2018 年 58 巻 9 号 p. 1745-1747
Cryogenic X-ray diffraction measurements demonstrated a split of the ε-martensite peak at 193 K in a hydrogen-charged austenitic steel. Only the higher angle peak remained after aging at room temperature. This phenomenon can be interpreted by a change in the interstitial hydrogen position. Particularly, the motion of the leading partial involved in ε-martensitic transformation can move interstitial hydrogen from a tetrahedron to an octahedron site, expanding the lattice. Subsequently, the hydrogen can move back to the tetrahedron site, which relatively shrinks the lattice. The two different hydrogen positions cause the peak to split.
In austenitic steels, martensitic transformations have significant interactions with solute hydrogen in terms of diffusion/desorption kinetics,1,2,3) transformation thermodynamics,4) and martensite morphology,5) including lattice dilatation.6,7) The interactions between martensitic transformations and solute hydrogen affect work hardening,8) fatigue resistance,9) and embrittlement susceptibility.10) Therefore, the nature of the hydrogen/martensitic transformation relation has long been a subject of research.
In particular, ε-martensite characteristics are strongly dependent on hydrogen uptake and desorption behaviors. One of the most distinct effects of hydrogen on ε-martensite is lattice dilatation. Although the hydrogen atom size is small, its lattice dilatation is significant even when compared with carbon. This abnormal lattice dilatation has been investigated through X-ray diffraction (XRD) analysis. However, the origin of the lattice dilatation has not been elucidated yet. Some factors complicating this phenomenon are the presence of hydrogen concentration gradient from a surface,11) occurrence of hydrogen diffusion and desorption at ambient temperature,11) and simultaneous formation of αʹ-martensite.6) In this study, we performed XRD measurements at cryogenic temperatures using a specimen with a saturated hydrogen content and without αʹ-martensite. We then propose an interpretation for the hydrogen effect on XRD profiles in terms of a geometric effect of the atomic sequence.
A Fe-15Mn-10Cr-8Ni alloy (mass%) was prepared by vacuum induction melting. The ingot was hot-forged and rolled at 1273 K. The rolled bar was solution-treated at 1273 K for 3.6 ks and subsequently water-quenched. All the samples were cut by spark machining, and the thickness was reduced to 0.2 mm by mechanical grinding and subsequent electrochemical polishing. The initial constituent phase of this alloy is fully austenite. The starting temperatures for both the ε-martensitic transformation (Ms) and its reverse (As), which were measured by differential scanning calorimetry at a heating rate of 40 K·min−1, are 244 and 336 K, respectively. Specimens were hydrogen-charged at a current density of 50 A·m−2 at room temperature (RT) in a 3% NaCl aqueous solution containing 3 g·L−1 NH4SCN. A platinum wire was used as the counter electrode. Hydrogen content was measured by thermal desorption spectroscopy at a heating rate of 400 K·h−1. Diffusible hydrogen content, which was defined as cumulative hydrogen content from RT to 573 K, is plotted as a function of charging time in Fig. 1. Diffusible hydrogen content was accomplished at 11 days, indicating that the hydrogen distribution is homogeneous in the specimen at the time. Therefore, the hydrogen-charging time for this study and the corresponding diffusible hydrogen content were determined to be 11 days and 70 mass ppm, respectively.
Diffusible hydrogen content plotted against charging time.
Cryogenic XRD measurements were performed using Co Kα at 40 kV and 35 mA with a scanning rate of 0.7° min−1. The cooling rate was 5 K·min−1. The specimens utilized for the XRD measurements were 15 mm wide, 15 mm long, and 0.2 mm thick. The specimens were electrochemically polished and subsequently hydrogen-charged. To remove contamination due to the hydrogen-charging, the specimen surface was slightly polished with a #2000 emery paper before the XRD measurements.
In addition, a temperature-controlled silver decoration experiment was carried out to visualize the microstructural location where hydrogen preferentially exists. Hydrogen was introduced with conditions identical to those mentioned above, except for the charging time. The hydrogen-charging time was shortened to 12 h to reduce contamination of the specimen surface. In order to promote hydrogen diffusion, the temperature during silver decoration was set at 323 K, which is lower than As. The specimen was then mechanically and electrochemically polished followed by cooling down to 77 K. The cooled specimen was hydrogen-charged and subsequently washed with water. Finally, the specimen was immersed in a 4.3 mM K[Ag(CN)2] solution for 24 h. Silver particles were identified by secondary electron (SE) imaging and energy dispersion spectroscopy (EDS) at 5 kV. After slight mechanical polishing, an electron backscatter diffraction measurement was carried out at 15 kV with a beam step size of 50 nm.
Figure 2 shows the XRD profiles of the specimen without hydrogen-charging. The height of the ε-martensite peaks increased with decreasing temperature, and no αʹ-martensite was detected even after cooling to 173 K.
XRD profiles in a specimen without hydrogen-charging at 253, 233, 213, 193, and 173 K.
Figure 3 shows the XRD profiles of the hydrogen-charged specimen. A large peak for ε-martensite is present even at 253 K in the hydrogen-charged specimen owing to hydrogen-induced stress and mechanical polishing. The (1011)ε peak is located at a lower angle and is broader compared to that for the specimen without hydrogen-charging. In addition, the (1011)ε peak is split at 193 K. The height of the higher angle part of the split peak increased when the specimen is cooled to 173 K. After aging at RT for 18 h, the lower angle side disappeared, and only the higher angle side remained. The remaining (1011)ε peak after aging is lower than that without hydrogen-charging as shown in Fig. 3(b). More specifically, hydrogen-charging caused a 0.4% linear expansion of the (001)γ interplanar spacing, and 1.9% and 0.4% linear expansions of the (1011)ε interplanar spacings, corresponding to the low and high angle parts of the split peak, respectively. Since even a linear expansion of the (001)γ interplanar spacing by 1.1 mass% carbon is 0.8%,12) the 1.9% expansion of the (1011)ε interplanar spacing is an abnormal phenomenon.
(a) XRD profiles in a hydrogen-charged specimen at 253, 233, 213, 193, and 173 K. After subsequent aging at RT for 18 h, a diffraction pattern was obtained with the identical specimen. For a comparison, the XRD profile at 173 K without hydrogen-charging is also added here. (b) Enlarged diffraction profiles of the (1011) peak.
Furthermore, the preferential hydrogen sites were identified by a silver decoration experiment, as shown in Fig. 4. The silver-deposited locations indicate the high hydrogen flux regions.13) Figure 4(a) indicates that silver particles appearing as black dots are aligned along specific planes. Figure 4(b) shows that the silver particles lie along surface reliefs. The surface reliefs are identified as ε-martensite by a comparison between Figs. 4(c) and 4(d). Particularly, the silver particles are mainly located on ε-martensite plates having basal planes perpendicular to the surface; when the planes are parallel to the surface, the silver particle density is lower, as shown in Fig. 4(e). These facts indicate that hydrogen preferentially exists along γ/ε interface or in ε-martensite bundles containing numerous thin ε-plates, which is consistent with a previous work reporting that the γ/ε interface traps hydrogen.14)
(a) Optical micrograph showing silver deposition. (b) SE image of the region outlined in (a). The inset indicates an EDS result showing silver distribution. (c) Further magnified SE image. The yellow dotted lines indicates examples of silver particle alignment lying on surface reliefs (d) Phase and (e) inverse pole figure maps.
Based on these results, we propose the following mechanism of the peak split and shift in the XRD profiles. Interstitial hydrogen preferentially enters an octahedron (O) site in the γ matrix. However, this interstitial atom position apparently changes from an O-site to a tetrahedron (T) site when a leading partial sweeps the atom position.15,16) Since ε-martensite grows via a leading partial motion, the interstitial hydrogen position changes to a T-site along the basal plane when the transformation occurs. Furthermore, hydrogen preferentially lies along γ/ε interface, as observed in Fig. 4, implying that it stays at the γ/ε interface or in the ε-bundle after the ε-martensitic transformation. Since the space of T-site is smaller than that of O-site, the transformation-induced O→T site shift expands the ε-martensite lattice. However, since the T-sites are unstable for interstitials, hydrogen atoms would jump to the O-sites spontaneously (this is also the case for carbon, as discussed in a previous work16)). Therefore, we can say that the hydrogen in the T-sites extraordinarily expands the ε-martensite lattice, resulting in the lower angle peak (Situation (1) in Fig. 5). Then, subsequent hydrogen jumps to the O-site to relatively shrink the lattice, thereby producing the higher angle peak. Thus, peak splitting occurs during the transient period of the jump from the T-sites to the O-sites (Situation (2) in Fig. 5). Since hydrogen jumps to the O-sites occur easily at RT, only the higher angle peak remains after RT aging (Situation (3) in Fig. 5). Moreover, hydrogen expands the ε-martensite lattice even at the O-sites; hence, the peak position after aging is lower than that without hydrogen-charging.
A schematic for the correlation between interstitial atom positions and peak configurations.
In this study, we found that ε-martensite containing hydrogen shows two peaks for (1011) in the XRD profiles at cryogenic temperatures. However, the lower angle peak, which indicates extraordinary lattice distortion, disappeared after aging at RT. This phenomenon can be interpreted by a change in the interstitial hydrogen position. Particularly, the motion of the leading partial involved in ε-martensitic transformation can alter the hydrogen position from an O-site to a T-site. This expands the lattice and accordingly, the lower angle peak appears. Subsequently, hydrogen can move back to the O-sites spontaneously, which explains both the appearance of the higher angle peak during cooling and the disappearance of the lower angle peak after RT aging.
The Materials Manufacturing and Engineering Station at the National Institute for Materials Science supported this work through the production of the samples. M. Koyama and K. Tsuzaki acknowledge the helpful discussion with Dr. Sawaguchi. The research project was supported by the Japan Science and Technology Agency (JST) (grant number: 20100113) under Industry-Academia Collaborative R&D Program “Heterogeneous Structure Control: Towards Innovative Development of Metallic Structural Materials” and JSPS KAKENHI (JP16H06365; JP17H04956). Y. Abe thanks Ms Yuriko Hirota in NIMS for her kind assistance of hydrogen-charging and hydrogen content measurement experiments.
