JEOL, Japan). Pre-strain at 4 K caused changes in the microstructure that were observed using the EBSD method (Electron Back Scattered Diffraction) by a field emission scanning electron microscope (SEM; acceleration voltage: 20 kV; JSM-7000F, JEOL, Japan) and the XRD method (tube voltage: 40 kV; tube current: 15 mA; SmartLab, Rigaku, Japan) with Cu radiation, as well as using a magnetic measurement method using ferrite meter (FM; FMP30C, Fischer, Germany). Samples were extracted from the cross-section at the center part of each specimen’s gauge length. The non-pre-strained BM and WM are also considered in this analysis as the standard microstructures, and they are the reference to indicate the changes; for example, martensite transformed by plasticity.

Regarding the relative reduction of area (RRA) calculations and area fraction measurements from the fracture surface, a procedure was adopted to obtain more reliable values. The reduction of area is commonly calculated using the measurement of the diameter by a caliper. Nevertheless, the measurements in this current study used the fracture surface images obtained by the microscope and SEM machine. RRA was calculated according to Eq. (1), which represents a common HE index to quantify degradation and indicates the temperature of maximum hydrogen embrittlement (TMHE).

  

$$\begin{array}{c}RRA[\%]=\frac{R{A}_{H2}}{R{A}_{re}}\times \text{100}\end{array}$$(1)

RA_re stands for reduction of area in the reference environment, while RA_H2 is the paramter in the hydrogen environment.

3. Microstructure Changes Due to Pre-strain

Microstructural changes are expected to occur during the pre-strain procedure at 4 K due to deformation-induced martensitic transformation of austenite. Different microstructures and phases have different susceptibilities to hydrogen embrittlement. Therefore, the condition of the microstructure after applying four different amounts of plastic strain (0%, 5%, 10%, and 15%) is evaluated by SEM-EBSD (Scanning Electron Microscope-Electron Back Scattered Diffraction) method, XRD (X-ray diffraction) method, and magnetic measurement method. The observed surfaces were first wet-polished with emery polishing papers, followed by buff polishing with diamond slurry. The strained layer was then removed by mechanical polishing via colloidal silica polishing for 30 min.

Figure 3 shows the crystallographic orientation maps and phase maps of the BM and WM obtained via SEM-EBSD experiments. In the phase maps, the green, red, and yellow regions represent face-centered cubic (FCC), body-centered cubic (BCC), and hexagonal close-packed (HCP) iron, respectively.

Fig. 3. Crystallographic orientation maps: (a) 0% pre-strained BM, (b) 5% pre-strained BM, (c) 10% pre-strained BM, and (d) 15% pre-strained BM, (e) 0% pre-strained WM, (f) 5% pre-strained WM, (g) 10% pre-strained WM, and (h) 15% pre-strained WM; and phase maps of: (i) 0% pre-strained BM, (j) 5% pre-strained BM, (k) 10% pre-strained BM, and (l) 15% pre-strained BM, (m) 0% pre-strained WM, (n) 5% pre-strained WM, (o) 10% pre-strained WM, and (p) 15% pre-strained WM. (Online version in color.)

The prior austenite grains of the non-pre-strained specimens in the WM are coarser than the BM grains. This suggests that austenite in the WM was formed at a very high temperature after solidification. Additionally, in the BM, grains for the 15% pre-strained specimens were visually smaller than those on the other pre-strained and non-pre-strained specimens.

Although the microstructural changes under pre-strain amounts up to 5% do not seem expressive for both BM and WM, the increase in α’ martensite content was apparent from the 10% pre-strain level, and also ε martensite (HCP phase) was observed. Figure 3 shows that the martensitic transformation progressed further at 15% pre-strain for the BM and at 10% pre-strain for the WM.

From the EBSD analysis, the fraction of martensite seemed to have a sharper increase for the WM than for the BM. Nevertheless, EBSD sampling does not account for a large proportion of the specimens’ cross-section, and verification using a ferrite meter and XRD can be more adequate to compare the fraction of induced martensite.

The non-pre-strained BM specimen is almost a single FCC phase. However, as the degree of pre-strain increased, the amount of BCC and HCP phases increased. The fraction of BCC phase (f_F) was determined using Eq. (2):

  

$$\begin{array}{c}{f}_F\text{[\%]}=\frac{\frac{1}{n}{{\displaystyle \sum }}_1^n\frac{I_{hkl}^F}{R_{hkl}^F}}{\left(\frac{1}{m}{{\displaystyle \sum }}_1^m\frac{I_{hkl}^A}{R_{hkl}^A}+\frac{1}{n}{{\displaystyle \sum }}_1^n\frac{I_{hkl}^F}{R_{hkl}^F}+\frac{1}{o}{{\displaystyle \sum }}_1^o\frac{I_{hkl}^E}{R_{hkl}^E}\right)}\text{×100}\end{array}$$(2)

where m, n, and o are the numbers of hkl diffraction peaks for FCC, BCC, and HCP used in the calculation, respectively. Similarly, the superscript A, F, and E refer to FCC, BCC, and HCP, respectively, and the theoretical diffraction intensity is denoted by R_hkl, while the measured diffraction intensity is represented by I_hkl.

Figure 4(a) shows an estimative of the volume fraction of the BCC phase obtained, using XRD and FM. There are three curves for the BM in Fig. 4(a), two of them were obtained from XRD and indicate that the fractions of α’ martensite and ε martensite are similar and follow the same trend. The measurements of the BCC phase through the ferrite meter in the BM were compatible with the α’ martensite calculated by XRD, showing agreement. In the case of WM, the XRD measurements were not successful, and it may be due to the coarser grains. Then, the volume fraction for WM is represented only by the FM results, in which the non-pre-strained specimen with approximately 10% delta ferrite is considered the reference value. Figure 4(b) considers the reference value measured in the non-pre-strained material to calculate the change in volume fraction due to α’ martensite. Minor uncertainties in the FM results of the welded region are expected to be produced by the presence of delta ferrite and its content variation. Furthermore, the delta ferrite content of WM predicted from the Scheffler diagram is 5 to 10%, and it is assumed that delta ferrite crystallized during solidification is dispersed in the WM.

Fig. 4. a) Phase fraction of pre-strained specimens; b) change in volume fraction of α’ martensite due to pre-strain. (Online version in color.)

4. Results

4.1. Ductility Comparison from SSRT Tests

Figure 5(a) shows the relative reduction of area (RRA) of the BM from −150°C to 0°C for the non-pre-strained and pre-strained materials. The lowest values of RRA, 61.5%, 64.2%, 49.5%, and 41.8% for the non-pre-strained, 5%, 10%, and 15% pre-strained material, respectively, were found at −75°C. This is approximately the temperature of maximum HE obtained in the test temperature range. Whereas the performance for the 5% pre-strained material at this range seemed slightly better in comparison to the non-pre-strained material, the RRA values for the 10% and 15% pre-strained specimens decreased as the pre-strain amount increased. Thus, the higher level of plastic strain at 4 K seemed to be deleterious for HE resistance. Moreover, the 15% pre-strained BM showed a lower RRA value at −40°C in comparison to the other specimens tested at the same temperature. From the RRA results for the BM and according to ANSI (2014),²²⁾ the hydrogen embrittlement effect at −150°C, −40°C (except the 15% pre-strained material), and 0°C may be not significant since RRA is higher than 90%. However, these specimens were inspected to verify any visible damage caused by hydrogen. In general, the external surface of all specimens tested in the hydrogen environment was observed, as well as their fracture surface in order to classify the HE mechanism for each case.

Fig. 5. Relative reduction of area (RRA) versus temperature for the non-pre-strained and pre-strained materials: a) BM; b) WM. (Online version in color.)

Similar to BM specimens, no significant embrittlement is expected to happen at −150°C for all WM specimens. Nevertheless, the general WM results varied as the test temperature and pre-strain amount increased. Comparing Figs. 5(a) with 5(b), it is possible to notice that changes mainly happened for the pre-strained weld metal tested from −100°C to −40°C. From −150°C to −75°C, the trend of decreasing RRA values as temperature increased to −75°C was observed for the WM, except for the 15% pre-strained specimen that showed a peak value at −100°C. Secondly, from −40°C to 0°C, RRA remained low for the pre-strained WM and then recovered as the test temperature increased to 0°C, while the non-pre-strained WM showed a full RRA recovery at −40°C.

Furthermore, the non-pre-strained, 5%, and 10% pre-strained WM showed similar degradation in ductility from −150°C to −75°C due to hydrogen embrittlement. The test results at −75°C indicated the agreement in the mechanical behavior between the WM and BM at TMHE. However, the performance of 15% pre-strained WM differed from the non-pre-strained, 5%, and 10% pre-strained at −100°C, and its RRA was highly impacted at −75°C.

The unexpected RRA value at −100°C for the 15% pre-strained WM can be explained by a higher reduction of area value in the hydrogen environment. While the WM specimens tested in the reference environment showed similar trends in reduction of area, the tendency for specimens tested in the hydrogen environment was deviated by some reason for the 15% pre-strained WM tested at −100°C, and RA_H2 was relatively high.

As the test temperature increased to −40°C and then 0°C, the non-pre-strained WM instantly recovered from the hydrogen embrittlement region, and 5% pre-strained WM showed also a slight increase in RRA value. Nevertheless, the results for 10% and 15% pre-strained specimens can imply that embrittlement caused by hydrogen may be extended from −75°C to −40°C in this WM. Finally, the mechanical performance in hydrogen was recovered for the pre-strained WM when tested at 0°C, and all of the RRA values were higher than 90%. It is necessary to investigate if the pre-strain at 4 K induced changes in the WM microstructure that could be harmful to HE resistance.

4.2. Fracture Surface Classification for the Specimens Tested in the Hydrogen Environment

The fracture surface appearance changed according to the temperature and sometimes as the pre-strain amount was added, as shown in Fig. 6. In order to qualify each specimen’s condition under the HE effect, Table 2 was created, summarizing all the fracture surface observations and relating their appearance with RRA value. Six levels of hydrogen embrittlement severity were used to describe the condition of the fracture surface, according to the appearance of quasi-cleavage regions that were induced by the hydrogen effect. The labels that qualified the fracture surface ranged from 0, indicating the lowest effect of hydrogen embrittlement or its absence, to 5, representing the most severe case. A fully ductile fracture is represented by 0, while a fully brittle fracture is represented by 5.

Fig. 6. Fracture surface observation: a) 10% pre-strained base metal tested at −75°C; b) marked region from the 10% pre-strained base metal; c) 15% pre-strained weld metal specimen tested at −75°C; d) marked region from the 15% pre-strained weld metal; e) the growth of cracks induced by HE and the expansion of quasi-cleavage patterns from the lateral surface as test temperature is increased from −150°C to −75°C. (Online version in color.)

Table 2. Classification of the fracture surface appearance considering six levels of hydrogen embrittlement. (Online version in color.)

In a general overview of the fracture surface observation, specimens tested at −150°C did not present quasi-cleavage patterns, and their fracture surfaces were covered by dimples, and similar situation was observed for the specimens tested at 0°C, which showed only very small areas with quasi-cleavage patterns, indicating low degradation. The appearance of the fracture surface was in agreement with the RRA values.

Some specimens’ fracture surfaces showed quasi-cleavage patterns, as well as quasi-cleavage crack formation, but with RRA above 90%. It is believed that hydrogen can induce cracks on the lateral surface, which are viewed as quasi-cleavage patterns on the fracture surface; however, its effect is not strong enough to affect the ductility of the material.

4.3. External Lateral Surface Inspections

Cracks were induced on the lateral surface of the specimens in the hydrogen environment, and they were visible for the BM and WM tested at −100°C and −75°C, as well as WM at −40°C. Specimens tested in the reference environment did not present cracks, this section analyzes the results of the external lateral surface inspections for the BM specimens tested in the hydrogen environment. Apparently, the number of lateral cracks and their size varied according to test temperature and the pre-strain amount. Figure 7 shows examples of the image used for the lateral surface inspection through the microscope VHX-8000. These cracks were measured and counted, and then they were related to the corresponding diameter of the cross-section (D_i) to calculate the plastic strain in the specimen’s axis direction. Considering a constant volume for the SSRT specimen, the plastic strain in the axial direction of the specimen was calculated by using Eq. (3):

  

$${\epsilon }_{yy}=\frac{D_0^2}{D_1^2}-\text{1}$$(3)

where D is the diameter of the tensile specimen, in which D₀ refers to the initial diameter, and D₁ refers to the diameter of the specified region after deformation. Figure 7(b) exemplifies a region with a constant diameter inside the red rectangle, where cracks were counted and measured to relate with the corresponding ε_yy. The counting and measurement procedure used high-resolution 2D images, Fig. 7(b), in which 2D measurements of cracks located near the specimens’ axial line could provide less uncertainty for the results. In order to encompass the whole outer surface, four observations were done, being the first on 0°, secondly rotating 90° the specimen, after rotating more 90° to set the position at 180°, and finally rotating once more to 270°. Since there was a non-uniform distribution of strain along the specimen’s axial line of WM specimens, the results would not be trustworthy, and then this topic discusses only measurement with the BM specimens only.

Fig. 7. Lateral surface scanned in four parts of 90° for the non-pre-strained BM tested at −100°C. The sketches in red account for the calculation of the corresponding area in Eq. (4) and D_i is equal to D₁ to calculate strain in Eq. (3). The green square is the delimited region closer to the longitudinal axis in which cracks were measured to minimize errors due to the cylindrical surface curvature. (Online version in color.)

A possible reasoning that could be produced by some readers is that the number of cracks on the lateral surface may have a relation with the material’s performance, as maintenance investigations of the stress corrosion cracking (SCC) phenomenon frequently verify the number of cracks.²³⁾ Table 3 indicates that the hydrogen-induced 304 cracks on the non-pre-strained material tested at −75°C, while 1122 cracks were induced by hydrogen in the same material tested at −100°C. This may be due to the higher martensitic transformation during plastic deformation as the temperature decreases, which is a typical behavior of MASSs.²⁴⁾ The number of cracks and maximum crack length for the non-pre-strained specimens tested at −75°C and −100°C suggested that the material’s mechanical performance is not degraded by the increase in the number of cracks that nucleated near the lateral external surface but it is probably affected by its extension on the cross-section. Moreover, the decrease in RRA at −75°C according to the pre-strain amount at 4 K may also have no strict relation with the number of cracks. Nevertheless, RRA seems to be relatable with the lowest plastic strain in the specimen’s axis direction, in which a crack was measured and counted for the BM specimens tested at −75°C. RRA, the lowest ε_yy with crack, and the maximum crack length in the lowest ε_yy region in Table 3 suggest that cracks probably nucleate in an earlier stage or grow faster as the pre-strain amount at 4 K is increased. Martensite nucleated during the tensile tests is one of the paths for hydrogen diffusion into the specimen the specimen microstructure, and the reduction of RRA values and the earlier crack nucleation as pre-strain amount increases are relatable with the pre-existing martensite content. Because of the non-uniform deformation of the WM specimens, this relationship is more complicated to establish.

Table 3. Brief information about the lateral surface inspection for the BM specimens.

Variable	Test Temperature
	−75°C				−100°C
Pre-strain amount	0%	5%	10%	15%	0%
Crack Amount	304	132	541	437	1122*
amax	1.16	0.84	1.17	1.15	0.58
amax in the region of ${\epsilon }_{y{y}_{\text{min}}}$	0.26	0.29	0.72	0.32	0.19
${\epsilon }_{y{y}_{\text{min}}}$	0.4	0.4	0.19	0.03	0.36
RRA	61%	63%	49%	41%	85%

Notes: ε_yy is the plastic strain in the specimen’s axis direction along the specimen’s center axis. ${\epsilon }_{y{y}_{\text{min}}}$ is the lowest plastic strain in the specimen’s axis direction with crack nucleated on the surface. a^max is the maximum crack length in mm

*  This crack amount is only for 270° inspection in Fig. 12(d), and thus the total number of cracks can be approximately four times.

Another important aspect of this lateral inspection of hydrogen-induced cracks is their distribution according to each level of plastic strain along the specimen’s gauge length. As the final specimen’s length after fracture varies for each case, the number of cracks and their lengths on the surface need to be computed with the corresponding area under the plastic strain. The area along the gauge length of the specimen is the corresponding area and calculated by Eq. (4), considering the external lateral surface under the same plastic strain:

  

$$Are{a}_{corresponding}=\pi D_i{L}_i$$(4)

where the corresponding diameter after fracture is D_i and the length for that region is L_i.

5. Discussions

5.1. Crack Density and Distribution

It is well-known that hydrogen concentration inside the strained tensile specimen in the hydrogen environment is not uniform. Since the test is conducted in a high-pressure hydrogen environment under the SSRT test conditions, a higher hydrogen concentration in the microstructure is expected to exist near the external lateral surface. It is not easy to discuss the detailed and quantitative distribution of hydrogen at the surface of tensile specimen and crack tip. However, the authors will try to quantify them by numerical analysis in future work. This higher concentration of absorbed hydrogen near the surface in the influence of the plastic strain leads to the formation of lateral cracks, which extend to form a quasi-cleavage zone on the fracture surface. Figures 8 and 9 show the distribution of these cracks per the corresponding area as a tensile specimen at −75°C plastically deforms until its failure, as well as the crack density represented by the amount of crack per inspected length, L, in Fig. 9(b). To elaborate Figs. 8 and 9, all the counted and measured cracks were grouped according to the strain level calculated as Eq. (3). The crack length and number of cracks were divided by the corresponding area calculated as Eq. (4) to rationalize them within the area that has the same strain level.

Fig. 8. Individual crack length per area versus plastic strain in the specimen’s axis direction for BM tested in the hydrogen environment at −75°C: a) 0% pre-strain and 5% pre-strain; b) 10% pre-strain and 15% pre-strain. (Online version in color.)

Fig. 9. a) Total crack length per area versus plastic strain in the specimen’s axis direction for BM tested in the hydrogen environment at −75°C, including estimations for the strain in which crack nucleates for the 0% and 5% pre-strained specimens (${\epsilon }_{yy}^{0\%}$, and ${\epsilon }_{yy}^{5\%}$) by drawing trendlines. (The red part of each curve indicates the necking zone); b) Crack density versus plastic strain in the specimen’s axis direction for BM tested in the hydrogen environment at −75°C. (The red portion of each curve indicates the necking zone). (Online version in color.)

Figure 8 shows that a general tendency for crack length is to increase as the plastic strain in the specimen’s axis direction increases for specimens tested at −75°C. Although this tendency is weaker for the regions in the necking zone that is close to the fracture point, the majority of results portray the crack growth as plastic strain in the specimen’s axis direction increases in the hydrogen environment for a smooth SSRT specimen. In addition, it is possible to indicate that pre-strain at 4 K may influence the crack nucleation at TMHE. In Figs. 8 and 9, it is possible to verify that cracks probably initiate earlier for specimens with high levels of pre-strain (10% and 15%). Furthermore, the plastic strain range in the presence of cracks was reduced, as shown in the coordinate of Fig. 9. Another perspective, that can be given about hydrogen-induced cracks by observing Figs. 9(a) and 9(b), is that cracks start in small size, grow, and then connect with adjacent cracks making the degradation even more severe, as they also grow in the thickness direction. The red part of each curve in Fig. 9 indicates the necking zone, which is very close to the fracture point and did not fully follow the relationship. It may be due to a stronger influence of the crack expansion in the thickness direction than along the lateral external surface. Note that 10% and 15% pre-strained base metal also showed a higher content of pre-existing martensite, which may be the cause of the earlier crack nucleation at −75°C.

By drawing a fitting curve for 0% and 5% pre-strained base metal, the crack nucleation point can be estimated when the total crack length per area reaches the plastic strain axis. In Fig. 9, it is estimated the values for the plastic strain in the specimen’s axis direction that corresponds to the crack nucleation point. If a fitting curve is used for the 10% or 15% pre-strained BM, the crack nucleation point would be at a plastic strain in the specimen’s axis direction, ${\epsilon }_{yy}^{10\%\mathrm{\hspace{0.17em} }\text{or}\mathrm{\hspace{0.17em} }15\%}$, lower than 0. Then, it is assumed that the lateral crack nucleates at a null plastic strain in the direction of the specimen’s axis or as soon as the yielding happens.

5.2. Effect of the Pre-Existing Martensite

The pre-strain amount at 4 K induced martensitic transformation, including α’ martensite and ε martensite. Figure 10 relates the contents of martensite with the RRA values at the most critical situations, that is at −75°C for the BM, and −40°C and −75°C for the WM. In the secondary vertical axis of Fig. 10, the area fractions of brittle zone are also plotted in function of martensite volume fractions to depict the extension of quasi-cleavage patterns on the fracture surface. The areas of the brittle zone for each specimen were measured during the classification of the fracture surface, described in section 4.2. The area fraction was computed by dividing this area of the brittle zone by the cross-section of the specimens before the SSRT tests. In the BM, Fig. 10(a), the volume fraction of martensite is closely related to the reduction of RRA and extension of quasi-cleavage pattern on the fracture surface.

Fig. 10. Pre-existing martensite influence on the RRA values and the brittle area fraction: a) BM at −75°C; b) WM at −75°C and −40°C. (Online version in color.)

Whereas the WM offered coarse grains that make XRD measurements difficult, the BM specimens presented a microstructure, in which the α’ martensite and ε martensite contents were measured by XRD. Nevertheless, martensite measurements for the WM were done with the ferrite meter, and the variation of α’ martensite was the only data for this material. BM showed a compatible decrease in the RRA according to the increase of the pre-existing α’ and ε martensite contents, in Fig. 10(a). As the martensite has a higher hydrogen diffusivity and lower solubility, hydrogen transport, and saturation are easier in this microstructure. Many studies have investigated the relationship between a decrease in hydrogen embrittlement resistance and the austenite instability.^17,19,25,26) In MASSs, this instability facilitates martensitic transformation at low temperatures when plastic deformation is applied. In this current work, it is also possible to guess that the pre-existing martensite may be affecting the mechanical performance of the SUS316L, increasing the fraction of brittle fracture on the fracture surface, and probably leading to earlier crack nucleation and propagation. On the other hand, the WM manufactured by the TIG welding process has the presence of delta ferrite, and four microstructures (austenite, delta ferrite, and α’ and ε martensite phases) are influencing its mechanical performance in the hydrogen environment. Therefore, the effect on the hydrogen embrittlement resistance, as the pre-strain amount increases, is somewhat complicated to be explained only by the increase in the α’ martensite content, and then the WM performances do not follow an expected trend in Fig. 10(b).

From the BM behavior, the RRA values can be related to the increase of martensite content through Eqs. (5) and (6), considering that only the measurement of α’ or ε martensite volume fraction is enough to determine the negative effect on the decrease in ductility of MASSs in the hydrogen environment.

  

$$RR{A}_{BM}^{-75°\text{C}}=0.6272e^{-4.182({\varnothing }_{\alpha \prime })}$$(5)

  

$$\begin{array}{c}RR{A}_{BM}^{-\text{75}°\text{C}}=0.6244e^{-4.638({\varnothing }_{\epsilon })}\end{array}$$(6)

where ∅_α′ and ∅_ε represent the volume fractions of the pre-existing α’ martensite and ε martensite, and $RR{A}_{BM}^{-75°\text{C}}$ represents the relative reduction of area of the BM at the temperature of maximum hydrogen embrittlement (TMHE).

6. Conclusions

This work demonstrated the effect of pre-strain at 4 K in the HE resistance of SUS316L and its weld joint. SSRT tests with smooth round bar specimens in the hydrogen environment and in a reference environment were performed with the BM and WM specimens at low temperatures. Cracks that nucleated on the surface of the BM specimens tested at −75°C were quantified and measured, and changes in cack nucleation due to the pre-strain effect can be understood in this current work. The results following main conclusions can be drawn from the results of this this research:

• The effect of pre-strain is determined by the crack distribution on the lateral external surface of the BM. Cracks were likely initiated early as the pre-strained amount increased to 10% and 15%. In addition, although the distribution of the crack density and length of cracks according to the plastic strain during the SSRT tests were not uniformly distributed, it was possible to note a tendency of increasing crack density and length as the plastic strain in the longitudinal axis direction increased.

• For the BM, RRA values were reduced due to the pre-strain at 4 K, especially for the specimens tested at −75°C. The pre-existing martensite is related to the loss in material performance owing to an increase in deterioration by HE.

• For the WM, the pre-strain effect caused multiple changes. The WM microstructure changed from a dual-phase (austenite and delta ferrite) to a quadruple-phase (austenite, delta ferrite, α’ and ε martensites) due to the pre-strain effect at 4 K. The multiple microstructures in the pre-strained WM, as well as the variations in the delta ferrite shape and distribution, hindered the relationship between the pre-existing martensite and the degradation of ductility in the hydrogen environment.

Acknowledgements

This article was based on the results obtained from a project commissioned by the New Energy and Industrial Technology Development Organization (NEDO).

References

• 1)   M. F.  McGuire: Stainless Steels for Design Engineers, ASM International, Ohio, (2008).
• 2)   Y.  Qiu,  H.  Yang,  L.  Tong and  L.  Wang: Metals, 11 (2021), 1101. https://doi.org/10.3390/met11071101
• 3)   T.  Horiya,  N.  Yamamoto,  T.  Iida,  A.  Yamamoto,  H.  Okaguchi,  N.  Yaegashi,  T.  Doko,  M.  Saito,  K.  Yokogawa and  T.  Ogata: Evaluation Test of Structural Materials Used for Liquid Hydrogen Storage and Transportation System (Study on Low Temperature Materials used in WE-NET 1), https://www.enaa.or.jp/WE-NET/ronbun/1996/e17/jr1996.html, (accessed 2024-06-25).
• 4)   E.  Herms,  J.  Olive and  M.  Puiggali: Mater. Sci. Eng. A, 272 (1999), 279. https://doi.org/10.1016/S0921-5093(99)00319-6
• 5)   M. I.  Luppo,  A.  Hazarabedian and  J.  Ovejero-García: Corros. Sci., 41 (1999), 87. https://doi.org/10.1016/S0010-938X(98)00083-3
• 6)   S.  Fukuyama,  D.  Sun,  L.  Zhang,  M.  Wen and  K.  Yokogawa: J. Jpn. Inst. Met., 67 (2003), 456 (in Japanese). https://doi.org/10.2320/jinstmet1952.67.9_456
• 7)   C. S.  Marchi: Gaseous Hydrog. Embrittlement Mater. Energy Tech., 2 (2012), 592. https://doi.org/10.1533/9780857093899.3.592
• 8)   C. S.  Marchi,  B. P.  Somerday,  X.  Tang and  G. H.  Schiroky: Int. J. Hydrog. Energy, 33 (2008), 889. https://doi.org/10.1016/j.ijhydene.2007.10.046
• 9)   T.  Michler and  J.  Naumann: Int. J. Hydrog. Energy, 33 (2008), 2111. https://doi.org/10.1016/j.ijhydene.2008.02.021
• 10)   T.  Michler,  Y.  Lee,  R. P.  Gangloff and  J.  Naumann: Int. J. Hydrog. Energy, 34 (2009), 3201. https://doi.org/10.1016/j.ijhydene.2009.02.015
• 11)   T.  Michler,  J.  Naumann,  M.  Hock,  K.  Berreth,  M. P.  Balogh and  E.  Sattler: Mater. Sci. Eng. A, 628 (2015), 252. https://doi.org/10.1016/j.msea.2015.01.054
• 12)   J.  Park,  T. T.  Nguyen,  H. M.  Heo and  U. B.  Baek: Mater. Charact., 197 (2023), 112666. https://doi.org/10.1016/j.matchar.2023.112666
• 13)   P.  Tao,  F.  Ye,  W.  Cen,  J.  Zhao,  Y.  Wang and  J.  Gong: Results Phys, 16 (2020), 102820. https://doi.org/10.1016/j.rinp.2019.102820
• 14)   C.  Zhou,  P.  Da,  H.  Wu,  M.  He,  X.  Liu and  P. K.  Chu: Appl. Surf. Sci., 606 (2022), 154866. https://doi.org/10.1016/j.apsusc.2022.154866
• 15)   H. S.  Yun,  J. S.  Koo,  Y. K.  Lee,  B. U.  Bong and  S. H.  Nahm: Int. J. Hydrog. Energy, 48 (2023), 19297. https://doi.org/10.1016/j.ijhydene.2023.01.025
• 16)   R. M. M.  Freire,  M.  Kimura and  T.  Kawabata: International Ocean and Polar Engineering Conference, (2023), ISOPE-I-23-495. https://onepetro.org/ISOPEIOPEC/proceedings/ISOPE23/All-ISOPE23/ISOPE-I-23-495/524370
• 17)   C.  Izawa,  S.  Wagner,  M.  Deutges,  M.  Martín,  S.  Weber,  R.  Pargeter,  T.  Michler,  H.  Uchida,  R.  Gemma and  A.  Pundt: Int Int. J. Hydrog. Energy, 44 (2019), 25064. https://doi.org/10.1016/j.ijhydene.2019.07.179
• 18)   C. L.  Lai,  L. W.  Tsay and  C.  Chen: Mater. Sci. Eng. A, 584 (2013), 14. https://doi.org/10.1016/j.msea.2013.07.004
• 19)   H.  Zhang,  L.  Zheng,  T.  Wang,  W.  Lv,  Q.  Shi,  J.  Ma,  Y.  Luo,  W.  Liang,  J.  Hu and  R. D. K.  Misra: Mater. Sci. Eng. A, 831 (2022), 142169. https://doi.org/10.1016/j.msea.2021.142169
• 20)   T.  Ogata,  T.  Yuri and  Y.  Ono: Teion Kogaku, 42 (2007), 10. https://doi.org/10.2221/jcsj.42.10
• 21)  ASTM G129-21: Standard practice for slow strain rate testing to evaluate the susceptibility of metallic materials to environmentally assisted cracking.
• 22)  ANSI/CSA CHMC 1-2014: Test methods for evaluating material compatibility in compressed hydrogen applications-metals.
• 23)   S.  Suzuki,  K.  Takamori,  K.  Kumagai,  A.  Sakashita,  N.  Yamashita,  C.  Shitara and  Y.  Okamura: Stress Corrosion Cracking in Low Carbon Stainless Steel Components in BWRs, E-Journal of Advanced Maintenance, 1 (2009), 1. https://jsm.or.jp/ejam/Vol.1.No.1/AP/EJAMVol.1(2009)1-29_S_SUZUKI_et_al.pdf, (accessed 2024-08-06).
• 24)   K.  Nalepka,  B.  Skoczeń,  M.  Ciepielowska,  R.  Schmidt,  J.  Tabin,  E.  Schmidt,  W.  Zwolińska-Faryj and  R.  Chulist: Materials, 14 (2020), 1. https://doi.org/10.3390/ma14010127.
• 25)   M.  Moallemi,  S. H.  Kim and  S. J.  Kim: Mater. Sci. Eng. A, 885 (2023), 145595. https://doi.org/10.1016/j.msea.2023.145595
• 26)   K.  Ichii,  M.  Koyama,  C. C.  Tasan and  K.  Tsuzaki: Scr. Mater., 150 (2018), 74. https://doi.org/10.1016/j.scriptamat.2018.03.003