2024 Volume 64 Issue 4 Pages 655-659
To understand the process and mechanism for hydrogen embrittlement in steels, visualization of the location of hydrogen is essential. In the present study, two visualization techniques, hydrogen microprint technique (HMT) and tritium autoradiography (TAR), were applied to a pure iron sheet 20% tensile-deformed with cathodic hydrogen charging. When the specimen was covered with photographic emulsion shortly (40 min) after the deformation, HMT showed that the charged hydrogen atoms diffused out at majorly grain boundaries and minorly in the grain interiors. The TAR, conducted on the same sample but completely de-hydrogenated and then charged with tritium, revealed that hydrogen enhances the formation of vacancies or vacancy clusters with plastic deformation, which are located along grain boundaries and deformation bands and act as relatively stable trapping sites for tritium.
Structural materials essentially need protection against hydrogen embrittlement (HE: degradation of mechanical properties caused by hydrogen atoms introduced into the materials from the environment), in addition to strength and formability. In steels, it is widely known that the susceptibility to HE increases with the strength.1,2,3) If we can elucidate the mechanism for HE with examining the behavior of hydrogen in the material, we will be able to find a strategy for developing materials both with high strength and high resistance to HE. The mechanisms proposed so far can be summarized as follows: (i) hydrogen enhanced decohesion (HEDE) theory where solute hydrogen is assumed to reduce the atomic bonding resulting in the embrittlement;4) (ii) hydrogen enhanced local plasticity (HELP) theory where hydrogen enhances the glide motion of dislocations resulting in the localization of plastic deformation into earlier ductile fracture;5) (iii) hydrogen-enhanced strain-induced vacancy (HESIV) theory where hydrogen accelerates the formation and clustering of vacancies caused by plastic deformation resulting in facilitation of the process of localized ductile fracture.6) In HESIV theory, what directly contributes to the fracture is vacancies or vacancy clusters that form and grow with the aid of hydrogen, not hydrogen atoms themselves. Takai et al.7) reported that a pure iron specimen tensile-deformed with cathodic charging, unloaded and then de-hydrogenated has ductility lower than that tensile-deformed to fracture without hydrogen charging. They also revealed from their thermal desorption analysis (TDA) that the amount of absorbed hydrogen of the specimen hydrogen-charged simultaneously with tensile deformation is larger than that simply hydrogen-charged. They claimed that the increment of the hydrogen amount stemmed from the hydrogen-induced vacancies or vacancy clusters (hydrogen-induced defects: HID). Sakaki et al.8) applied positron annihilation lifetime spectroscopy (PALS) to the pure iron specimen hydrogen-charged simultaneously with 20% tensile deformation, and found that the vacancy concentration was 4 times higher than that without hydrogen charging. By means of TDA and PALS, the acceleration of vacancy formation and clustering during plastic deformation by hydrogen has been demonstrated also in a tempered martensitic steel,9) cold-drawn pearlitic steel10) and metastable austenitic stainless steel.11) However, these technique cannot reveal the correlation between the microscopic distribution of HID and the microstructure. To understand the process and mechanism for HE, visualization of hydrogen in relation to the microstructure is effective.
A variety of visualization methods for hydrogen in metals have been developed so far.12) In the present study, we used two methods where the hydrogen atoms can be observed as silver particles in microscopic scale by covering the specimen surface with photographic emulsion containing silver bromide particles.
The first method is hydrogen microprint technique (HMT)13,14) that enables diffusive hydrogen. In this method, a chemical reaction is used where metallic silver is produced as a result of reduction of silver ions in the bromide particles caused by the hydrogen atoms that diffuse from the interior to the surface of the specimen. The location of the desorbed hydrogen atom can be observed as the metallic silver particles by keeping the specimen for a certain duration after emulsion covering, followed by fixing treatment to remove unreacted silver bromide particles.
The previous HMT studies on steels15,16,17,18) focused mainly on the hydrogen diffusion path and trapping site related to second gamma and inclusion phases without straining, where the detection side was opposite to the hydrogen entry side (opposite-side HMT). Under this configuration, Nagao et al.19) investigated the effect of plastic strain up to 20% in a 0.005%C steel. Ghorani et al. investigated the effect of incubation period at room temperature between hydrogen charging and emulsion covering in a duplex stainless steel under the same configuration20) as well as that where the detection side was the same as the hydrogen entry side (same-side HMT).18) No work seems to have been made on pure iron or ferritic steel under the same-side HMT condition and on the effect of incubation period.
The second method is tritium autoradiography (TAR)21) where photographic effect of β− ray emitted from disintegrating tritium (radioisotope of hydrogen) atom is used. To do this, the specimen is charged with tritium, covered with photographic emulsion, held at a cryogenic temperature for a certain duration to prevent tritium diffusion and to cause tritium disintegration, photographically developed, fixed, and then observed microscopically. By the development, the latent images formed in the silver bromide particles receiving the β−-rays during the holding are converted into metallic silver particles. The role of fixing is the same as in HMT to remove unreacted bromide particles. Since the range of the β−-rays in steel is sufficiently short, an observed metallic silver particle shows that a tritium atom was just underneath the particle, which can visualize the location of trapped (non-diffusive) hydrogen.
All the TAR studies on steels16,21,22,23) were aimed to reveal stable hydrogen trapping sites of hydrogen, because the sample must be left at room temperature at a sufficient duration to eliminate the HMT effect by diffusive hydrogen atoms. The reported hydrogen trapping sites were second phase/matrix boundaries, grain boundaries and dislocation networks (sub-boundaries). No attempt has been made to visualize HID.
In the present study, firstly, we have confirmed desorption behavior of hydrogen at room temperature by examining the effect of incubation period by means of HMT together with TDA in a pure iron deformed in tension under hydrogen charging. Secondly, we have attempted to visualize the microscopic distribution of hydrogen related to HID by means of TAR.
From a 1 mm thick cold-rolled sheet of pure iron with 99.99% purity, test pieces with length and width of the parallel portion of 10 mm and 4 mm, respectively, were cut out, and then annealed in an argon gas flow atmosphere at 800°C for 1 h, followed by grinding using #80, #400 and #800 abrasive papers to remove surface oxide film.
The parallel portion of the tensile test pieces ( the other portions were masked), was cathodic-pre-charged for 1 h at room temperature with a platinum anode in a solution of 3% NaCl plus 3 g·L−1 NH4SCN at a current density of 100 A·m−2, and then tensile-tested with continuing the charging at an initial strain rate of 1.67×10−5 s−1. For comparison, some test pieces were tensile-tested in laboratory air without any charging. After the test, fracture surfaces were observed with a scanning electron microscope (SEM, JSM-6510LA).
To investigate the behavior of formation of HID with tensile deformation, specimens for TDA, HMT and TAR were cut out from the parallel portion of the test pieces tensile-deformed by 20% under the same condition as above and then unloaded.
For TDA, the specimens were ground up to #800, washed ultrasonically in acetone, inserted in a quarts tube, heated under argon gas of 99.999% purity with a flow rate of and 20 mL·min−1 at a heating rate of 100 K·h−1 up to 300°C. Measurement of desorbed hydrogen was made at a constant interval of 2 min, using a gas chromatograph (NISSHA FIS, PDHA-2100) with a semi-conductor gas sensor. To investigate the hydrogen desorption during holding at room temperature, the duration between the unloading and the onset of TDA measurement (incubation period) was changed from 30 min (as quick as possible) and 168 h.
For HMT, the specimens were tensile-deformed in the same way as above, ground up to #2000, electrolytically polished, etched with 3% nital, covered with photographic emulsion for nuclear experiment (Ilford L-4, grain size: 0.13 μm) distilled to double in a darkroom by wire-loop method, kept for a certain duration, fixed in a 15% sodium thiosulfate solution, rinsed in running water, dried naturally, and finally observed using an SEM (JSM-6510LA) equipped with an energy dispersive X-ray spectroscopic (EDS) device. For the dilution of the photographic emulsion and preparation for the fixing solution, 10% solution of sodium nitrite was used. Here, as in TDA, to examine the effect of the incubation period, the durations between the unloading and the emulsion covering were chosen to be 40 min (as quick as possible) and 24 h. Holding time from emulsion covering to fixing were 24 h and 144 h, respectively.
For TAR, the specimens 20% deformed with and without hydrogen charging were held at ambient atmosphere for about 365 d to remove normal hydrogen trapped inside HID thoroughly, and then charged with tritium so that the tritium would be trapped by HID. For tritium charging, the specimens were exposed for 4 h in a quartz tube at room temperature with tritium-deuterium mixture gas (T/(D+T)=0.05, T: number of tritium atoms, D: number of deuterium atoms ) at a total pressure of 1.2 kPa, cooled with liquid nitrogen from outside of the tube with evacuating up to a total pressure of 10−5 Pa, taken out from the tube after vented to atmospheric pressure, covered with the emulsion in the same way as in HMT, and then held in liquid nitrogen for 257 h so that the latent images were formed in the emulsion by the β− ray emitted from disintegrating tritium atoms. The duration from the end of the exposure to the tritium-containing gas to the emulsion covering was about 4 h, during which tritium atoms trapped by sites with smaller binding energy (dislocations and grain boundaries) were supposed to be desorbed to outside of the specimen. After holding for 257 h, the emulsion films together with the specimens were developed and fixed using commercially available developer and fixer (Super prodol and Super Fuji fix), rinsed in running water, and then observed with an SEM (JSM-6701F) equipped with an EDS device.
Figure 1 shows stress-strain curves of the specimens with and without hydrogen charging. The sample with hydrogen charging fractured at a small strain compared to the un-charged sample. Elongation to failure calculated by a change in a gauge length was 28.4 and 44.9% for with and without hydrogen charging, respectively, which indicated degradation of ductility by HE. In the tensile test, as the strain was derived from the crosshead displacement and some plastic deformation in the R portion was observed, the values of elongation to failure estimated from Fig. 1 are somewhat larger than the value mentioned above (28.4 and 44.9%). Figure 2 shows fracture surfaces of the specimens with and without hydrogen charging. Quasi-cleavage and intergranular cracking, which are typical fracture surfaces resulted from HE, are observed in the sample with hydrogen charging. At a higher magnification (Fig. 2(c)), ledges, as an evidence of plasticity, are observed on intergranular fracture surfaces. The un-charged sample is seen to be fractured after significant reduction of area with dimples on the fracture surface. In Fig. 3, TDA spectra of the specimens are shown with different incubation durations at room temperature after 20% tensile deformation with hydrogen charging. In the sample with an incubation period of 30 min after unloading, peak temperature of hydrogen desorption rate is at about 50°C and a total amount of desorbed hydrogen from room temperature to 200°C is 2.61 ppm. As the incubation duration increases to 24 h, hydrogen desorption below 50°C decreases drastically (note that the full scale of the vertical coordinate in Fig. 3(b) is enlarged by twenty times from that in Fig. 3(a)). The peak shifts to higher temperature and the total amount of desorbed hydrogen decreases to 0.18 ppm. In the sample with a longer incubation period of 168 h, hydrogen desorption below 90°C was not detected and the peak temperature is observed at about 120°C. During holding the sample at room temperature, hydrogen trapped at grain boundaries and dislocations, which have relative low binding energies, as well as solute hydrogen in the lattice, are desorbed from these sites, diffuse to the sample surface and then evolve out. Therefore, hydrogen desorption rate in lower temperature range decreases with increasing of incubation period. It is presumed that HID trapping site with higher binding energy,24) trapped hydrogen for the time as long as 168 h.
HMT/SEM images of the specimens tensile-deformed by 20% with hydrogen charging are shown in Fig. 4, with incubation periods of 40 min and 24 h, holding durations of 24 h or 144 h from emulsion covering to fixing, respectively. White particles are observed on grain interior and grain boundaries in the sample that was covered with emulsion with incubation periods of 40 min at room temperature after unloading (Figs. 4(a) and 4(b)). These particles were confirmed as metallic silver particles with EDS analysis. Based on the principle of HMT, the location of these silver particles corresponds to the hydrogen emission location. Hence, hydrogen atoms in the sample are found to diffuse during the holding after the emulsion covering at room temperature both through the grain interior and along grain boundaries, reach the surface, and reduce the silver bromide in the emulsion to form metallic silver particles. In Figs. 4(c) and 4(d), no silver particle can be seen (the particle around the center of Fig. 4(d) was confirmed as a contaminant with EDS analysis), which means most of the hydrogen atoms charged in the specimen have been diffused out during the incubation period from 40 min to 24 h even though the duration at room temperature after the emulsion covering was prolonged to 144 h. As shown in Fig. 3, most of diffusible hydrogen desorbed during the incubation at room temperature for 24 h. Therefore, little hydrogen reaching to the surface resulted in no detection of silver particles.
Regarding the location of silver particles, i.e., the desorption site of hydrogen in Figs. 4(a) and 4(b), it can be seen that both grain boundary and interior act as the desorption site. As mentioned in the introduction section, Nagao et al.19) reported a result of the opposite-side HMT for the specimen hydrogen-charged, emulsion-covered, tensile-deformed up to 20% and then fixed, stating that silver particles were detected both grain boundary and interior, which is similar to the present result. They attributed their result to lattice diffusion, pipe diffusion along the dislocation core and hydrogen transport with gliding dislocations. However, the number density of the particles on the grain boundaries in Figs. 4(a) and 4(b) in the present study is found to be much larger than what can be expected assuming that the particles are randomly distributed, which implies that the role of grain boundary is great in terms both of diffusion and trapping. Further study is needed to draw a conclusion on the desorption mechanism because many experimental conditions are different between their report19) and present study.
TAR/SEM images are shown in Fig. 5. White particles, confirmed to be silver with EDS analysis, are observed on a grain boundary (G.B.) and deformation bands indicated by dashed lines in the tensile-deformed sample with hydrogen charging (Fig. 5(a)). Silver particles were not detected in the tensile-deformed sample without hydrogen charging (Fig. 5(b)). In TAR, since the location of silver particles corresponds to that of the trapped tritium, this result means that certain sites that can trap tritium are only present in the sample deformed with hydrogen charging, not in the sample without hydrogen charging, and that the defects should be HIDs acting as relatively stable trapping sites. This is in agreement with the results obtained by Takai et al.,7) where HIDs are present even after the hydrogen is defused out. A notable feature in the present study is that HIDs are present adjacent to grain boundaries and deformation bands. These locations correspond to the fractography results showing intergranular cracking and quasi-cleavage fracture with traces of local plastic deformation, assuming that vacancy formation occurred at these locations as a result of intersection of dislocations, dragging of jogs, etc. As mentioned in the introduction section, since no attempt has been made to visualize the HIDs, the present result should be the first evidence for the location of HIDs.
With respect to effect of plastic deformation, although we did not examine the specimen without deformation in the present HMT and TAR studies, it is expected that both the amount of charged hydrogen and the amount of hydrogen diffusing out prior to the emulsion covering will be reduced because of a significant decrease in the number of hydrogen trapping sites, resulting in far fewer silver particles for the same incubation period as in the present HMT result shown in Figs. 4(a) and 4(b). This is supported by the result by Nagao et al.19) that no silver particle was observed when the deformation amount was 5% and 0% in 0.005%C steel, and by the TDA study by Takai et al. that the amount of charged hydrogen without deformation was lower than that with deformation.7) Also, in the present TAR condition, the density of the trapping sites for tritium is much smaller in the sample that would have been hydrogen charged without plastic deformation, than that with deformation. Accordingly, the number of silver particles will be smaller than the sample with deformation (zero; Fig. 5(b)).
In the present study, hydrogen distribution was investigated in a pure iron deformed by 20% in tension with cathodic electrolytic hydrogen charging by means of HMT and TAR in relation to the microstructure. The results obtained can be summarized as follows.
(1) When the specimen was covered with photographic emulsion 40 min after unloading and then kept for 24 h in the ambient atmosphere, HMT revealed silver particles both at grain boundaries and inside the grains. However, the density of particles at grain boundaries was larger than that expected assuming the random distribution of the particles. Hence, hydrogen atoms are concluded to be diffused out majorly at grain boundaries and minorly in the grain interiors. When emulsion covering was carried out 24 h after the unloading, no silver particle was observed even the specimen was kept in the ambient atmosphere for 144 h after the emulsion covering, which means that the majority of the hydrogen atoms weakly trapped were desorbed and diffused out within 24 h at room temperature and that remaining small fraction of hydrogen atoms was not detectable by means of HMT.
(2) TAR, carried out on the sample 20% deformed with hydrogen charging, complete de-hydrogenation and then charged with tritium, revealed that silver atoms were observed at grain boundaries and deformation bands, while no silver particle was detected in the sample deformed without hydrogen charging even though the tritium charging was conducted under the same condition as above. This can be reasonably understood by assuming that hydrogen enhances the formation of HIDs with plastic deformation, which act as relatively stable trapping sites for tritium and hence for hydrogen as well.
The present work was supported by the joint research program of the Hydrogen Isotope Research Center, Organization for Promotion of Research, University of Toyama (HRC-2022-10) and by JSPS KAKENHI Grant Number JP20K05128. Also, one of the authors (T.M.) is grateful for financial support by the 29th ISIJ Research Promotion Grant.
