2022 年 62 巻 5 号 p. 832-839
The interactions between hydrogen, dislocations and vacancies that lead to the hydrogen embrittlement of iron-based materials have remained largely unknown with major impediments for the development of the infrastructure for hydrogen transport and storage. Recent studies of the hydrogen-induced lattice defects formed in pure iron and common austenitic stainless steels by positron annihilation lifetime spectroscopy have provided a breakthrough in understanding the controlling factors of the hydrogen embrittlement process. In this review, the main results of those studies are summarized and discussed together with current knowledge of hydrogen-related defects. The formation of vacancy-hydrogen complexes coupled to a plastic strain localization which is large enough to lead to hydrogen-enhanced vacancy clustering during the plastic deformation appears to be the likely factor that triggers the hydrogen embrittlement of bcc(α-) and fcc(γ)-iron.
Hydrogen embrittlement (HE) of metallic materials is one of the most urgent issues that needs be solved for industrial applications such as the development of the hydrogen infrastructure for a more sustainable society.1,2) HE is characterized by the deterioration of the mechanical properties of susceptible materials, such as iron and stainless steels, due to hydrogen.3,4) The deterioration mechanism consists in the lattice embrittlement in which hydrogen penetrates within the crystal lattice and reduces the interatomic bonding force.5) Although a variety of mechanisms3,6) have been proposed to explain the origin of the embrittlement process, HE remains to date an unsolved problem.7) Amongst those models, the hydrogen-enhanced local plasticity (HELP) mechanism8) is a notable one according to which hydrogen promotes the dislocation motion and induces local plastic deformation. Alternatively, the hydrogen-enhanced strain-induced vacancy (HESIV) formation model9) argues that hydrogen promotes the formation of strain-induced vacancy cluster which destabilize the local plastic deformation.
A huge number of papers on HE in steel-related materials have been published by researchers in a large variety of fields3,6,10,11) such as metallurgy, applied physics and mechanical engineering, using techniques within their own specialized research field. In addition, steel-related company engineers have reported many phenomenological studies, but the results are often challenging.12,13,14) Regarding material properties, studies commonly focus on the observation of the microstructure and dislocations by transmission electron microscopy or scanning electron microscopy,15,16,17) but it is important to understand the interaction between hydrogen, dislocations and vacancies to clarify the critical elements in HE. In the research from the viewpoint of the mechanical properties, the discussion is based on cracks and voids, while the factors leading to them are rarely discussed.18,19,20) Dislocations, stacking faults, and atomic vacancies have been proposed as the potential controlling factors of HE.21) The conventional method for investigating the vacancy behavior in iron-based materials is thermal desorption analysis (TDA), which is a technique that determines the hydrogen-trapping defect species from the hydrogen desorption temperature.9,22) TDA is a powerful analytical method, but it is often difficult to identify the specific defect species because the spectral analysis itself is challenging and requires numerical simulations.23) One main disadvantage of TDA lies in the possibility that the defects change when raising the temperature. While TDA is an indirect technique for the detection of lattice defects, positron annihilation lifetime spectroscopy (PALS) allows one to directly determine the defect species in which positrons are trapped from their annihilation lifetime.24) By comparison of the measured positron lifetime with first-principle calculations, it is possible to determine the type and size of vacancies (Fig. 1). As such, positrons are non-destructive, highly sensitive and direct probes of vacancy-type defects26) and are considered possibly the ideal tool for the atomic vacancy analysis of HE.27)
Positron lifetime at vacancy clusters in α-Fe as a function of cluster size obtained from the first-principle calculations by Ohkubo et al.25) The line is drawn just to guide the eye.
In this article, we review recent progress in our understanding of the HE mechanism in pure α-iron and common austenitic stainless steels from an atomistic point of view. On top of summarizing our general understanding of the HE behavior, we report the results of studies involving PALS, which has been shown to be a very promising technique for the elucidation of the HE mechanism. In particular, we provide a focus on current studies that uncovered the controlling defects of HE by the direct detection of atomic vacancies using novel PALS-based approaches.28,29,30) The elucidation of the HE mechanism is extremely important for the design and development of hydrogen-resistant steel materials to meet the requirements of the new hydrogen era.
In iron-based materials, hydrogen is generally added by electrochemical or gaseous charging.31) For cathodic electrolytic charging an electrochemical cell is employed, in which the sample acts as the cathode and a piece of platinum acts as the anode, both submerged in an electrolyte. Typically, a NH4SCN + H2SO4 or NaOH or NaCl electrolytic aqueous solution is used. By applying an electrical potential across the electrodes, the electrolytic solution decomposes to generate hydrogen ions which accumulate on the surface of the electrodes. Charging is performed at room temperature or higher temperatures with varying current densities on the order of several 10 A/m2 and for variable times ranging from less than 1 h for α-iron to a few days for austenitic stainless steels. With this charging method, from the known large hydrogen diffusivity in α-iron at around room temperature (Fig. 2), hydrogen is expected to diffuse within the entire specimen. Note that since the diffusion coefficient of hydrogen in α-iron at room temperature is very large, the plastic deformation of pure iron needs be performed directly in a hydrogen environment in order to avoid hydrogen desorption during the tensile testing. On the other hand, in γ-iron, which has a hydrogen diffusion coefficient 6 orders of magnitude smaller than α-iron at room temperature (Fig. 2), hydrogen tends to be mostly concentrated in the ~10-μm topmost layer and the hydrogen concentration in the sample decreases exponentially with the depth from the surface (Fig. 3).33) Hydrogen can also be charged by exposing the samples to hydrogen gas at high-temperature and high-pressure, typically in the range of ~100 MPa and 80–250°C for times ranging from a few days to weeks. The advantage of this method is that hydrogen is fairly uniformly introduced into the entire sample with relatively large concentrations of tens ppm (Fig. 3).35)
Diffusion coefficient of hydrogen in ferrite and austenite as a function of temperature. Reprinted from Bhadeshia.32) Copyright (2016) The Iron and Steel Institute of Japan.
Typical depth distribution of H content in a high-entropy alloy and austenitic stainless steel 316L by gaseous (G) and electrochemical (E) charge. Reprinted from Zhao et al.34) with permission from Elsevier. (Online version in color.)
Pure iron is the main component of ferritic steels and represents the simplest model to study HE due to the absence of impurities which complicate the interpretation of empirical results and numerical simulations. In α-iron, HE strongly depends on the tensile strain rate: deformation at slow strain rate (~10−5/s) was found to induce HE, whereas fast straining (~10−3/s) does not affect the material properties or fracture mode (Fig. 4).36,37) Numerous first-principles calculations have shown that hydrogen reduces the vacancy formation energy and that the vacancy-hydrogen (VH2) complex composed of two hydrogen atoms trapped to a monovacancy is the most stable form of vacancies.38,39,40) This suggests that hydrogen trapping stabilizes monovacancies, which are unstable at room temperature. Different kind of experiments confirmed that the presence of hydrogen in the material during the straining process leads to the formation of hydrogen-induced defects.41,42,43,44,45) In particular, by the strain-assisted transport of hydrogen, e.g. by dislocation drag, deep into the bulk during the plastic deformation, the formation of atomic vacancies is promoted. These hydrogen-induced vacancies acting as trapping sites of hydrogen highlight the important role of vacancies in the HE mechanism.
Nominal stress vs. nominal strain of hydrogen-charged pure iron deformed at different strain rates. (
) slow strain rate, (
) fast strain rate. Reprinted from Chiari et al.28) with permission from Elsevier. (Online version in color.)
Low-temperature TDA measurements have shown that the amount and distribution of hydrogen trapping sites, such as dislocations and vacancies, in hydrogen-charged strained α-iron are enhanced.9,46,47,48) The activation energy of trapped hydrogen in hydrogen-embrittled pure α-iron was found to be very small, and all the trapped hydrogen is desorbed by aging at room temperature for about 1 hour in the atmosphere.47) In addition, α-iron pre-strained in a hydrogen environment exhibited a ductility reduction, even though hydrogen was absent in the subsequent deformation stage.22) This suggests that HE is due to the hydrogen-related vacancies created during the plastic deformation, rather than to hydrogen itself.
Defect analysis in α-iron by PALS42,43,44,45) has revealed that although only dislocations formed by tensile deformation of pure α-iron, straining in a hydrogen environment promoted the formation and accumulation of vacancies. This result stressed not only the importance of the interaction of hydrogen with strain-induced vacancies but also that the presence of hydrogen enhances the generation of vacancies. However, the fact that the deformation in most of those studies was carried out after the hydrogen charge and the consequent hydrogen desorption leaves open the possibility that the detected defects might not correspond to the defects generated during the deformation in the hydrogen environment.
In a recent study, Chiari et al.28) carried out tensile deformation of pure α-iron in a hydrogen environment at slow and fast strain rate up to a unified strain of 16% (i.e. the elongation at break of the hydrogen-embrittled sample) and subsequently quenched the samples in liquid nitrogen to preserve the state of defects formed during the deformation. Temperature-variable PALS measurements confirmed the formation of vacancy clusters in both samples (Fig. 5), i.e. irrespective of HE taking place or not, suggesting that the vacancy clusters are not the primary defects of HE. However, at low temperatures smaller vacancy clusters were induced in the sample strained at slow strain rate than in that strained at fast rate, but the former grew much bigger in size when the temperature was raised. Such trend in the vacancy growth of the slowly strained sample only was explained in terms of the local accumulation of vacancies in high concentrations. Since monovacancies are unstable above 200 K, this can only happen if hydrogen is trapped at the vacancies, i.e. the formation of VH2 complexes was speculated. By annealing at higher temperatures, the VH2 complexes supposedly dissociated into two hydrogen atoms and monovacancies that, by regaining their mobility, aggregated into vacancy clusters. Although the re-trapping of hydrogen at the vacancies in this process cannot be ruled out, eventually all hydrogen was desorbed at high temperatures.
Results of the temperature-variable PALS results of hydrogen-charged pure iron deformed at different strain rates. Fast strained sample: (a) lifetime and (b) intensity, slowly strained sample: (c) lifetime and (d) intensity. (
) bulk, (
) dislocations/monovacancies, (
) vacancy clusters. Reprinted from Chiari et al.28) with permission from Elsevier. (Online version in color.)
Since austenitic stainless steels are employed as structural materials in the hydrogen infrastructure, a large amount of empirical data on HE in these materials has been accumulated over the years3,32) and yet the fundamental mechanism that drives it could not be completely elucidated.6) Amongst the austenitic stainless steels, the common AISI 304 grade greatly suffers from HE at room temperature.49) In fact, it is recognized that a low γ-phase stability increases the hydrogen susceptibility in austenitic stainless steels.50)
TDA measurements have shown that hydrogen is not easily desorbed from austenitic stainless steels at room temperature, because of the small hydrogen diffusivity in γ-iron at room temperature.51,52) For this reason, defect analysis in austenitic stainless steels by TDA has remained quite limited.51,52)
It has long been known from XRD measurements that hydrogen charging of austenitic stainless steel 304 induces the transformation from the austenite phase to the ε and α′ martensite phases, as well as the formation of austenite hydrides.53,54) Under cathodic electrolytic charging, even though hydrogen is typically added only to the topmost ~10-μm layer in a sample with a thickness of 200 μm, and consequently the ε and α′-phase are formed only in the charged layer, HE was nonetheless confirmed.35) Note that the effect of the hydrogen-charged layer may be strongly dependent on the thickness of the sample, so that HE might not take place in an overly thick sample (i.e. more than several hundred μm). However, HE was found to occur even when no strain-induced α′-phase was observed,27) so that the formation of the α′-phase was understood not to be a direct cause of HE. In addition, the strain-induced martensite phase was not detected in hydrogen-embrittled 304, so that the γ-phase itself is thought to be embrittled.27)
Several measurements and numerical simulations demonstrated that hydrogen enhances the velocity of dislocations,55) highlighting a close relationship between hydrogen and lattice defects, as well as the microstructure.56) In particular, PALS measurements27,57,58,59) have found that the interaction between dislocations and hydrogen enhances the formation of strain-induced vacancies which agglomerate into vacancy clusters. As a consequence, the role of hydrogen-enhanced strain-induced vacancies in the HE of austenitic stainless steels was thought to be crucial.9) Nonetheless, the presence of vacancy clusters was later confirmed not only in hydrogen-embrittled austenitic stainless steels but also in non-embrittled austenitic stainless steels,27) so that it was understood that they cannot be directly regarded as the controlling defects of HE, but rather the final product.
A comprehensive review of the PALS studies of lattice defects in hydrogen-charged strained steels has been recently compiled by Sugita et al.60) (Fig. 6) A strong tendency towards larger positron lifetimes of the vacancy cluster component for tensile-deformed metals with lower fracture strains was found in many of the PALS studies summarized there.
Dependence of the positron lifetime of the vacancy cluster component and the fracture strain of hydrogen-free and hydrogen-charged metals. Reproduced from Sugita et al.60) Copyright (2021) The Iron and Steel Institute of Japan.
A very recent PALS study by Chiari et al.29) has recently shed new light on the defects responsible for HE in austenitic stainless steel 304. As in earlier studies,27) they confirmed the generation of vacancy clusters after hydrogen addition and application of 10% strain due to the effect of hydrogen introduced by cathode electrolysis. However, when the topmost 10-μm hydrogen-charged layer was electropolished, only a positron lifetime component of 180 ps longer than that of dislocations (130–160 ps) was detected (Fig. 7), which suggested that this lifetime includes a vacancy component. Hydrogen is required for the generation of vacancies, so that the formation of vacancies suggests the promotion of hydrogen diffusion. When the sample was further strained, the ductility was found to be reduced and vacancy clusters were detected again. These are considered to be the defects in which the vacancies observed in the sample strained by 10% and electropolished evolved and lead to the decrease in ductility, or HE. Since monovacancies are expected to be unstable at room temperature even in γ-iron (defect energies are similar to those in α-iron61)) and hydrogen is known to stabilize vacancies, these defects were determined to be VH2 complexes. The VH2 complexes were found to be the precursor of the vacancy clusters that develop into cracks and cause the HE, so that they were identified as the defects responsible for the HE in austenitic stainless steel 304.
PALS results of hydrogen-charged austenitic stainless steel 304 strained by 10%. From left to right: as prepared; 10 μm-polished; 10 μm-polished and fractured samples. (
) Bulk, (
) dislocations, (
) dislocations and vacancy-hydrogen complexes, (
) vacancy clusters. Reproduced from Chiari et al.29) Copyright (2021) The Iron and Steel Institute of Japan. (Online version in color.)
Unlike in metastable austenitic stainless steels, the hydrogen susceptibility of austenitic stainless steels with a higher Ni-equivalent is greatly reduced, e.g. in the extra-low carbon grade and stable austenitic stainless steel 316L.27,33,62) Nonetheless, even stable austenitic stainless steels can suffer from severe HE in a hydrogen environment under certain environmental conditions, such as low temperatures and high pressures.63,64) Specifically, austenitic stainless steel 316L is known to become hydrogen embrittled when deformed at low temperatures from about −100°C to −40°C, while it has high hydrogen resistance at other temperatures.50,65)
It has been well established experimentally that, the lower the temperature, the more easily the strain-induced martensitic transformation occurs.54,66) In austenitic stainless steels with a high Ni equivalent, which is a strong austenite stabilizer, such as 316L, the lower HE susceptibility is ascribed to the high γ-phase stability.33,50,67,68) However, this phenomenon cannot be explained simply by the decrease in the diffusion coefficient of hydrogen.69) The fraction of martensite phase in the fractured material is not affected by the addition of hydrogen irrespective of the straining temperature, so that it is thought to be almost entirely a strain-induced structure. Nonetheless, hydrogen has been shown to enhance the localization of deformation which, in turn, supports the martensitic transformation.70,71,72) In fact, a correlation between the HE susceptibility of this steel at −70°C and the fraction of the α’-phase was found.50)
In hydrogen-charged austenitic stainless steel 316L by cathodic electrolysis, TDA measurements have revealed a hydrogen content as high as 80 ppm concentrated within 90 μm from the charged surface for a charging time of 72 h.73) The hydrogen concentration rapidly decreased during the first 48 h due to hydrogen desorption and remained almost constant after that as a result of trapped hydrogen. Also upon application of stress, the hydrogen desorption was found to rapidly increase due to dislocation-assisted hydrogen drag to the surface.57)
Defect analysis in austenitic stainless steel 316L by PALS has shown that the average concentration of lattice defects is enhanced by plastic deformation in the presence of hydrogen owing to the interaction of hydrogen with dislocations and vacancies.27) The formation of vacancy clusters in hydrogen-charged austenitic stainless steel 316L was confirmed and they were found to ultimately lead to the brittle fracture.27) However, vacancy clusters were detected even in non-hydrogen-embrittled 316L, so that HE cannot be directly linked to their presence.27)
In austenitic stainless steel 316L with a Ni content of 12% by mass, Komatsu et al.30) visualized the effect of the deformation on the microstructure with kernel average misorientation (KAM) maps obtained by electron backscatter diffraction (EBSD). They found that hydrogen promotes a higher level of strain inhomogeneity and the generation of high-strain zones within the grains in the hydrogen-embrittled sample at −70°C (Fig. 8). Low-temperature PALS measurements after straining at −150°C in a hydrogen environment exhibited the formation of dislocations (140–160 ps) and vacancies with a lifetime of about 200 ps, which quickly grew in size as the temperature was raised. At the same time their concentration decreased as indicated by the reduction in the intensity of their component. This suggests that small vacancies bound to hydrogen were generated locally in high density in high-strain regions, such as the boundaries of the α’ and ε-phase or stacking faults, during the plastic deformation in order to rapidly agglomerate into large vacancy clusters as hydrogen desorbed with increasing temperature. However, the VH2 complexes are not mobile at the lowest temperatures, so that they could not trigger the HE. Note that, in general, excess vacancy clusters can change to stable Frank dislocation loops in γ-iron during annealing, so that a contribution from the formation of such defects in the concentration reduction of the vacancy clusters cannot be excluded. In the specimen fractured at −70°C (Fig. 9), vacancy clusters with a positron lifetime of 325–375 ps were formed at the lowest temperatures, i.e. somewhat larger in size than in the sample strained at −150°C. The vacancies could not grow in size as much as in in the sample fractured at −150°C, because the initial vacancies had already agglomerated due to the VH2 complexes becoming mobile, but they became large enough to grow into large voids and lead to the HE. Straining at room temperature produced vacancy clusters with a positron lifetime of 250–300 ps, i.e. a size in between that in the samples fractured at low temperatures, and exhibited no temperature dependence of the vacancy cluster component. This implies that even if the vacancies are formed, they are heterogeneously distributed and without a suitable environment for the vacancy agglomeration the HE could not be triggered. These results clearly showed that the localized accumulation of vacancies in high-strain regions is the key of the HE mechanism in austenitic stainless steel 316L.
KAM maps of austenitic stainless steel 316L strained by 10% at −70°C: (a) hydrogen free, (b) hydrogen charged. Reprinted from Komatsu et al.30) with permission from Elsevier. (Online version in color.)
Low temperature PALS results of hydrogen-charged austenitic stainless steel 316L fractured at −70°C: (
) bulk, (
) vacancy clusters. Reprinted from Komatsu et al.30) with permission from Elsevier. (Online version in color.)
Despite the vigorous research activity of the recent decades, the reason why the defects responsible for HE had not yet been identified likely lies in the fact that the vacancy-hydrogen complex, which is hypothesized to be the controlling defect of HE in ferritic steel, is unstable at room temperature.9) Hydrogen facilitates the formation of monovacancies, but even if vacancies-hydrogen complexes are formed, hydrogen is desorbed by aging at room temperature and the remaining monovacancies disappear or aggregate into vacancy clusters, as they are unstable.74) Therefore, simple measurements of hydrogen-embrittled samples fell short of clarifying the core defects of the HE process.
Recent measurements have addressed these issues by focusing on empirical approaches that preserve hydrogen during the deformation as well as during the measurements, thus providing a picture that represents the real status of the hydrogen-related defects. This enabled us to make important steps forward in the elucidation of the defects responsible for HE. In pure iron,28) the formation of vacancy-hydrogen complexes was found to be essential to explain the growth of vacancies in hydrogen-embrittled iron. In austenitic stainless steels,29,30) vacancy-hydrogen complexes were found to locally form and aggregate in high-strain zones, such as the boundaries between different phases or stacking faults (e.g. see the Graphical abstract). By application of tensile stress and the concomitant hydrogen desorption, the vacancy-hydrogen complexes dissociate and vacancies agglomerate into vacancy clusters which grow in size and become the starting point of the fracture which causes HE.
Regarding the HE modality, many earlier PALS studies have observed a strong correlation between the hydrogen-enhanced strain-induced vacancy size and the premature fracture of the material due to hydrogen.60,75,76) At the same time, more recent PALS studies have observed a close relationship between the localization of hydrogen-enhanced strain-induced vacancies and the HE.28,29,30) This suggests that, in general, the plastic strain localization in a hydrogen environment is consistent with the hydrogen-enhanced vacancy clustering during the plastic deformation. In other words, the interaction between the HELP and HESIV mechanisms appears at present to be the most likely critical factor behind the HE phenomenon. A comprehensive model encompassing multiple embrittlement mechanisms rather than a unified view of the HE process is in agreement with the conclusions of the earlier study by Sugita et al.60)
The mechanism of deterioration of the mechanical properties of iron-based materials in a hydrogen environment has not been elucidated and is one of the major issues for a hydrogen-based society. Recent developments in clarifying the controlling defects of hydrogen embrittlement in pure iron and austenitic stainless steels 304 and 316L through innovative approaches based on positron annihilation lifetime spectroscopy represent an important step forward in our understanding of the embrittlement mechanism from an atomistic point of view. These results are expected to have an important impact on future studies on hydrogen-related defects and open up new pathways for research. Nonetheless, more research is warranted in the future to clarify the role of the relationship between plastic strain localization and the formation of hydrogen-enhanced strain-induced vacancies in the HE mechanism.
