2022 Volume 62 Issue 7 Pages 1540-1547
To investigate the effect of high-pressure hydrogen gas on the fracture of high-strength low-alloy steels, rising load tests were conducted on JIS SNCM439 steel in high-pressure (20 MPa) hydrogen gas at room temperature (20–25°C), and its hydrogen-induced crack initiation behavior was investigated. The load-crack opening displacement curve obtained for rising load tests in hydrogen began to deviate from that obtained in air at very low loads, indicating that the stress intensity factor at crack initiation was significantly smaller in hydrogen. Scanning electron microscopy observations of the fractured surfaces of the specimens unloaded during the middle of the rising load test confirmed that hydrogen-induced cracks had already occurred at a load lower than the deviation point. The stretch zone that appeared in the rising load test in air was not observed for the test in hydrogen, and the hydrogen-induced cracks were found to directly initiate from the tip of the fatigue pre-crack. The hydrogen-induced cracks were initiated at almost the same stress intensity factor value as that at which the stretch zone was observed in air, indicating that plastic slip and the resulting hydrogen ingress from the new surface were the causes of the hydrogen embrittlement. In addition, it was shown that the stress intensity factor for crack initiation in hydrogen increased and the effect of hydrogen decreased with the increase in loading rate, inferring that dislocation migration and hydrogen penetration into the steel are key factors for hydrogen embrittlement.
Steel hydrogen pressure vessels used at refueling hydrogen stations are composed of materials such as high-strength low-alloy steels that offer good material strength, enable facile processing, and are economical, among other advantages.1) As hydrogen pressure vessels are receptacles designed to store large amounts of hydrogen (a combustible gas) in urban settings, it is necessary to ensure that there is no hydrogen leakage; this can be done by elucidating the effects of hydrogen on the materials used in these pressure vessels. These steels are known to be affected by high-pressure hydrogen gas and exhibit different mechanical properties under hydrogen and air atmospheres.2,3,4,5) One property affected by hydrogen is the fatigue crack growth rate of these steels, which is reported to be several dozen times greater in hydrogen gas than in air; furthermore, above certain values, fatigue cracks in these steels have been reported to spread in a time-dependent manner.6) Therefore, the presence of large cracks on the internal surfaces of pressure vessels in contact with hydrogen gas may abruptly progress to the point of failure through the repeated amplification of pressure, with the crack as the origin. One method of managing safety regarding hydrogen pressure vessels is to inspect for cracks on the inner surface of the vessels; it is extremely important to confirm whether the discovered cracks are potentially large enough to propagate.
Fracture-mechanics evaluation methods for crack initiation and propagation are well established, and numerous studies have been reported on hydrogen-induced crack initiation and propagation behavior.7,8,9) Regarding the effect of hydrogen on fracture, Kayakabe et al. conducted fracture toughness tests using hydrogen-charged materials and pointed out that subcritical flow growth occurs before rapid fracture.10) There are also several reports on the crack initiation behavior of carbon steel and low-alloy steel in high-pressure hydrogen gas. Lam et al. investigated various mechanical properties, such as the critical stress intensity factor (KIC), of such steels in hydrogen gas and reported that factors such as gas pressure and purity affect them.11) Cialone et al. conducted rising load tests in hydrogen gas, H2S gas, and seawater to obtain the lower bound stress intensity factors of steels in various environments; they revealed that these values are greatly affected by the gas pressure and loading rate, and that the test conditions of such quantitative evaluations particularly affected these values of materials with low yield stress.12) Nibur et al. evaluated the threshold of crack initiation in hydrogen gas using the constant displacement method and the rising load method and found that the latter was more conservative due to differences in the strain history relative to the environmental exposure history.13) Matsuoka et al. reported that the mechanism of crack propagation in hydrogen is ductile where the slip is localized in a narrow region, the plastic deformation hardly appeared under small-scale-yielding conditions, and the hydrogen-induced threshold stress intensity factor (KIH) is obtained from the load-crack opening displacement (COD) curves.14,15) In these previous reports, ASTM E399 and ASTM E1820 methods were used to determine the lower bound stress intensity factors in hydrogen. However, the relationship between the obtained values and the extent of microscopic hydrogen effects is not well documented and therefore remains unclear.
In this study, using compact tension (CT) samples in which fatigue pre-crack had been induced, rising load tests were conducted in air and hydrogen gas, and the specimens were unloaded at each stage of the test to examine the K values at which cracks would begin to progress. Herein, we report the results of our investigation of the effects of hydrogen on crack initiation and propagation of SNCM439 steels in high-pressure hydrogen gas.
The chemical compositions of the samples are presented in Table 1. The sample material used was JIS SNCM439 steel, and after melting via 50-kg vacuum induction melting (VIM), 20 mm-thick sheets were hot forged and then normalized for 3 h at 950°C. For quenching, the sheets were heated to 830°C and held at that temperature for 1.5 h, and then oil cooled. For tempering, the sheets were heated to 550–650°C and held within that temperature range for 5 h, then air-cooled; the tensile strength of the samples was then adjusted to 800–1050 MPa. The tensile properties and fracture toughness values of the samples after the heat treatment are listed in Table 2.
| C | Si | Mn | P | S | Ni | Cr | Mo |
|---|---|---|---|---|---|---|---|
| 0.41 | 0.22 | 0.76 | <0.003 | <0.0012 | 1.72 | 0.82 | 0.20 |
| Tempering Temperature (deg.) | 550 | 600 | 625 | 650 |
| 0.2%Y.S. (MPa) | 921 | 782 | 714 | 655 |
| T.S. (MPa) | 1045 | 925 | 861 | 805 |
| E.L. (%) | 15.3 | 18.3 | 19.0 | 21.0 |
| R.A. (%) | 57.2 | 62.4 | 63.2 | 65.3 |
| Y.S./T.S. | 0.88 | 0.85 | 0.83 | 0.81 |
| KIC(J) (MPa-m1/2) | 154 | 205 | 233 | 245 |
The shapes of the CT specimen used in the rising-load tests are shown in Fig. 1. Notably, the final stress-intensity range associated with the pre-cracking operation did not exceed 12.5 MPa·m1/2 so that the plastic region near the crack tip would not affect the test results. After the introduction of pre-cracks, the tests were conducted at loading rates ranging from 0.02–20 MPa·m1/2·s−1 in air and in 20 MPa hydrogen gas at room temperature. Since the steel exhibited hydrogen embrittlement behavior even at low hydrogen gas pressure,16) it was considered the effect of hydrogen could be evaluated even at the test pressure of 20 MPa.
Compact tension (CT) specimen drawing showing important dimensions.
The rising load test was conducted using a testing machine with a pressure vessel capable of holding high-pressure hydrogen gas. The test specimen was placed in the pressure vessel and the vessel was evacuated using a rotary pump. Next, high-purity nitrogen gas was introduced into the vessel at a pressure of approximately 3 MPa and released at 0.1 MPa. This operation was repeated two times so as to reduce the oxygen concentration in the vessel, after which the vacuum was evacuated again. High-purity hydrogen gas was then introduced at a pressure of approximately 5 MPa and then released at 0.3 MPa. This operation was repeated three times to purge the inside of the vessel with hydrogen gas. Next, the pressure of the gas was increased to 20 MPa and sustained for 20 min before the evaluation test was conducted. The crack tip opening displacement (CTOD) was measured using a clip gauge manufactured by MTS, which has a proven track record of accurate measurements in high-pressure hydrogen gas.
To focus on the stretch zone (SZ) formed by plastic blunting and crack propagation in hydrogen gas at the tip of the pre-cracks, the samples were unloaded in the middle of the tests. After the tests, the fracture surfaces were opened and examined via scanning electron microscopy (SEM) using a JEOL JEM-6060 microscope.
Figure 2 shows the load-COD curve for the 800 MPa-class high-strength specimens tempered at 650°C. Figure 3 shows the results of the fracture surface observations by SEM after unloading at each position. As shown in the figure, the SZ expanded with the increase in load. The SZ width (SZW) was measured using these SEM images; specifically, the width of the SZ was measured multiple times in various spots over a length of 5 mm, with the average value being the SZW of the sample. The stress intensity factor (KI) for each load was determined using Eq. (1), which is as follows:
| (1) |
Load - COD curve for SNCM43 with a tensile strength of 800 MPa class in air. The circle symbols on the curve represent the positions where the fracture surfaces were observed upon unloading.
Scanning Electron Microscope (SEM) images of fracture surface at test specimens with a tensile strength of 800 MPa class in air. (a) unloaded at 82 MPa-m1/2, (b) unloaded at 115 MPa-m1/2, and (c) after fracture toughness testing.
Similar appraisals were performed on samples of varying strengths, with SZW measurements performed upon each unloading of the samples at various loads and after fracture toughness testing. The relationship between KI and SZW was determined and is illustrated in Fig. 4. As shown in Fig. 4, regardless of the tensile strength of the material, SZW and KI coexist on the same line prior to the initiation of stable fractures. The point of intersection with the vertical axis marks the point at which the SZ begins to form. However, the stress intensity factor at the crack initiation of a stable fracture (KIi) and the critical SZW (SZWc) decrease with the increase in tensile strength.
Relationship between stretch zone width and stress intensity factor in air.
Figure 5 shows the load-COD curves for the 800 MPa-class, 860 MPa-class, 930 MPa-class, and 1050 MPa-class high-strength specimens in air and hydrogen gas. The hydrogen gas diagram deviated from the air load-displacement curve in the region where the latter increases in a linear fashion, making its maximum load point lower than that of the air curve; the loads decreased thereafter. Additionally, these values decreased as the tensile strength of the material increased.
Load - crack opening displacement curve in air and in hydrogen gas. Tensile strength (a) 800 MPa class, (b) 860 MPa class, (c) 930 MPa class, and (d) 1050 MPa class.
Figure 6 shows the SEM images of the fracture surface of the crack tip of the 930 MPa-class specimens unloaded at 20 kN, 23 kN, and 27 kN during the rising load tests in hydrogen gas. For the specimen unloaded at 20 kN, no cracks were observed at the crack tip; however, in the specimen unloaded at 23 kN, hydrogen gas initiated cracks at the pre-crack tip. Further, for the specimen unloaded at 27 kN, whose load-COD deviated from linearity, cracks were observed to have grown and propagated significantly. These cracks initiated before the hydrogen load-displacement curve began to deviate from the results obtained by the air curve, and the cracks propagated significantly at the deviation point. In addition, SEM observation did not indicate the presence of cracks at the leading edges nor any traces of plastic deformation, such as SZs; further, the cracks induced by hydrogen were generated directly from the pre-crack tip. This trend was also observed for specimens with other tensile strengths. For instance, the 1050 MPa-class specimens showed no crack initiation up to 9 kN, but exhibited partial crack initiation at 10 kN. For the 800 MPa-class specimens, the deviation point was approximately 40 kN, but hydrogen-induced cracks were observed on those unloaded at 30 kN, which was much lower than the deviation point.
Fracture surface of crack tip from 930 MPa class specimens tested under rising load in hydrogen gas. Unloaded at (a) 20 kN, (b) 23 kN, and (c) 27 kN.
Figure 7 shows the relationship between KI for the tip of the crack at the time of loading and the amount of crack progression (Δa) in hydrogen gas observed at the leading edge of the fatigue pre-crack. Its intersection with the vertical axis of the graph represents the stress intensity factor for crack initiation in hydrogen (KIHi). Therefore, it was noted that as the tensile strength increases, the crack-generation stress intensity factor decreases, and the effect of hydrogen gas is stronger. In addition, the slope of the graph represents the rate of increase in the stress intensity factor necessary for cracks to progress, which decreased with the increase in the tensile strength. It is apparent from the graph that, owing to the effects of hydrogen gas, even a small increase in load can cause significant crack progression. In other words, the experiment demonstrated that the resistance necessary for crack progression depends on the tensile strength of the material, and that the greater the tensile strength of the material, the lighter the load at which the cracks progress.
Crack initiation and crack growth resistance in hydrogen gas.
Figure 8 shows the load-COD curves of the 1050 MPa-class specimens in hydrogen gas at different loading rates. In air, displacement occurred at a loading rate of 0.2 MPa·m1/2/s; however, the different loading rates did not have any overall significant impact on the curve. However, in hydrogen, the maximum load increased with an increase in the loading rate, and the shape of the load-displacement curve was very different from that of the curve obtained in air. The specimens with loading rates of 0.02 MPa·m1/2/s and 0.2 MPa·m1/2/s deviated at almost the same load; meanwhile, the specimens with a loading rate of 0.02 MPa·m1/2/s decreased in load with deviation and fractured immediately, whereas those with a loading rate of 0.2 MPa·m1/2/s increased in load after the deviation and then fractured. For the specimens with a loading rate of 2 MPa·m1/2/s, their rupture load increased after deviation. For the specimens with a loading rate of 20 MPa·m1/2/s, the deviation point was lower than 30 kN, but the maximum load before rupture reached approximately 70 kN, the value almost equal to that observed for the specimens in air.
Load - crack opening displacement curve of 1050 MPa class with different loading rate in hydrogen gas.
Figure 9 shows the SEM observations regarding the crack tip after unloading in the middle of the rising load test. For tests with a loading rate smaller than 2 MPa·m1/2/s, specimens unloaded at 13 kN were observed, and for tests with a loading rate of 20 MPa·m1/2/s, specimens unloaded at 28 kN were observed. In all the specimens, hydrogen-induced crack propagation of 100–200 μm was observed. No traces of plastic blunting, such as SZs, were observed at the boundary between the hydrogen-induced crack initiation region and the fatigue pre-crack tip.
Fracture surface of crack tip from 1050 MPa class specimens tested under rising load in Hydrogen gas. (a) Loading rate is 0.02 MPa-m1/2/s and unloaded at 13 kN, (b) Loading rate is 0.2 MPa-m1/2/s and unloaded at 13 kN, (c) Loading rate is 2 MPa-m1/2/s and unloaded at 13 kN, and (d) Loading rate is 20 MPa-m1/2/s unloaded at 28 kN.
Figure 10 shows the stress intensity factors at the time of crack initiation in air and in hydrogen gas. In the figure, the value of the stress intensity factor for SZ initiation, as obtained in Fig. 4, is also indicated with a dashed line. The initiation of cracks in hydrogen gas occurred at stress intensity factors with values far lower than those in air; with the increase in the tensile strength, these values decreased even further. In air, the higher the tensile strength of the material, the larger the stress intensity factor at the start of stable fracture linearly. However, in hydrogen, for materials with tensile strengths lower than 860 MPa, the stress intensity factor saturated to a constant value. The value was equal to that at which the SZs began to form in air. Any traces of plastic deformation, such as the SZ observed shown in Fig. 4, were observed on the fracture surface in hydrogen, and the hydrogen-induced cracks were generated directly from the pre-crack tip. Since SZs are created by slippage, the fact that crack initiation in hydrogen was saturated at the stress intensity factor for SZ initiation presumably indicates that hydrogen-induced cracking occurred simultaneously with slip initiation. Since hydrogen-induced cracking is caused by the penetration of hydrogen into the steel and stress,17,18) it is inferred that hydrogen was transported into the steel by plastic sliding and cracks were generated immediately. This inference is supported by the SEM image of the fracture surface of the crack tip of an 800 MPa-class specimen (Fig. 11), which shows no SZ (unlike that observed in Fig. 3) and demonstrates that cracking occurred without evidence of significant plastic slip. This result is also consistent with the mechanism by which fatigue cracks propagate forward with little opening at fatigue crack propagation tests in hydrogen.19) On the contrary, for materials with tensile strengths exceeding 900 MPa, their KIHi values drop progressively, and hydrogen-induced cracks likely formed in the small-scale-yielding state before plastic blunting occurred. As the tensile strength and yield strength increased, the values of elongation and the reduction in area decreased progressively, and the effects of hydrogen gas became pronounced;20,21) this suggests that hydrogen-induced crack initiation is affected by local plastic deformation at the crack tip before plastic slip occurs and hydrogen penetrates into the steel, resulting in early crack initiation.
Relationship between stress intensity factor for crack initiation and tensile strength of SNCM439 in air and in hydrogen gas.
Fracture surface of crack tip from 800 MPa class specimens tested under rising load in hydrogen gas.
Furthermore, as shown in Fig. 5, as the tensile strength decreases, the curves for hydrogen and air become closer and the effect of hydrogen becomes smaller, which can be explained by the value of the stress intensity factor for crack initiation in hydrogen and the effect of crack growth resistance after crack initiation. The slope in Fig. 7 represents the crack propagation resistance. The smaller the tensile strength of the material, the greater the resistance to crack propagation; therefore, the maximum loads were large even though the cracks occurred early because more energy was required to reach the fracture. On the other hand, the higher the tensile strength of the material, the smaller the slope, the earlier the crack initiation, and the lower the resistance to crack propagation.
Thus, the effect of hydrogen is more pronounced as the tensile strength of the material increases. The localization of plastic deformation and the enhancement of void formation due to the presence of hydrogen in steel has been observed,22,23) and it is believed that the interaction between dislocations activated by plastic deformation and hydrogen is involved in hydrogen-induced fracture. Figure 12 shows the transmission electron microscopy images of precipitation in the 800 MPa-class specimen (Fig. 12(a)) and 1050 MPa-class specimen (Fig. 12(b)) by a carbon extraction replica method. In Fig. 12(a), spherical precipitates are observed and the density of the precipitates is low. However, in the material with a higher tensile strength (Fig. 12(b)), before coarsening at the lath interface and block interface of the martensitic structure, long precipitates and short precipitates were observed and the density of the precipitates was high. In addition, since the tensile strength was varied by changing the tempering temperature, it was expected that the dislocation density would be higher for materials with higher tensile strength because of their lower tempering temperature; this is because the recovery of dislocations introduced by quenching does not progress sufficiently in high-tensile-strength materials. It is thought that these factors act as obstacles to dislocation motion, causing dislocations to accumulate in high strength materials even at small deformations, leading to crack initiation. Hydrogen penetration into the steel, which promotes localized plastic deformation and void formation, is thought to reduce the stress level at crack initiation, leading to early fracture in steels in hydrogen. However, they are also thought to be closely related to the microstructure and dislocation structure of the material; thus, the elucidation of these mechanisms is a future issue.
TEM images of precipitation by a carbon extraction replica method: (a) 800 MPa class specimen; (b) 1050 MPa class specimen.
Figure 13 shows the relationship between the loading rate and the stress intensity factor for crack initiation in hydrogen for specimens with four classes of tensile strengths. Specimens with the same class of tensile strength were unloaded at two or more load conditions at the same loading rate to measure the amount of crack growth, and the stress intensity factor for crack initiation in hydrogen was determined. At a loading rate of 0.2 MPa·m1/2·s−1 or lower, the stress intensity factor of crack initiation in hydrogen was hardly affected by velocity and therefore remained almost constant. The value of the stress intensity factor for crack initiation in hydrogen slightly increased when the loading rate was 2 MPa·m1/2/s, and the value of the stress intensity factor for crack initiation in hydrogen rapidly increased by increasing the loading rate, up to 20 MPa·m1/2/s. It was considered that hydrogen affected the steel because it is transported into the steel together with the dislocation. Therefore, upon increasing the loading rate, hydrogen could not follow the migration rate of the dislocations; the dislocation moved faster than hydrogen and the hydrogen was released from the mobile dislocation.24) Consequently, the effect of hydrogen on the material was reduced.
Relationship between loading rate and stress intensity factor at crack initiation in hydrogen gas.
Figure 14 presents a schematic of the process by which cracks develop in air and hydrogen gas. As can be seen, the plastic region at the tip of the crack expanded with an increase in load. In air, during the loading process, SZs were formed depending on the degree of plastic slip, with stable cracks subsequently forming in front of these zones as the SZW became saturated. In contrast, in hydrogen gas, as plastic deformation occurs and hydrogen penetrates into the steel along with the dislocations, hydrogen-induced cracking occurs before the SZ occurs owing to the effects of plastic deformation localization and void formation enhancement; subsequently, these cracks propagated and then expanded. In other words, this embrittlement behavior demonstrated in hydrogen gas at the formation of crack tips can be explained by the generation of new surfaces through plastic deformation at the crack tips or plastic slip, as well as by the infiltration and diffusion of hydrogen through these new surfaces. However, more detailed studies are required to elucidate the dominant factors.
Schematic diagram of crack initiation process.
In this study, a rising load experiment was conducted using SNCM439 steel in both air and high-pressure hydrogen gas environments at room temperature to investigate the crack initiation and propagation behavior due to hydrogen gas embrittlement. In hydrogen gas, the stress intensity factors at the time of crack initiation were significantly lower than those in air. Moreover, the stress intensity factors for crack initiation decreased with the increase in the tensile strength of the material. SEM observations of the fracture surfaces of the specimens unloaded during the middle of the rising load tests, as indicated by load-COD curves; these indicated that cracks affected by hydrogen gas were already initiated at the load value on the linear curve where the curve of the rising load test in hydrogen was lower than the load value at which the hydrogen curve deviated from the curve in air.
The SZs that appeared in the rising load tests in air were not observed for those conducted in hydrogen gas. It was confirmed that hydrogen embrittlement cracks were initiated and propagated directly from the tips of the fatigue pre-cracks. For materials with tensile strengths below 860 MPa, hydrogen-induced cracks appeared starting at K values roughly equivalent to those at which SZs began to be observed in the air tests. The hydrogen gas embrittlement was induced by hydrogen penetration and transportation into the steel, which was caused by plastic slip at the crack tip. For the materials with greater tensile strengths, namely those in the 930 MPa- and 1050 MPa-classes, cracks were initiated at K values lower than those observed at the onset of plastic slip. This was considered to be due to the fact that hydrogen-induced crack initiation of the high-tensile-strength materials occurs at the small scale yielding of the pre-crack tip, and the resistance of crack propagation in hydrogen was weak, indicating that a slight increase in the load could cause significant crack propagation in these materials. Since this phenomenon may be related to the generation of the local plastic strain at the crack tip and microstructures such as precipitate dispersions; more detailed investigations are needed to clarify the dominant factors in these processes.
When the loading rate was increased to 20 MPa·m1/2·s−1, the stress intensity factor for crack initiation in hydrogen increased rapidly. This was attributed to the fact that the hydrogen could not match the dislocation movement speed, and as a result, the dislocations moved away from the hydrogen, thereby reducing the effect of hydrogen on the steel.
We would like to thank Dr. Ono, Dr. Itoh, and the late Dr. Ohnishi, formerly of Japan Steel Works, Ltd., for their collaboration during the early stages of this work and for their useful discussions.
There are no conflicts to declare.
