Hydrogen Embrittlement Mechanism of Ultrafine-grained Iron with Different Grain Sizes

Abstract

To investigate the effect of grain sizes on hydrogen embrittlement of 4N-purity iron, miniature tensile tests were conducted after hydrogen charging for the ultrafine-grained specimens produced by high-pressure torsion and subsequent annealing. Hydrogen embrittlement indexes defined from reduction of area were increased with decreasing grain size, and shear-type fracture occurred with fine dimples on the fracture surface of the diagonally raptured tensile specimen with a smaller grain size. The formation and growth of microvoids at triple junctions of grain boundaries ahead of propagated cracks were responsible for such earlier shear-type fracture because necking between adjacent microvoids more likely and extensively occurred. In the specimens with larger grain sizes or without hydrogen charging, on the other hand, local coalescence and growth of microvoids were predominant due to longer distances between triple junctions, resulting in void coalescence-type fracture with coarser dimple patterns. Therefore, hydrogen atoms introduced by hydrogen charging are considered to enhance the formation of deformation-induced vacancies in ultrafine-grained iron, resulting in shear-type fracture with finer dimple patterns.

1. Introduction

In recent years, hydrogen energy is expected to be utilized to realize a low-carbon society because hydrogen can be produced from various resources and does not emit CO₂ in use. However, in addition to the development of infrastructures for storage and transportation of hydrogen in the social system, hydrogen embrittlement of the containers is also an urgent issue for metallic materials, because their strength and fracture toughness are gradually degraded, leading to premature failures. Especially, high-strength steels are well-known to be more susceptible to hydrogen embrittlement with increasing their strength,¹⁾ and thus it is important to elucidate hydrogen embrittlement mechanism of iron for further expansion of application ranges of high-strength steels.

Usually, hydrogen atoms dissolved into metallic materials not only exist in crystal lattices, but also are trapped by lattice defects such as dislocations,^2,3,4,5,6,7) vacancies^8,9) and grain boundaries.^6,10) Since these lattice defects are mainly responsible for plastic deformation and fracture of metallic materials, the relationship between hydrogen atoms and each type of lattice defects has been studied by many researchers. For example, the hydrogen-enhanced localized plasticity (HELP) theory¹¹⁾ is known as a mechanism of hydrogen embrittlement originating from interaction between hydrogen atoms and dislocations. This mechanism assumes that hydrogen atoms enhance local plastic deformation in crack tip regions and promote the propagation of cracks because of the increased mobility of dislocations.¹²⁾ In contrast, the hydrogen-enhanced strain-induced vacancy (HESIV) theory¹³⁾ differently explains the promoted propagation of cracks, i.e. the concentration of vacancies is increased by plastic deformation, and thus the formation and growth of microvoids are enhanced by agglomeration of the vacancies.¹⁴⁾

As for the effect of grain boundaries, furthermore, it is reported that intergranular fracture more likely occurs if larger hydrogen content is charged in metallic materials.^15,16,17,18) Some results of molecular dynamics calculation attributed this intergranular fracture to a decrease in cohesive energies of grain boundaries by hydrogen atoms.^19,20) However, there are contradictory reports that hydrogen embrittlement is rather suppressed by grain refinement of metallic materials.^{21,22,23,24,25,26,27,28,29)} This discrepancy may arise from the fact that those reports detect not only the effect of grain boundaries but also the effect of accompanying microstructural changes. For example, elongation to fracture of tensile-tested high-Mn steels was kept at larger levels, because deformation twins are less formed, and thus hydrogen atoms introduced by hydrogen charging are less trapped^24,25) with decreasing grain sizes. In tensile-deformed Fe-Ni alloys, furthermore, dislocation densities within slip bands decreased with decreasing grain sizes, resulting in the suppression of hydrogen embrittlement due to the decreased hydrogen contents transported to grain boundaries via dislocations.²³⁾ Therefore, it is essential to investigate the effect of interaction between hydrogen atoms and grain boundaries separately from that of accompanying microstructural changes by grain refinement.

In this study, ultrafine-grained 4N-purity iron with various grain sizes was prepared by high-pressure torsion (HPT)³⁰⁾ and subsequent heat treatment to eliminate the effect of microstructural changes caused by grain refinement. Then, miniature tensile tests were conducted after hydrogen charging for different charging times to investigate the effect of grain sizes on hydrogen embrittlement of 4N-purity iron. It should be noted that HPT processing can produce ultrafine grains of less than 1 μm by applying a large amount of strain even for pure metals without contaminated by impurities. Mine et al.³¹⁾ and Todaka et al.³²⁾ conducted tensile tests of HPT-processed Fe-0.01 mass% C after hydrogen charging and small punch tests of HPT-processed ultra-low carbon steel in a hydrogen environment, respectively, to investigate hydrogen-induced changes in their mechanical properties. While those results suggested that hydrogen embrittlement more likely occurs in the specimens with smaller grain sizes, the fracture mechanism has not been fully discussed, and thus the effect of grain boundaries on hydrogen embrittlement of iron is not clear yet. Therefore, the purpose of this study is two-fold: To elucidate the fracture mechanism based on the observed fracture surface after tensile tests as well as that just before fracture, and to clarify the effect of grain boundaries on hydrogen embrittlement of ultrafine-grained 4N-purity iron with various grain sizes.

2. Experimental Procedures

2.1. Samples and Heat Treatment

A 4N-purity 1 mm-thick iron sheet with a composition shown in Table 1 was utilized in this study. The disks with 20 mm diameter were cut from the sheet by electrical discharge machining and annealed at 750°C for 3 h in Ar atmosphere (Anneal sample). Then, HPT processing was performed on the samples under a pressure of 1.5 GPa with a rotation speed of 0.25 rpm for 5 turns (HPT sample). After that, dumbbell-shaped specimens with gauge length of 4 mm^l×1 mm^w×0.6 mm^t and disk-shaped specimens with 8.6 mm in diameter were cut from the HPT samples by electrical discharge machining for miniature tensile tests and thermal desorption spectrometry (TDS), respectively, as shown in Fig. 1. Both shapes of specimens were polished by emery papers (#100 – #2000) and cloths with alumina suspension of 0.3 μm particles, and then heat-treated at 250, 300 and 350°C for 1 h in Ar atmosphere to obtain various grain sizes (HPT + 250, 300 and 350°C samples). The microstructures of those samples were observed by electron backscatter diffraction (EBSD) analysis with a 7.5 nm step size using orientation imaging microscope (OIM) attached to field-emission scanning electron microscope (FE-SEM) of JSM-7001F.

Table 1. Chemical composition of the investigated 4N-purity iron.

Fe	C	Si	Cu	Ni	Cr	Mn
Bal.	0.002	<0.002	0.0005	0.0004	0.0003	0.0001
Unit: mass%

Fig. 1.

Sampling positions of tensile specimen and TDS specimen from the disk processed by HPT. (Online version in color.)

2.2. Hydrogen Pre-charged Tensile Tests

In this study, hydrogen pre-charged and uncharged tensile tests were conducted for the HPT + 250, 300 and 350°C samples using a cathode charge method. As shown in Fig. 2, miniature tensile test specimens were set on a testing machine and immersed in 0.2N NaOH solution at a constant temperature of ice water. Then, a constant current of 50 A/m² was applied to the HPT + 250°C sample for 0.5, 1, 3, 6, 12 h, to the HPT + 300°C sample for 3, 4.5, 6, 12, 18 h and to the HPT + 350°C sample for 3, 6, 12, 18, 24 h for hydrogen charging. Immediately after the hydrogen charging, tensile tests were started with a strain rate of 8.3×10⁻⁵ s⁻¹ in the NaOH solution, and the nominal stress-crosshead displacement curves were obtained. It should be noted that since crosshead displacement includes not only the elongation of tensile specimens but also elastic deformation of machine parts, reduction of area after fracture F was used as the ductility of specimens through   

$$F=\frac{S_0-S}{S_0},$$(1)

where S₀ and S are cross-sectional areas before tensile test (i.e. 1 mm^w×0.6 mm^t) and after tensile test obtained from SEM observation, respectively.

Fig. 2.

Schematic illustration of miniature tensile test after cathode hydrogen charging. (Online version in color.)

2.3. Observation by SEM and Laser Microscope

The fracture surface of raptured specimens after hydrogen pre-charged and uncharged tensile tests was observed by SEM with an accelerating voltage of 15 kV. Three-dimensional observation of raptured specimens was also conducted by a laser microscope of KEYENCE VK-X250 to investigate the changes in fracture morphology with grain size and pre-charged hydrogen content.

Furthermore, for some specimens, tensile tests were interrupted just before fracture to investigate the formation behavior of internal microvoids and cracks. The necked specimens were polished in the above-mentioned manner, and then examined by SEM observation and EBSD analysis.

2.4. TDS Measurements

TDS measurement was conducted to evaluate hydrogen contents released from pre-charged specimens during heating from −130°C to 300°C with a heating rate of 20°C/min. From the comparison of hydrogen contents between the HPT+250°C sample with 0.5, 1, 3, 6, 12 h charging and the HPT+300, 350°C samples with 3 h charging, the charging time and grain size dependence of pre-charged hydrogen content was investigated. Here, hydrogen charging conditions for TDS measurement were the same as those of the hydrogen pre-charged tensile tests.

3. Results

3.1. Average Grain Size and Hydrogen Content before Pre-charging

Figures 3(a)–3(f) shows the orientation maps and grain boundary maps of the HPT+250, 300 and 350°C samples obtained by EBSD analysis. All the samples consisted of randomly oriented ultrafine grains having high-angle grain boundaries with misorientations over 15°, and the average grain sizes of the HPT+250, 300 and 350°C samples were ~0.50, ~0.82 and ~1.01 μm, respectively. These grain sizes are larger than those of the HPT samples of Fe-0.01 mass%C subjected to heat treatment at 300, 400 and 500°C for 0.5 h (i.e. ~0.167, ~0.239 and ~0.395 μm, respectively³¹⁾). This result indicates that the applied heat treatment at lower temperatures is enough for the investigated 4N-purity iron to cause recovery of dislocations introduced by HPT processing, growth of ultrafine grains and increase in misorientation of grain boundaries. The corresponding Kernel Average Misorientation (KAM) maps obtained by defining misorientations over 5° as grain boundaries revealed that misorientation θ_KAM is larger at grain boundaries but smaller within grains (Figs. 3(g)–3(i)). Since θ_KAM relates to geometrically necessary (GN) dislocations,³³⁾ it was confirmed that dislocation densities within grains are much smaller than those near grain boundaries.

Fig. 3.

(a)(b)(c) Orientation maps, (d)(e)(f) grain boundary maps and (g)(h)(i) KAM maps of (a)(d)(g) HPT+250°C, (b)(e)(h) HPT+300°C and (c)(f)(i) HPT+350°C samples.

Figure 4 shows the results of TDS measurement for the Anneal, HPT and HPT+250°C samples without hydrogen charging. For the HPT sample with a higher dislocation density and smaller grain size, 0.667 mass ppm of hydrogen was detected even without hydrogen charging, suggesting that hydrogen atoms were introduced into the specimens during wet polishing. However, most of hydrogen atoms within the HPT sample were released during heating to 250°C because only 0.00175 mass ppm of hydrogen was detected in the HPT+250°C sample. Therefore, it was confirmed that the heat treatment applied in this study not only changes the grain sizes of the specimens, but also removes hydrogen introduced during wet polishing.

Fig. 4.

TDS profiles of annealed, HPT and HPT+250°C samples without hydrogen charging. Total contents of detected hydrogen are numerically indicated. (Online version in color.)

3.2. Hydrogen Pre-charged Tensile Tests and Hydrogen Embrittlement Index

Figure 5 shows the stress-crosshead displacement curves obtained from uncharged and hydrogen pre-charged tensile tests for the HPT+250, 300 and 350°C samples. Regardless of the applied heat-treatment temperature and charging time, discontinuous yielding with a maximum stress and the following sharp stress drop was observed in all the stress-crosshead displacement curves. When this critical (maximum) stress is taken as yield strength, the HPT+250°C sample with a smaller grain size was found to exhibit the highest yield strength (Fig. 5(a)), suggesting that grain refinement strengthening is achieved. On the other hand, the charging time (t) dependence of yield strength was not clearly detected, but crosshead displacement to fracture was certainly decreased with increasing t. This tendency is also obvious in Fig. 6, where reduction of area F is first decreased with t, and then saturated at 3 h for the HPT+250°C sample and at 18 h for the HPT+350°C sample.

Fig. 5.

Stress-crosshead displacement curves of (a) HPT+250°C, (b) HPT+300°C and (c) HPT+350°C samples after hydrogen charging for different charging times. (Online version in color.)

Fig. 6.

Charging time dependence of reduction of area for (a) HPT+250°C, (b) HPT+300°C and (c) HPT+350°C samples obtained by miniature tensile test. (Online version in color.)

Figure 7 shows the charging time dependence of hydrogen embrittlement index   

$$I=\frac{F_0-F_{\text{H}}}{F_0},$$(2)

where F₀ and F_H are reductions of area for uncharged and hydrogen pre-charged specimens, respectively. Under this definition, the higher I value, the greater decrease in ductility by hydrogen charging, and thus more severe hydrogen embrittlement occurs. It is shown from Fig. 7 that I becomes larger with decreasing grain size, especially in the HPT+250°C sample, where rapid hydrogen embrittlement occurs even with a shorter charging time of hydrogen t.

Fig. 7.

Charging time dependence of hydrogen embrittlement index for (a) HPT+250°C, (b) HPT+300°C and (c) HPT+350°C samples obtained by miniature tensile test. (Online version in color.)

3.3. Observation of Fracture Surface and Transition of Fracture Modes

Figures 8 and 9 show SEM images of the fracture surface of the HPT+250°C sample without charging and with 12 h charging, and of the HPT+350°C sample without charging and with 24 h charging. The cross-sectional areas of the uncharged HPT+250 and 350°C samples were significantly decreased (Figs. 8(a), 8(b)), and fine (< 1 μm) and coarse dimples were observed on the fracture surface of the well-necked tensile specimens (Figs. 9(a), 9(b)). On the other hand, similar ductile fracture was observed in the HPT+350°C sample with 24 h charging (Fig. 9(d)), whereas the cross-sectional area of the HPT+250°C sample with 12 h charging was hardly decreased (Fig. 8(c)) with the fracture surface of fine (< 1 μm) and elongated shallow dimples (The latter dimples were indicated by arrows in Fig. 9(c)). Such a difference of fracture morphology is also confirmed by laser scanning images in Fig. 10, and the fracture modes of the investigated specimens could be classified into two types: void coalescence-type fracture of well-necked tensile specimens and shear-type fracture of diagonally ruptured tensile specimens. Although not shown here, the HPT+250°C sample with 0.5 h charging was classified into void coalescence-type fracture, whereas the HPT+250°C sample with 1 h charging into shear-type fracture. This result indicates that the rapid increase in I for the HPT+250°C sample in Fig. 7 is attributed to the transition of fracture modes from void coalescence-type fracture to shear-type fracture. This transition of fracture modes is also reported for the HPT sample of 2N-purity Fe-0.01%C.³¹⁾

Fig. 8.

SEM images of fracture surface for (a) HPT+250°C sample without charging, (b) HPT+350°C sample without charging, (c) HPT+250°C sample with 12 h charging and (d) HPT+350°C sample with 24 h charging.

Fig. 9.

Magnified SEM images of fracture surface for (a) HPT+250°C sample without charging, (b) HPT+350°C sample without charging, (c) HPT+250°C sample with 12 h charging and (d) HPT+350°C sample with 24 h charging.

Fig. 10.

Laser scanning images color-coded with the difference in height of fracture surface for (a) HPT+250°C sample without charging, (b) HPT+350°C sample without charging, (c) HPT+250°C sample with 12 h charging and (d) HPT+350°C sample with 24 h charging. (Online version in color.)

3.4. Hydrogen Content and Hydrogen Embrittlement Index after Hydrogen Pre-charging

Figure 11 shows the charging time dependence of hydrogen content (C) obtained from TDS measurement for the HPT+250°C sample with 0.5, 1, 3, 6, 12 h charging. C rapidly increased with increasing charging time and then saturated around 0.7 mass ppm of hydrogen. This result suggests that the saturation of hydrogen embrittlement index I around 3 h charging (Fig. 7) is attributed to the saturated hydrogen content around 0.7 mass ppm by hydrogen charging.

Fig. 11.

Charging time dependence of hydrogen content in HPT+250°C sample. (Online version in color.)

Figure 12 shows the grain size dependence of hydrogen content C for the HPT+250, 300 and 350°C samples with the same charging time of 3 h. C monotonously increased with decreasing grain size, suggesting that hydrogen atoms generated during cathode charging are entered from the specimen surface through grain boundaries. It should be noted that the HPT+250°C sample with a smaller grain size contains higher C due to the greater area of grain boundaries.

Fig. 12.

Grain size dependence of hydrogen content in HPT+250°C, HPT+300°C and HPT+350°C samples after hydrogen charging for 3 h. (Online version in color.)

4. Discussion

4.1. Fracture Mechanism of Uncharged Specimens

First, the fracture mechanism of uncharged specimens with smaller hydrogen contents is discussed, followed by that of hydrogen pre-charged specimens. As shown in Figs. 9(a) 9(b), fine (< 1 μm) and coarse dimples were observed in the fracture surface of the uncharged HPT+250 and 350°C samples, suggesting that microvoids formed during tensile tests partially coalesced and grew, resulting in the void coalescence-type fracture. In general, second phase particles of precipitates and inclusions are considered as initiation points of voids formation, because voids are formed by fracture of those particles or by separation of particle/matrix interfaces. However, such second phase particles are seldom dispersed in the investigated 4N-purity iron, and thus voids should be formed and grown in a different mechanism. Kumer et al.³⁴⁾ conducted in-situ tensile tests in transmission electron microscopy for a nanocrystalline nickel and suggested that voids formation is initiated by atomic-scale displacement at triple junctions of grain boundaries or at grain boundaries where intragranular and intergranular slips are intersected. Therefore, local coalescence and growth of microvoids formed by such a mechanism is considered to lead to void coalescence-type fracture of the investigated ultrafine-grained 4N-purity iron with almost no second phase particles (Figs. 10(a) 10(b)). In fact, SEM images just before fracture of the uncharged HPT+250 and 350°C samples contained coalesced and grown internal microvoids as shown in Figs. 13(a) 13(b).

Fig. 13.

SEM images of tensile-tested specimens just before fracture of (a) HPT+250°C sample without charging, (b) HPT+350°C sample without charging, (c) HPT+250°C sample with 12 h charging and (d) HPT+350°C sample with 24 h charging. Magnified images of rectangular area are also shown on the right side. (Online version in color.)

From a report that grain boundary sliding is activated by absorption of dislocations at grain boundaries,³⁵⁾ therefore, the fracture mechanism of uncharged ultrafine-grained iron can be schematically illustrated in Fig. 14(a). First, microvoids are formed at triple junctions of grain boundaries because of grain boundary sliding caused by absorption of dislocations at grain boundaries before discontinuous yielding^36,37) in Fig. 5, and then local coalescence and growth of microvoids follows by absorption of deformation-induced vacancies. In the uncharged specimens, however, triple junctions do not necessarily act as initiation points of voids formation, and thus void coalescence-type fracture occurs with coarser dimple patterns (Figs. 9(a) 9(b)).

Fig. 14.

Schematic illustration of fracture mechanisms of ultrafine-grained iron (a) without hydrogen charging, (b) with hydrogen charging (in case of smaller grains) and (c) with hydrogen charging (in case of larger grains). (Online version in color.)

4.2. Fracture Mechanism of Hydrogen Pre-charged Specimens

Next, the fracture mechanism of the HPT+250°C sample with 12 h charging is discussed to reveal the effect of pre-charged hydrogen on the formation and growth of microvoids in the investigated ultrafine-grained iron. As shown in section 3.2, pre-charged hydrogen of ~0.7 mass ppm resulted in almost no tensile elongation in the stress-crosshead displacement curve (Fig. 5(a)) because shear-type fracture occurred in the diagonally raptured tensile specimen (Figs. 8(c) and 10(c)). SEM images and the corresponding orientation map of the HPT+250°C sample with 12 h charging confirmed that cracks were already propagated along grain boundaries before fracture (Figs. 13(c) and 15(a)) unlike the case of the uncharged specimen where only coalesced and grown internal microvoids were formed (Fig. 13(a)). However, the fracture surface showed no trace of intergranular fracture, but finer dimple patterns (Fig. 8(c)), and furthermore microvoids were formed at triple junctions of grain boundaries ahead of propagated cracks, as indicated by arrows in Fig. 15(a). Therefore, the fracture mechanism of ultrafine-grained iron with hydrogen charging can be schematically illustrated in Fig. 14(b). First, in a similar manner to the uncharged specimen (Fig. 14(a)), microvoids are formed at triple junctions of grain boundaries because of grain boundary sliding, and then further growth of microvoids follows by absorption of deformation-induced vacancies. However, the frequency and degree of the formation and growth of microvoids are much greater in the hydrogen pre-charged specimens because hydrogen atoms introduced by hydrogen charging enhance the formation of deformation-induced vacancies. Therefore, more noticeable crack propagation can be observed due to shorter distances between adjacent microvoids (Fig. 14(b)).

Fig. 15.

SEM images and the corresponding orientation maps around propagated cracks in (a) HPT+250°C sample with 12 h charging and (b) HPT+350°C sample with 24 h charging. Orientation maps are depicted for rectangular area in SEM images.

Such a transition of fracture modes from void coalescence-type fracture to shear-type fracture is also reported in an advanced high-strength steel (AHSS),³⁸⁾ where shear-type fracture occurred after hydrogen charging with the fracture surface of fine and elongated shallow dimples like Fig. 9(c). It should be noted that those fracture surfaces are typical of shear-stress fracture, which can be seen in periphery area of cup-and-cone type fracture.³⁹⁾ However, a difference in crack initiation was observed in the HPT+250°C sample with 12 h charging. While crack propagation started at corners of a rectangular cross section of the AHSS steel (i.e. intergranular fracture), cracks were propagated as if internal microvoids at triple junctions are connected each other. This indicates that not only hydrogen content but also grain size changes fracture modes of ultrafine-grained iron.

According to the HESIV theory,¹³⁾ hydrogen atoms introduced by hydrogen charging enhance the formation and growth of microvoids because the increased concentration of deformation-induced vacancies are more efficiently absorbed by microvoids. This could promote the propagation of cracks, resulting in the transition of fracture modes from void coalescence-type fracture to shear-type fracture.

4.3. Grain Size Dependence of Fracture Mechanism of Hydrogen Pre-charged Specimens

Finally, the fracture mechanism of the HPT+350°C sample with 24 h charging is discussed. Regardless of hydrogen pre-charging for 24 h, void coalescence-type fracture occurred, not shear-type fracture (Figs. 8(d), 9(d) and 10(d)) unlike the case of the HPT+250°C sample with 12 h charging. The corresponding SEM images and orientation map confirmed that not only coalesced and grown microvoids at triple junctions of grain boundaries but also cracks propagated along grain boundaries were observed before fracture (Figs. 13(d) and 15(b)), similarly to HPT+250°C sample with a smaller grain size (Fig. 15(a)). However, the larger grain size of the HPT+350°C sample possesses longer distances between triple junctions, and thus local coalescence and growth of microvoids become predominant, resulting in void coalescence-type fracture with coarser dimple patterns, as schematically illustrated in Fig. 14(c).

A molecular dynamics calculation by Wan et al.⁴⁰⁾ suggested that hydrogen atoms promote voids formation by annihilation of dislocations at grain boundaries in body-centered cubic (bcc) iron. However, if voids formation occurs not only at triple junctions but also at grain boundaries, all the investigated pre-charged specimens should rupture by shear-type fracture, independent of their grain sizes. Therefore, it can be concluded that hydrogen embrittlement of ultrafine-grained iron is dominated not only by pre-charged hydrogen content but also by grain size because the coalescence and growth of microvoids depend on distances between triple junctions of grain boundaries, i.e. grain size, leading to the transition of fracture modes from void coalescence-type fracture to shear-type fracture.

5. Conclusions

In this study, grain sizes of 4N-purity iron were changed by high-pressure torsion (HPT) and subsequent annealing, and then miniature tensile tests were conducted for the ultrafine-grained specimens after hydrogen charging for different charging times. From the observed fracture surface after tensile tests as well as that just before fracture, the transition of fracture modes has been observed depending on grain size and pre-charged hydrogen content of the specimens. The obtained results are summarized below.

(1) Randomly oriented ultrafine grains having high-angle grain boundaries with misorientations over 15° were produced in the HPT+250, 300 and 350°C samples with average grain sizes of ~0.50, ~0.82 and ~1.01 μm, respectively. The cross-sectional areas of the uncharged HPT+250 and 350°C samples were significantly decreased after tensile tests, and fine (< 1 μm) and coarse dimples were observed on the fracture surface of the well-necked tensile specimens. This fracture mode of void coalescence-type fracture is attributed to local coalescence and growth of microvoids formed at triple junctions of grain boundaries.

(2) On the other hand, the cross-sectional area of the HPT+250°C sample with pre-charged hydrogen of ~0.7 mass ppm was hardly decreased with fine (< 1 μm) and elongated shallow dimples on the fracture surface of the diagonally ruptured tensile specimens. This fracture mode of shear-type fracture originates from hydrogen atoms generated and entered from the specimen surface through grain boundaries during cathode charging.

(3) Regardless of grain size and pre-charged hydrogen content, discontinuous yielding with a maximum stress and the following sharp stress drop was observed in the stress-crosshead displacement curves. However, hydrogen embrittlement index evaluated from reduction of area was increased with decreasing grain size due to the transition of fracture modes from void coalescence-type fracture to shear-type fracture. This transition is attributed to the fact that cracks are propagated as if internal microvoids at triple junctions are connected each order.

(4) In the HPT+350°C sample with a larger grain size, however, void coalescence-type fracture occurred, not shear-type fracture, regardless of hydrogen pre-charging for 24 h. Therefore, it can be concluded that hydrogen embrittlement of ultrafine-grained iron is dominated not only by pre-charged hydrogen content but also by grain size because the coalescence and growth of microvoids depend on distances between triple junctions of grain boundaries, i.e. grain size, leading to the transition of fracture modes from void coalescence-type fracture to shear-type fracture.

Acknowledgements

This research was financially supported by the 28th ISIJ Research Promotion Grant and JSPS KAKENHI Grant No. 16K18268. Sample preparation by HPT processing was performed using the facility in Kyushu Institute of Technology. The authors deeply acknowledge their generous supports.

References

• 1)   S.  Matsuoka,  N.  Homma,  H.  Tanaka,  Y.  Fukushima and  Y.  Murakami: J. Jpn. Inst. Met., 70 (2006), 1002 (in Japanese). https://doi.org/10.2320/jinstmet.70.1002
• 2)   C. M.  Sturges and  A. P.  Miodownik: Acta Metall., 17 (1969), 1197. https://doi.org/10.1016/0001-6160(69)90097-2
• 3)   R. A.  Oriani: Acta Metall., 18 (1970), 147. https://doi.org/10.1016/0001-6160(70)90078-7
• 4)   A.  Zielinski,  E.  Lunarska and  M.  Smialowski: Acta Metall., 25 (1977), 551. https://doi.org/10.1016/0001-6160(77)90194-8
• 5)   A. J.  Kumnick and  H. H.  Johnson: Acta Metall., 28 (1980), 33. https://doi.org/10.1016/0001-6160(80)90038-3
• 6)   W. Y.  Choo and  J. Y.  Lee: Metall. Trans. A, 13 (1982), 135. https://doi.org/10.1007/BF02642424
• 7)   S.  Asano and  M.  Shibata: Scr. Metall., 16 (1982), 1171. https://doi.org/10.1016/0036-9748(82)90090-4
• 8)   S. M.  Myers,  D. M.  Follstaedt,  F.  Besenbacher and  J.  Bottiger: J. Appl. Phys., 53 (1982), 8734. https://doi.org/10.1063/1.330473
• 9)   F.  Besenbacher,  S. M.  Myers,  P.  Nordlander and  J. K.  Norskov: J. Appl. Phys., 61 (1987), 1788. https://doi.org/10.1063/1.338020
• 10)   K.  Ono and  M.  Meshii: Acta Metall. Mater., 40 (1992), 1357. https://doi.org/10.1016/0956-7151(92)90436-I
• 11)   H. K.  Birnbaum and  P.  Sofronis: Mater. Sci. Eng. A, 176 (1994), 191. https://doi.org/10.1016/0921-5093(94)90975-X
• 12)   T.  Tabata and  H. K.  Birnbaum: Scr. Metall., 18 (1984), 231. https://doi.org/10.1016/0036-9748(84)90513-1
• 13)   M.  Nagumo: Mater. Sci. Technol., 20 (2004), 940. https://doi.org/10.1179/026708304225019687
• 14)   K.  Sakaki,  T.  Kawase,  M.  Hirato,  M.  Mizuno,  H.  Araki,  Y.  Shirai and  M.  Nagumo: Scr. Mater., 55 (2006), 1031. https://doi.org/10.1016/j.scriptamat.2006.08.030
• 15)   M.  Nagumo and  H.  Matsuda: Philos. Mag. A, 82 (2002), 3415. https://doi.org/10.1080/01418610208240452
• 16)   S.  Wang,  M. L.  Martin,  P.  Sofronis,  S.  Ohnuki,  N.  Hashimoto and  I. M.  Robertson: Acta Mater., 69 (2014), 275. https://doi.org/10.1016/j.actamat.2014.01.060
• 17)   A.  Shibata,  T.  Murata,  H.  Takahashi,  T.  Matsuoka and  N.  Tsuji: Metall. Mater. Trans. A, 46 (2015), 5685. https://doi.org/10.1007/s11661-015-3176-x
• 18)   Y.  Ogawa,  D.  Birenis,  H.  Matsunaga,  O.  Takakuwa,  J.  Yamabe,  Ø.  Prytz and  A.  Thøgersen: Mater. Sci. Eng. A, 733 (2018), 316. https://doi.org/10.1016/j.msea.2018.07.014
• 19)   K. N.  Solanki,  M. A.  Tschopp,  M. A.  Bhatia and  N. R.  Rhodes: Metall. Mater. Trans. A, 44 (2013), 1365. https://doi.org/10.1007/s11661-012-1430-z
• 20)   S.  Wang,  M. L.  Martin,  I. M.  Robertson and  P.  Sofronis: Acta Mater., 107 (2016), 279. https://doi.org/10.1016/j.actamat.2016.01.067
• 21)   H.  Fuchigami,  H.  Minami and  M.  Nagumo: Philos. Mag. Lett., 86 (2006), 21. https://doi.org/10.1080/09500830500482316
• 22)   K.  Takasawa,  R.  Ikeda,  N.  Ishikawa and  R.  Ishigaki: Int. J. Hydrog. Energy, 37 (2012), 2669. https://doi.org/10.1016/j.ijhydene.2011.10.099
• 23)   S.  Chen,  M.  Zhao and  L.  Rong: Mater. Sci. Eng. A, 594 (2014), 98. https://doi.org/10.1016/j.msea.2013.11.062
• 24)   I. J.  Park,  S. M.  Lee,  H. H.  Jeon and  Y. K.  Lee: Corros. Sci., 93 (2015), 63. https://doi.org/10.1016/j.corsci.2015.01.012
• 25)   N.  Zan,  H.  Ding,  X.  Guo,  Z.  Tang and  W.  Bleck: Int. J. Hydrog. Energy, 40 (2015), 10687. https://doi.org/10.1016/j.ijhydene.2015.06.112
• 26)   Y.  Bai,  Y.  Momotani,  M. C.  Chen,  A.  Shibata and  N.  Tsuji: Mater. Sci. Eng. A, 651 (2016), 935. https://doi.org/10.1016/j.msea.2015.11.017
• 27)   C.  Park,  N.  Kang and  S.  Liu: Corros. Sci., 128 (2017), 33. https://doi.org/10.1016/j.corsci.2017.08.032
• 28)   Y.  Mine,  N.  Horita,  Z.  Horita and  K.  Takashima: Int. J. Hydrog. Energy, 42 (2017), 15415. https://doi.org/10.1016/j.ijhydene.2017.04.249
• 29)   Y. H.  Fan,  B.  Zhang,  J. Q.  Wang,  E. H.  Han and  W.  Ke: J. Mater. Sci. Technol., 35 (2019), 2213. https://doi.org/10.1016/j.jmst.2019.03.043
• 30)   A. P.  Zhilyaev and  T. G.  Langdon: Prog. Mater. Sci., 53 (2008), 893. https://doi.org/10.1016/j.pmatsci.2008.03.002
• 31)   Y.  Mine,  S.  Matsumoto and  Z.  Horita: Corros. Sci., 53 (2011), 2969. https://doi.org/10.1016/j.corsci.2011.05.052
• 32)   Y.  Todaka,  K.  Morisako,  M.  Kumagai,  Y.  Matsumoto and  M.  Umemoto: Adv. Mater. Res., 89–91 (2010), 763. https://doi.org/10.4028/www.scientific.net/AMR.89-91.763
• 33)   M.  Calcagnotto,  D.  Ponge,  E.  Demir and  D.  Raabe: Mater. Sci. Eng. A, 527 (2010), 2738. https://doi.org/10.1016/j.msea.2010.01.004
• 34)   K. S.  Kumar,  S.  Suresh,  M. F.  Chisholm,  J. A.  Horton and  P.  Wang: Acta Mater., 51 (2003), 387. https://doi.org/10.1016/S1359-6454(02)00421-4
• 35)   S.  Itoh,  K.  Nakazawa,  T.  Matsunaga,  Y.  Matsukawa,  Y.  Satoh and  H.  Abe: ISIJ Int., 54 (2014), 1729. https://doi.org/10.2355/isijinternational.54.1729
• 36)   S.  Gao,  M.  Chen,  S.  Chen,  N.  Kamikawa,  A.  Shibata and  N.  Tsuji: Mater. Trans., 55 (2014), 73. https://doi.org/10.2320/matertrans.MA201317
• 37)   H.  Li,  S.  Gao,  Y.  Tomota,  S.  Ii,  N.  Tsuji and  T.  Ohmura: Acta Mater., 206 (2021), 116621. https://doi.org/10.1016/j.actamat.2021.116621
• 38)   J.  Venezuela,  Q.  Zhou,  Q.  Liu,  H.  Li,  M.  Zhang,  M. S.  Dargusch and  A.  Atrens: Mater. Today Commun., 17 (2018), 1. https://doi.org/10.1016/j.mtcomm.2018.07.011
• 39)   T.  Matsuo,  J.  Yamabe,  S.  Matsuoka and  Y.  Murakami: Proc. 2008 Int. Hydrogen Conf. - Effects of Hydrogen on Materials (Jackson, WY), ed. by B. Somerday et al., ASM International, Materials Park, OH, (2009), 105.
• 40)   L.  Wan,  W. T.  Geng,  A.  Ishii,  J. P.  Du,  Q.  Mei,  N.  Ishikawa,  H.  Kimizuka and  S.  Ogata: Int. J. Plast., 112 (2019), 206. https://doi.org/10.1016/j.ijplas.2018.08.013