Hydrogen Embrittlement and Local Characterization at Crack Initiation Associated with Phase Transformation of High-strength Steel Containing Retained Austenite

Abstract

States of hydrogen present in high-strength steels for use as bearing steel SUJ2 and hydrogen embrittlement susceptibility were examined using thermal desorption analysis (TDA) and tensile tests. SUJ2 specimens containing the retained austenite phase (γ_R) in the martensitic phase exhibited three hydrogen desorption peaks in the TDA profile. Two of the peaks desorbed at higher temperatures decreased with a decreasing amount of γ_R, indicating they corresponded to desorption associated with γ_R. Fracture strength in the presence of hydrogen increased with a decreasing amount of γ_R and with an increasing strain rate. For the specimens containing γ_R and hydrogen, a flat facet at the crack initiation site and a quasi-cleavage (QC) fracture in the initial crack propagation area were observed on the fracture surface. Local characterization using electron back-scattered diffraction (EBSD) revealed that the flat facet on the fracture surface corresponded not to γ_R but to strain-induced martensite. In addition, the facet was on the {112} plane of martensite, which is the slip plane or deformation twin plane of body-centered-cubic metals. The reason for high hydrogen embrittlement susceptibility of the specimens containing γ_R was attributed to the strain-induced phase transformation at the crack initiation site of the flat facet and in the initial crack propagation area of the QC fracture. Furthermore, the strain rate dependency of hydrogen embrittlement susceptibility is presumably ascribable to local plastic deformation, i.e., the interaction between dislocations and hydrogen.

1. Introduction

Improvements in vehicle mileage in recent years resulting from weight reductions and the introduction of fuel cell vehicles without any CO₂ emissions are expected to help resolve environmental issues. Understanding the fundamental phenomena involved in hydrogen embrittlement is important for the usage of higher-strength steels for the component materials of these vehicles under various conditions that are not assumed conventionally since these materials are required to have higher strength and to be used under severer environments such as in the presence of corrosion or high pressure hydrogen.

High carbon chromium bearing steel (SUJ2) has been widely used as a bearing material because SUJ2 has high tensile strength above 2000 MPa and high abrasion resistance. Assuming that SUJ2 can be used under high pressure hydrogen gas, its range of usage can be expected to expand. However, the mechanical properties of SUJ2 were reported to deteriorate due to rolling contact fatigue induced by hydrogen.^1,2) A white etching area (WEA) was observed on the fracture surface due to rolling contact fatigue, which was presumably associated with hydrogen embrittlement. In experiments to reproduce WEA in rolling contact fatigue tests, it was reported that WEA formation was associated with phase transformation of the retained austenite phase (γ_R), with the result that a number of voids were formed on the interfaces between martensite and γ_R.³⁾

A number of previous studies have dealt with hydrogen embrittlement of steels containing γ_R such as transformation-induced plasticity (TRIP) steel. For example, delayed fracture susceptibility increased with an increasing amount of γ_R in a high-strength steel sheet. Hence, reducing the amount of γ_R by annealing at higher temperatures was effective in preventing delayed fracture.⁴⁾ In contrast, it was reported that the presence of γ_R reduced hydrogen embrittlement susceptibility in tensile tests and constant load four-point bending tests because γ_R trapped a large amount of hydrogen, thereby suppressing the occurrence of fracture on prior austenitic grain boundaries.⁵⁾ In addition, one of the reasons why Al-added TRIP steel displays superior resistance to delayed fracture is because it suppresses hydrogen-induced and stress-induced martensitic transformation.⁶⁾

Conflicting results have been reported as to whether the presence of γ_R promotes or suppresses hydrogen embrittlement, thus its effects have not been made clear yet. Additionally, a few direct observations of microstructural fracture associated with phase transformation and crack initiation of quasi-cleavage around the microstructural fracture were performed using local characterization. Therefore, clarifying the role of the presence of γ_R and hydrogen on crack initiation and propagation will provide valuable information for the development of materials with high resistance to hydrogen embrittlement.

The purpose of the present study was to establish a fundamental understanding of hydrogen embrittlement in SUJ2 with high strength above 2000 MPa. The change in the states of hydrogen associated with phase transformation of γ_R and the effects of temperature, strain rate and the amount of γ_R on hydrogen embrittlement susceptibility were investigated. Furthermore, the role of the phase transformation of γ_R and the effect of interaction between dislocations and hydrogen on crack initiation and propagation were also investigated on the basis of local characterization near the crack initiation site.

2. Experimental Procedure

2.1. Material

High carbon chromium bearing steel SUJ2 quenched at 850°C, tempered at 160°C and sub-zero treated at −60°C was used in the present study. The chemical composition is shown in Table 1. Specimens with two different geometries were prepared; one was 8 mm in diameter and 20 mm in length for hydrogen analysis and the other was 3.4 mm in diameter and 15 mm in G.L. for tensile testing. The tensile strength of the specimens was 2119 MPa. Figure 1(a) shows the microstructure of SUJ2 observed with a scanning electron microscope (SEM). Figure 1(b) shows the phase map obtained by electron back-scattered diffraction (EBSD); the red areas represent martensite, the green areas austenite and the yellow areas carbide. Spherical cementites of approximately 1 μm in diameter were observed in the martensitic phase. In addition, γ_R was also observed in spite of conducting subzero treatment at −60°C.

Table 1. Chemical composition of SUJ2 specimen (mass%).

C	Si	Mn	P	S	Cr
0.99	0.22	0.26	0.008	0.006	1.41

Fig. 1.

(a) SEM image and (b) phase map obtained by EBSD of a SUJ2 specimen: red represents α’, yellow represents carbide and green represents retained γ. (Online version in color.)

In order to vary the amount of γ_R in the specimens, three types of specimens were prepared. The first type denoted here as “SUJ2” was heat treated under the above-mentioned conditions. The second type denoted as “SUJ2+σ” was prepared by prestressing the SUJ2 specimen at 1200 MPa in liquid nitrogen at −196°C to transform γ_R into strain-induced martensite. The third type denoted as “SUJ2+300°C” was annealed at 300°C for 30 min to decompose γ_R into ferrite and cementite. Figure 2 shows diffraction patterns of the three types of specimens obtained by X-ray diffraction (XRD). The XRD conditions were a Cu tube, a tube voltage of 40 kV, a tube electric current of 30 mA, a diffraction rate of 2°∙min⁻¹ and a diffraction angle (2θ) between 30 and 90°. The amount of γ_R was calculated from the integrated intensity of the diffraction line of the martensitic phase (211) and austenite phase (220). The volume of γ_R in the SUJ2, SUJ2+σ and SUJ2+300°C specimens was 13.4, 7.3 and 0 vol%, respectively.

Fig. 2.

Diffraction patterns of (a) SUJ2, (b) SUJ2+σ and (c) SUJ2+300°C specimens analyzed by XRD.

2.2. Hydrogen Charging and Hydrogen Analysis

The specimens were charged electrochemically with hydrogen at a current density of 10 A·m^–2 in a 0.1 N NaOH aqueous solution containing a NH₄SCN additive of 1 g·L⁻¹ kept at a temperature of 30°C. The charging time was determined as 48 h so as to reach an equilibrium hydrogen concentration at both the surface and center of the specimens. Immediately after hydrogen charging, the amount of hydrogen and states of hydrogen were analyzed by thermal desorption analysis (TDA) using a gas chromatograph at a heating rate of 400°C·h^–1 in the temperature range from room temperature to 800°C. A standard gas mixture of Ar + 50 vol. ppm of H₂ was used for calibration of the amount of hydrogen.

2.3. Hydrogen Embrittlement Susceptibility

Hydrogen embrittlement susceptibility of SUJ2 was evaluated on the basis of the fracture strength obtained in tensile tests. Hydrogen was precharged under the above-mentioned conditions. The amount of hydrogen in the specimens was approximately 6 mass ppm. Tensile tests were conducted at strain rates of 2×10⁻⁶, 2×10⁻⁴ and 2×10⁻² s⁻¹ under various temperatures of −196, −60, −30, and 30°C. Hydrogen charging was conducted concurrently with the tensile test at 30°C under the same charging conditions after precharging to keep the hydrogen content of the specimens constant.

Hydrogen embrittlement susceptibility of the SUJ2+σ and SUJ2+300°C specimens containing different amounts of γ_R was also evaluated. Tensile tests of the three types of specimens precharged under identical conditions were conducted at strain rates of 2×10⁻⁶, 2×10⁻⁴ and 2×10⁻² s⁻¹ at a temperature of −30°C.

2.4. Local Characterization at Crack Initiation Site Using EBSD

Crystallographic orientation and phase transformation at and near the crack initiation site were investigated in detail using EBSD for the SUJ2 specimen. The SUJ2 specimen was precharged concurrently with the tensile test at a strain rate of 2×10⁻⁶ at 30°C, followed by microstructural characterization at and near the crack initiation site on the fracture surface cut by a focused ion beam (FIB) using EBSD.

3. Results and Discussion

3.1. Change in Hydrogen States Associated with Phase Transform of Retained Austenite

TDA profiles of the SUJ2, SUJ2+σ and SUJ2+300°C specimens charged with hydrogen under identical conditions are shown in Fig. 3. Hydrogen desorption from the SUJ2 specimen consisted of a large peak at 190°C, a shoulder at 300°C and a small peak at 500°C. Hydrogen desorption from the SUJ2+σ specimen also consisted of two peaks and a shoulder at the same temperatures as the SUJ2 specimen, although the peak heights were different from those of the latter specimen. Hydrogen desorption from the SUJ2+300°C specimen consisted of a large peak at 190°C. In previous studies, it was reported that two peaks appeared in TRIP steel⁷⁾ and that two peaks and one shoulder appeared in SAE 52100 steel.⁸⁾ The ratio of the amount of hydrogen in the three specimens was SUJ:SUJ2+σ:SUJ2+300°C = 1:1:1.4.

Fig. 3.

Thermal desorption profiles of SUJ2, SUJ2+σ and SUJ2+300°C specimens hydrogen-charged at 10 A·m⁻² in 0.1N NaOH solution containing 1 g·L⁻¹ NH₄SCN.

The hydrogen of each peak and shoulder probably desorbed at not the true desorption temperatures but rather at a higher temperature because hydrogen desorption from the specimens during TDA measurement was not detrap-controlled but diffusion-controlled under the experimental conditions such as these specimen diameters and heating rate. However, the microstructure of specimens will change gradually during heating using TDS because of the thermal instability of γ_R in SUJ2 assuming that the heating rate is reduced. On the basis of these reasons, separation of the SUJ2 peaks was attempted under a diffusion-controlled condition, though the peaks and shoulders appeared at higher temperatures.

Figure 4 shows the fitting curves obtained with a Gaussian function to separate the experimental profiles shown in Fig. 3. Curve fitting made it possible to separate three peaks at 190°C (peak 1 hydrogen), 300°C (peak 2 hydrogen) and 525°C (peak 3 hydrogen) for the SUJ2 specimen as shown in Fig. 4(a). On the basis of these three peak temperatures, the fitting curves of the SUJ2+σ and SUJ2+300°C specimens are also shown in Figs. 4(b) and 4(c), respectively. The experimental profile of the SUJ2+σ specimen was fitted by the Gaussian function with the same three peak temperatures as the SUJ2 specimen. For the SUJ2+300°C specimen, the experimental profile was fitted with both peak 1 and peak 2 hydrogen, though it was seen as a single peak at first.

Fig. 4.

Comparison of thermal desorption profiles between experiment and curve fitting using a Gaussian function of (a) SUJ2, (b) SUJ2+σ and (c) SUJ2+300°C specimens.

The presence of the peak 2 hydrogen that appeared as the shoulder in the TDA profiles was confirmed experimentally. Figure 5 shows TDA profiles of the SUJ2 specimen immediately after hydrogen charging (0 d) and aging at a temperature of 30°C for 10 d to degas the hydrogen. Although the amount of hydrogen was reduced by hydrogen degassing for 10 d, a small peak at 325°C remained. This peak temperature corresponded to the temperature of peak 2 hydrogen obtained by fitting the Gaussian function. Hence, the presence of the peak 2 hydrogen obtained by curve fitting was confirmed experimentally.

Fig. 5.

Thermal desorption profiles of SUJ2 specimens immediately after hydrogen charging and following degassing for 10 d at 30°C.

It was reported that peak 1 hydrogen desorbed from medium carbon tempered martensitic steels at weak trapping sites such as vacancies, elastic stress fields around dislocations, dislocation cores, grain boundaries and interfaces of cementite without strain.⁹⁾ Presumably, the peak 1 hydrogen desorbed from the high carbon martensitic steel SUJ2 in the present study also corresponded to diffusible hydrogen trapped at the above-mentioned sites.

Peak 2 and peak 3 hydrogen, which were not seen for medium carbon martensitic steels, were investigated. The SUJ2 specimen contained γ_R of 13.4 vol%. It was reported that γ_R was not easily transformed even though subzero treatment was conducted because γ_R on lath boundaries was relatively stable.¹⁰⁾ In addition, it was reported that block-like γ_R was observed between martensitic laths by TEM.³⁾ This suggests that γ_R also remained in the SUJ2 specimen, although subzero treatment was conducted at −60°C. The heights of peak 2 and peak 3 hydrogen desorbed from the SUJ2+σ specimen decreased to 7.3 vol% γ_R by a strain-induced martensitic transformation at the temperature of liquid nitrogen decreased compared with that of the SUJ2 specimen. The height of peak 2 hydrogen desorbed from the SUJ2+300°C specimen containing 0 vol% γ_R as a result of annealing at 300°C decreased in comparison with that of the SUJ2 and SUJ2+σ specimens, especially peak 3 hydrogen disappeared. Peak 2 and peak 3 hydrogen decreased with a decreasing amount of γ_R. This result suggests that peak 2 and peak 3 hydrogen were related to desorption associated with the presence of γ_R.

The states of peak 2 and peak 3 hydrogen associated with the presence of γ_R were then investigated. Possible reasons for desorption at higher temperatures are as follows.

(i) Diffusion in γ_R

(ii) Decomposition and disappearance of γ_R above 280°C during the heating process

(iii) Strong trapping on the interfaces between γ_R/martensite

First, as the possibility of (i), the apparent diffusion coefficient of hydrogen (D_H) in γ_R of a face centered cubic (fcc) lattice is much smaller than that in martensite of a body centered cubic (bcc) lattice. The D_Hs of two types of steels having the same tensile strength of 1200 MPa, tempered martensite steel and steel containing 26.3 vol% γ_R, were 1.5×10⁻¹⁰ m²·s⁻¹ and 1.3×10⁻¹¹ m²·s⁻¹, respectively.¹¹⁾ Hydrogen desorption was presumably diffusion-controlled in SUJ2 because of the decrease in D_H due to the solution hydrogen in γ_R and the passing across γ_R. Therefore, hydrogen desorbed at higher temperatures in the TDA profile.

Next, the possibility of (ii) is considered. γ_R disappears because it decomposes into ferrite and cementite in the temperature range from 200 to 300°C. Some of the hydrogen present and diffusing in γ_R probably desorbed at a higher temperature at which γ_R decomposed and disappeared when it was unstable and thermally transformed to diffusible ferrite quickly during heating of the TDA measurement.

Finally, the possibility of (iii) is discussed. It was reported that the γ_R/α’ interfaces trapped hydrogen and cracked when γ_R having large solution hydrogen transformed to the α’_stress phase having small solution hydrogen as the hydrogen embrittlement mechanism of SUS304 and steels containing γ_R.^12,13,14,15) Summarizing the changes in the interfaces during testing, γ_R/α’ interfaces present in the SUJ2 specimen decreased and α’_stress/α’ interfaces appeared in the SUJ2+σ specimen. Additionally, γ_R/α’ interfaces disappeared and α/α’ interfaces appeared in the SUJ2+300°C specimen. Since γ_R (fcc)/α’ (bcc) interfaces have different crystallographic structures, hydrogen was probably trapped more strongly than that at the other interfaces, so it desorbed at higher temperatures as peak 2 or peak 3 hydrogen as shown in Fig. 4.

It is widely recognized that the hydrogen desorption temperature measured by TDA rises as the binding energy between trap sites and hydrogen in steels increases. However, peak 2 hydrogen remained regardless of the lower desorption temperature compared with peak 3 hydrogen; in contrast, peak 3 hydrogen with a higher desorption temperature disappeared in Fig. 5. This result implies that peak 2 hydrogen possibly corresponded to desorption from interfaces between γ_R and martensite, i.e., reason (iii) above. Although peak 3 hydrogen was not trapped strongly, it might be ascribable to a time lag due to the low diffusion coefficient of hydrogen in γ_R, i.e., reason (i), or to desorption due to the decomposition and disappearance of γ_R, i.e., reason (ii).

As shown in Fig. 4(c), the amount of peak 1 hydrogen increased for the SUJ2+300°C specimen in spite of annealing at 300°C. In addition, the experimental curve was fitted with not only peak 1 hydrogen but also peak 2 hydrogen, regardless of the 0 vol% of γ_R. The reason can be explained without any contradiction on the basis of the change in the microstructure. The amount of peak 1 hydrogen presumably increased by annealing at 300°C because of the effect of the increase in trapping sites, i.e., cementite precipitation due to decomposition of γ_R into cementite and α¹⁰⁾ was greater than that of the decrease in dislocation density. Therefore, D_H decreased and diffusion-controlled desorption became the dominant factor during the thermal desorption process. This indicates that higher temperature desorption is not attributed to γ_R but to diffusion-controlled desorption of peak 1 hydrogen.

3.2. Hydrogen Embrittlement Susceptibility of SUJ2

Figure 6 shows the stress-displacement curves of SUJ2 specimens at a strain rate of 2×10⁻⁶ s⁻¹ under various temperatures of −196, −60, −30 and 30°C. Fracture strengths of hydrogen-charged SUJ2 specimens decreased in comparison with non-hydrogen-charged SUJ2 specimens under the various temperatures and depended on the tensile test temperature. The relationship between fracture strength and temperature at strain rates of 2×10⁻⁶, 2×10⁻⁴ and 2×10⁻² s⁻¹ is shown in Fig. 7. Although fracture strengths of non-hydrogen-charged specimens were independent of the temperature, fracture strengths of hydrogen-charged specimens were the lowest at −30°C under the various strain rates, and then increased with decreasing temperature.

Fig. 6.

Stress-displacement curves of SUJ2 specimens with/without hydrogen in tensile tests at a strain rate of 2×10⁻⁶ s⁻¹ under various temperatures of (a) 30°C, (b) −30°C, (c) −60°C and (d) −196°C.

Fig. 7.

Effect of tensile test temperatures on fracture strength of SUJ2 specimens with/without hydrogen at strain rates of 2×10⁻⁶, 2×10⁻⁴ and 2×10⁻² s⁻¹.

In early studies, the mechanism of the temperature dependence of hydrogen embrittlement susceptibility was explained based on the interaction between dislocations and hydrogen. The relationship between an average dislocation velocity and D_H in metals affects the degree of the interaction since D_H depends on the temperature. Thus, hydrogen embrittlement susceptibility depends on the test temperature.^16,17) It was reported that hydrogen embrittlement susceptibility of 0.2%C and 0.7%Mn steel having a bcc lattice was maximum at approximately 0°C.¹⁸⁾ The susceptibility of austenitic stainless steel and Al-8%Mg alloy having an fcc lattice was also maximum at −53°C¹⁹⁾ and −13 to 17°C,²⁰⁾ respectively. These results indicate that the temperature dependence of hydrogen embrittlement susceptibility caused by the interaction between dislocation slip and hydrogen diffusion is a common phenomenon for bcc and fcc lattices. However, these results were obtained by fracture in the plastic deformation region. In contrast, since the SUJ2 specimens used in the present study fractured in the elastic deformation region, the mechanism of the SUJ2 specimens must be considered from other viewpoints.

Figure 8 shows microscopic fracture surfaces near the crack initiation site and schematic diagrams of the distribution of fracture modes for hydrogen-charged and non-hydrogen-charged SUJ2 specimens. The results of tensile tests at 30°C are also shown as representative results because the fracture surfaces and modes at 30°C were similar to those at −60 and −30°C. Crack initiation of the non-hydrogen-charged SUJ2 specimen was intergranular (IG) fracture originating from inclusions or grain boundaries and surrounded by microvoid coalescence (MVC) as shown in Fig. 8(a). Crack initiation of the hydrogen-charged SUJ2 specimen was a flat facet which was surrounded by quasi-cleavage (QC) fracture, IG and MVC as shown in Fig. 8(b). Since secondary-phase particles were not observed and a similar flat facet was observed when the opposite fracture surface was examined, this flat facet was caused by microstructural fracture, not by secondary-phase particle fracture. Additionally, crack initiation of hydrogen- charged SUJ2 specimens was caused not only by flat facets but also by inclusions. As for the fracture surfaces at −196°C, crack initiation of the non-hydrogen-charged SUJ2 specimen was due to cleavage fracture, not IG. Crack initiation of the hydrogen-charged SUJ2 specimen was due to a flat facet or inclusion. However, the QC was surrounded by MVC, not IG.

Fig. 8.

Fracture surfaces of SUJ2 specimens after a tensile test at a strain rate of 2×10⁻⁶ s⁻¹ at 30°C. (a) SUJ2 specimen without hydrogen shows intergranular (IG) fracture. (b) SUJ2 specimen with hydrogen shows quasi-cleavage (QC) fracture. (c) and (d) show schematic diagrams of distribution of fracture modes for SUJ2 specimens with/without hydrogen.

These results indicated that crack initiation and fracture modes changed owing to hydrogen charging. In early studies, it was reported that crack initiation changed from QC to IG with an increasing amount of hydrogen in tempered martensitic steel.²¹⁾ In contrast, crack initiation of the SUJ2 specimens without hydrogen or with a small amount of hydrogen was due to IG as shown in Fig. 8(a). Crack initiation changed to a flat facet surrounded by QC and IG with an increasing amount of hydrogen. It was also reported that ASTM A490 steel for high-strength bolts changed into similar fracture modes because stable initial crack propagation in the grains was caused by film-like γ_R on lath boundaries.²²⁾ These observations indicate that hydrogen embrittlement susceptibility of the SUJ2 specimens involved γ_R at crack initiation. In the next section, the effect of γ_R on crack initiation and hydrogen embrittlement susceptibility of SUJ2 will be discussed.

3.3. Relationship between the Amount of γ_R and Hydrogen Embrittlement Susceptibility

The relationship between the amount of γ_R and hydrogen embrittlement susceptibility was investigated by comparing the fracture strengths of the SUJ2 specimen with 13.4 vol% γ_R, the SUJ2+σ specimen with 7.3 vol% γ_R and the SUJ2+300°C specimen with 0 vol% γ_R, which were charged with hydrogen under the same conditions as in section 2.2. The stress-displacement curves of the three types of specimens at a strain rate of 2×10⁻⁶ at −30°C are shown in Fig. 9. Facture strengths of the hydrogen-charged specimens decreased compared with the non-hydrogen-charged specimens in the order of SUJ2+300°C > SUJ2+σ > SUJ2. Since the three types of specimens had different strengths, hydrogen embrittlement susceptibility was evaluated as the ratio of the fracture strengths of the hydrogen-charged and non-hydrogen-charged specimens. As a result, the order was the same in both evaluations. Although it was reported that hydrogen embrittlement susceptibility increased with increasing tensile strength for tensile strengths above 1200 MPa, the fracture strength and its ratio of the hydrogen-charged SUJ2+300°C specimen were the highest regardless of the maximum strength of the specimens. Furthermore, although the SUJ2+300°C specimen absorbed 1.4 times more hydrogen than the SUJ2 specimen as shown in Fig. 4(c), its fracture strength was the highest. Figure 10 shows the fracture strengths of the three types of specimens at strain rates of 2×10⁻⁶, 2×10⁻⁴ and 2×10⁻² s⁻¹ at −30°C. The order of the fracture strength of the hydrogen-charged specimens was SUJ2+300°C > SUJ2+σ > SUJ2 under these strain rates. These results suggest that the difference in hydrogen embrittlement susceptibility of these three types of specimens needs to be considered in terms of some factor other than tensile strength, hydrogen content and strain rate.

Fig. 9.

Stress-displacement curves of (a) SUJ2, (b) SUJ2+σ and (c) SUJ2+300°C specimens with/without hydrogen at a strain rate of 2×10⁻⁶ s⁻¹ at a tensile test temperature of −30°C.

Fig. 10.

Relationship between fracture strength and strain rate of SUJ2, SUJ2+σ and SUJ2+300°C specimens with hydrogen at a tensile test temperature of −30°C.

Figure 11 summarizes the relationship between the fracture strength of the hydrogen-charged specimens and the initial amount of γ_R prior to tensile testing. The fracture strengths of the hydrogen-charged specimens decreased with an increasing initial amount of γ_R under the various strain rates. The results reveal that hydrogen embrittlement susceptibility of SUJ2 is associated with its initial γ_R content.

Fig. 11.

Relationship between fracture strength and the amount of retained austenite in SUJ2, SUJ2+σ and SUJ2+300°C specimens charged with hydrogen at strain rates of 2×10⁻⁶, 2×10⁻⁴ and 2×10⁻² s⁻¹ at a tensile test temperature of −30°C.

Microscopic fracture surfaces near the crack initiation sites of the three types of hydrogen-charged specimens containing different amounts of γ_R are shown in Fig. 12. Crack initiation sites of the SUJ2 and SUJ2+σ specimens containing γ_R were flat facets due to microstructural fracture, which was surrounded by QC as shown in Figs. 12(a) and 12(b). Some previous studies showed that crack initiation sites of hydrogen-charged SUJ2 were non-metallic inclusions.²³⁾ In the present study, not only flat facets but also inclusions were observed at crack initiation sites. On the other hand, crack initiation sites of hydrogen-charged SUJ2+300°C specimens containing no γ_R were not flat facets surrounded by QC but IG fracture that occurred at grain boundaries (some fractures occurred at inclusions) as shown in Fig. 12(c). Crack initiation sites and propagation modes of the three types of specimens containing different amounts of γ_R are summarized in Table 2.

Fig. 12.

Crack initiation and propagation areas of (a) SUJ2, (b) SUJ2+σ and (c) SUJ2+300°C specimens with hydrogen after a tensile test at a strain rate of 2×10⁻⁶ s⁻¹ at −30°C. (a) SUJ2 specimen shows quasi-cleavage (QC) fracture around a flat facet. (b) SUJ2+σ specimen also shows QC and intergranular (IG) fracture around a flat facet. (c) SUJ2+300°C specimen shows IG fracture.

Table 2. Crack initiation and propagation modes of SUJ2, SUJ2+σ and SUJ2+300°C specimens with/without hydrogen at tensile test temperatures of −60, −30 and 30°C.

		SUJ2: 13.4 vol% γ	SUJ+σ: 7.3 vol% γ	SUJ2+300°C: 0 vol% γ
without H	crack initiation	G.B. or inclusion	G.B. or inclusion	G.B. or inclusion
	propagation	IG → MVC	IG → MVC	IG → MVC
with H	crack initiation	flat facet or inclusion	flat facet or inclusion	G.B. or inclusion
	propagation	QC → IG → MVC	QC → IG → MVC	IG → MVC

Since fracture strengths decreased and crack initiation sites of SUJ2 specimens containing γ_R changed to flat facets surrounded by QC due to hydrogen charging, crack initiation sites and their vicinity were analyzed and the results are presented in the next section.

3.4. Local Characterization at Crack Initiation Sites of SUJ2

The fracture process in hydrogen embrittlement of SUJ2 was investigated by dividing it into three stages.

(a) Crack initiation (flat facet)

(b) Initial crack propagation (QC)

(c) Unstable crack propagation (IG and MVC)

In stage (a) of crack initiation, the reason why flat facets appeared at crack initiation sites only in the case of hydrogen-charged specimens containing γ_R was considered. Figure 13(a) shows the microscopic fracture surface near the crack initiation site of a hydrogen-charged SUJ2 specimen that fractured at a strain rate of 2×10⁻⁶ s⁻¹ at 30°C. Figure 13(b) shows inverse pole figure (IPF) maps at flat facets analyzed by EBSD. The flat facet of the crack initiation site was in the martensitic phase of bcc and the crystallographic plane of the facet on the fracture surface was in the {112} plane in the IPF maps.

Fig. 13.

Crack initiation and propagation areas of SUJ2 specimen with hydrogen after a tensile test at a strain rate of 2×10⁻⁶ s⁻¹ at 30°C. (a) Image of crack initiation on the fracture surface observed by SEM. (b) IPF map of the flat facet at the crack initiation site on the fracture surface analyzed by EBSD. The flat facet is on the {112} plane of martensite which is the slip plane in the bcc lattice. (c) IPF map of a cross section near the flat facet at the crack initiation site analyzed by EBSD. The flat facet is on the {112} plane of martensite, which is the slip plane in the bcc lattice. (d) Phase map of a cross section near the flat facet at the crack initiation site analyzed by EBSD. The phase of the flat facet and surrounding area is not retained austenite but martensite. (Online version in color.)

Figures 13(c) and 13(d) show an IPF map and a phase map analyzed by EBSD in a cross section near the flat facet of the crack initiation site cut out using FIB. The regions surrounded by squares are the cross-sectional surface of the same flat facet. One result seen in Fig. 13(c) is that the flat facet was a specific crystallographic plane of {112}. The red and green areas in Fig. 13(d) represent martensite and austenite, respectively. The presence of γ_R was hardly detected in the range of several microns just beneath fracture surface. The SUJ2 specimens originally contained 13.4 vol% γ_R, which was distributed uniformly as shown in Fig. 1(b). The amount of γ_R apart from the fracture surface decreased to 5.3 vol%, but remained for the hydrogen-charged SUJ2 specimens fractured in the tensile test at −30°C. However, these local characterization results revealed that almost all γ_R was transformed into martensite in the range of several microns just beneath the fracture surface as shown in Fig. 13(d).

Since the flat facet was not {001}, which is the cleavage plane, but {112}, which is the slip plane of the bcc lattice, crack initiation at this site was probably caused by plastic deformation due to local dislocation slip and not by a decrease in cohesive energy. Although the SUJ2 specimens fractured under elastic stress macroscopically, the γ_R phase presumably deformed more than the martensite phase and was transformed into strain-induced martensite because of the large difference in strength between γ_R and martensite. Therefore, the {112} plane of the bcc slip plane appeared at crack initiation sites. Alternatively, there is the possibility that fracture occurred on the twin plane in martensite because the internal twin in lens martensite of high-carbon steels generally forms along the {112} plane. For hydrogen-charged Fe–Mn–C steels, it was reported that crack initiation occurred in the deformation twin plane.²⁴⁾

Next, stage (b) of initial crack propagation around the crack initiation site was investigated. For the SUJ2 and SUJ2+σ specimens containing γ_R, fracture occurred at crack initiation sites (grain boundaries or inclusions) and then propagated as IG in the case of the non-hydrogen-charged specimens, while fracture occurred at crack initiation sites (grain boundaries or inclusions) and then propagated as QC and then IG only in the case of hydrogen charging. For the SUJ2+300°C specimens containing no γ_R, fracture occurred at crack initiation sites (grain boundaries or inclusions) and then propagated as IG with/without hydrogen charging. These results reveal that QC, which is one type of transgranular fracture, appeared only when stress was applied to the specimens containing both γ_R and hydrogen. There were cases where several flat facets were observed on a QC fracture surface.

Although QC fracture is one type of brittle fracture without necking macroscopically, TEM observation^25,26) revealed that crack initiation sites were not {001} of the cleavage fracture plane observed in low temperature brittle fracture but {011} of the slip plane along with lath boundaries. The electron diffraction pattern obtained using TEM clarified experimentally the involvement of plastic deformation just beneath the QC fracture surface.²⁵⁾ EBSD analysis also indicated that the QC fracture surface corresponded to the {011} plane.²⁷⁾ Furthermore, when the QC region was cut with FIB and analyzed by TEM, a high-density slip band was observed. These observations prove that QC fracture is associated with plastic deformation.^28,29) Assuming that QC fracture is a trace associated with dislocation slip, the strain rate dependency of hydrogen embrittlement susceptibility probably results from dislocation slip in the initial crack propagation area.

In stage (c) of unstable crack propagation, every fracture surface of the SUJ2, SUJ2+σ and SUJ2+300°C specimens was the MVC fracture mode with/without hydrogen charging. As a result of XRD analysis of a 2×2 mm area of the MVC fracture surface, γ_R was not detected for the SUJ2 specimen. Hence, brittle fracture such as flat facets did not appear, although strain-induced martensitic transformation occurred in the MVC area in the unstable crack propagation stage. Figure 14 shows a schematic diagram of crack initiation and propagation during the fracture process of hydrogen-charged and non-hydrogen-charged SUJ2 and SUJ2+300°C specimens on the basis of the results in Figs. 12 and 13.

Fig. 14.

Schematic diagram of crack initiation and propagation during the fracture process of SUJ2 and SUJ2+300°C specimens with/without hydrogen at tensile test temperatures of −60, −30 and 30°C. Crack initiations site of SUJ2 specimen without hydrogen and that of SUJ2+300°C specimen with/without hydrogen are grain boundaries or inclusions and initial crack propagation of these specimens is IG. In contrast, the crack initiation site of the SUJ2 specimen with hydrogen is a flat facet or inclusion and initial crack propagation of the specimen is QC.

3.5. Relationship between Phase Transformation and Hydrogen Embrittlement of SUJ2

Since the flat facet was plate-like with specific crystallographic orientation perpendicular to the tensile direction as shown in Figs. 13(c) and 13(d), the largest γ_R perpendicular to the tensile direction among various sizes and directions will preferentially cause crack initiation owing to strain-induced martensitic transformation. Although the present study did not examine whether flat facets or inclusions were the preferred crack initiation site, it was presumably determined by the size and distribution. It was observed that the volume expanded by approximately 2% during transformation from γ_R into martensite, although the expansion ratio depended on the carbon content of γ_R.⁴⁾ It was previously reported that void formation and dislocation pile-up on the interfaces were caused by local expansion.³⁾ The following possibilities were considered as the role of hydrogen because specimens without hydrogen formed neither flat facets nor QC fracture and did not result in a decrease in fracture strength, either.

① Hydrogen promotes embrittlement due to an increase in the M_d temperature, i.e., maximum temperature for starting strain-induced transformation, and enhances the transformation.

② Hydrogen promotes embrittlement due not to phase transformation but to an increase in hydrogen trapped at voids and dislocation cells on the interfaces formed by strain-induced transformation.

③ Hydrogen promotes embrittlement due to its supersaturation in martensite instead of hydrogen in γ_R when strain-induced martensitic transformation occurs in the presence of hydrogen.^13,14,15)

④ Lattice defects formed during strain-induced martensite transformation in the presence of hydrogen promote hydrogen embrittlement.³⁰⁾

⑤ Strain-induced lattice defects enhanced by hydrogen associated with local plastic deformation promote hydrogen embrittlement.^31,32,33)

The strain rate dependency of fracture strength as shown in Fig. 10 was considered on the basis of roles ①–⑤. Fracture strengths increased with a decreasing amount of γ_R under various strain rates. This difference in hydrogen embrittlement susceptibility is probably ascribable to the phase transformation of γ_R in the crack initiation stage, resulting from roles ①–③. However, fracture strengths increased with an increasing strain rate; in addition, flat facets in crack initiation and QC fracture in the initial crack propagation stage appeared at a greater strain rate of 2×10⁻² s⁻¹ for the SUJ2 and SUJ2+σ specimens containing γ_R. This change in fracture strength owing to the strain rate was probably caused not by phase transformation but by lattice defect formation associated with the interaction between dislocations and hydrogen in local areas as mentioned in ④ and ⑤.

Based on these considerations, hydrogen embrittlement susceptibility is determined by combined factors; firstly, the effects of strain-induced martensitic transformation on crack initiation as flat facets, and secondly, the effect of the interaction between dislocations and hydrogen on initial crack propagation as QC fracture.

4. Conclusions

States of hydrogen present in high carbon chromium bearing steel SUJ2 and the fundamental evaluation of hydrogen embrittlement susceptibility were examined in relation to the temperature, strain rate and the amount of retained austenite. In addition, the local characterization at crack initiation sites and initial crack propagation were also investigated. The results obtained in the present study can be summarized as follows.

(1) The TDA profile of SUJ2 containing γ_R was divided into three states of hydrogen (peak 1, peak 2, and peak 3 hydrogen). Since the amount of peak 2 and peak 3 hydrogen decreased with a decreasing initial amount of γ_R, peak 2 and peak 3 hydrogen were associated with desorption from γ_R. In contrast, peak 1 hydrogen was associated with desorption from various trapping sites such as dislocations and interfaces between cementite and martensite in tempered martensite phase.

(2) The fracture strengths of hydrogen-charged SUJ2 specimens depended on the strain rate, the amount of γ_R and test temperature. Hydrogen embrittlement susceptibility increased with a decreasing strain rate and an increasing amount of γ_R. Additionally, it showed a maximum value at −30°C.

(3) Flat facets at crack initiation sites and quasi-cleavage fracture at initial crack propagation were observed on the fracture surface of hydrogen-charged SUJ2 specimens containing γ_R. Almost all γ_R was transformed into martensite in local areas of several microns just beneath the flat facets and quasi cleavage fracture. Flat facets consisted of the specific crystallographic plane of {112}.

(4) The presence of γ_R in SUJ2 caused crack initiation (flat facet) and increased hydrogen embrittlement susceptibility due to strain-induced transformation into martensite in the presence of hydrogen. Hydrogen embrittlement susceptibility also increased with a decreasing strain rate, although strain-induced martensitic transformation also occurred under various strain rates. The hydrogen embrittlement susceptibility of the specimens containing γ_R was attributed to the strain-induced phase transformation at the crack initiation site of the flat facet and in the initial crack propagation area of the QC fracture. Furthermore, the strain rate dependency of hydrogen embrittlement susceptibility was presumably ascribable to local plastic deformation, i.e., the interaction between dislocations and hydrogen.

References

• 1)   H.  Uyama,  H.  Yamada,  H.  Hidaka and  N.  Mitamura: Tribol. Online, 6 (2011), 123.
• 2)   B. A.  Szost and  P. E. J.  Rivera-Diaz-del-Castilo: Scr. Mater., 68 (2013), 467.
• 3)   H.  Harada,  T.  Mikami,  A.  Yamamoto and  H.  Tsubakino: Tetsu-to-Hagané, 91 (2005), 567.
• 4)   K.  Yamazaki and  Y.  Mizuyama: Tetsu-to-Hagané, 83 (1997), 754.
• 5)   T.  Hojo,  K.  Sugimoto,  Y.  Mukai,  H.  Akamizu and  S.  Ikeda: Tetsu-to-Hagané, 92 (2006), 83.
• 6)   T.  Hojo,  K.  Sugimoto,  Y.  Mukai and  S.  Ikeda: Tetsu-to-Hagané, 93 (2007), 234.
• 7)   D. P.  Escobar,  T.  Depover,  L.  Duprez,  K.  Verbeken and  M.  Verhaege: Acta Mater., 60 (2012), 2593.
• 8)   J.  Yamabe,  T.  Matsumoto,  S.  Matsuoka and  Y.  Murakami: Int. J. Fract., 177 (2012), 141.
• 9)   K.  Takai: Zairyo-to-Kankyo, 60 (2011), 230.
• 10)   H.  Suto: Tekkouzairyou, ed. by The Japan Institute of Metals, Maruzen, Tokyo, (1985), 39.
• 11)   H.  Tsubakino and  H.  Harada: Tetsu-to-Hagané, 83 (1997), 587.
• 12)   S. L. I.  Chan,  H. L.  Lee and  J. R.  Yang: Metall. Trans. A, 22A (1991), 2579.
• 13)   T.  Omura,  K.  Kobayashi,  M.  Miyahara and  T.  Kudo: Zairyo-to-Kankyo, 55 (2006), 139.
• 14)   S.  Ohmiya and  H.  Fujii: ISIJ Int., 52 (2012), 247.
• 15)   L.  Zhang,  Z.  Li,  J.  Zheng,  Y.  Zhao,  P.  Xu,  X.  Liu,  C.  Zhou and  X.  Li: Int. J. Hydrog. Energy, 38 (2013), 11181.
• 16)   H.  Shoda,  H.  Suzuki,  K.  Takai and  Y.  Hagihara: ISIJ Int., 50 (2010), 115.
• 17)   T.  Doshida and  K.  Takai: Acta Mater., 79 (2014), 93.
• 18)   N.  Ohtani: Tetsu-to-Hagané, 60 (1974), 304.
• 19)   G.  Han,  J.  He,  S.  Fukuyama and  K.  Yokogawa: Acta Mater., 46 (1998), 4559.
• 20)   T.  Ohnishi,  K.  Higashi,  N.  Inoue and  Y.  Nakatani: J. Jpn. Inst. Met., 45 (1981), 972.
• 21)   M.  Wang,  E.  Akiyama and  K.  Tsuzaki: Scr. Mater., 52 (2005), 403.
• 22)   M.  Oyanagi,  T.  Suzuki and  K.  Yamada: J. Soc. Mater. Sci. Jpn., 51 (2002), 561.
• 23)   S.  Fujita,  S.  Matsuoka and  Y.  Murakami: Tetsu-to-Hagané, 95 (2009), 870.
• 24)   M.  Koyama,  E.  Akiyama,  K.  Tsuzaki and  D.  Raabe: Acta Mater., 61 (2013), 4607.
• 25)   Y. H.  Kim and  J. W.  Morris, Jr.: Metall. Trans. A, 14A (1983), 1883.
• 26)   M.  Gao and  R. P.  Wei: Acta Metall., 32 (1984), 2215.
• 27)   A.  Shibata,  H.  Takahashi and  N.  Tsuji: ISIJ Int., 52 (2012), 208.
• 28)   A.  Nagao,  C. D.  Smith,  M.  Dadfarnia,  P.  Sofronis and  I. M.  Robertson: Acta Mater., 60 (2012), 5182.
• 29)   M. L.  Martin,  J. A.  Fenske,  G. S.  Liu,  P.  Sofronis and  I. M.  Robertson: Acta Mater., 59 (2011), 1601.
• 30)   K.  Yokoyama,  T.  Ogawa,  K.  Takashima,  K.  Asaoka and  J.  Sakai: Mater. Sci. Eng. A, 466 (2007), 106.
• 31)   K.  Takai,  H.  Shoda,  H.  Suzuki and  M.  Nagumo: Acta Mater., 56 (2008), 5158.
• 32)   T.  Doshida,  M.  Nakamura,  H.  Saito,  T.  Sawada and  K.  Takai: Acta Mater., 61 (2013), 7755.
• 33)   M.  Hatano,  M.  Fujinami,  K.  Arai,  H.  Fujii and  M.  Nagumo: Acta Mater., 67 (2014), 342.