1 Introduction

High-strength steel sheets of 780-MPa- and 980-MPa-grade are used for automobile frame parts to secure crash safety of passengers and improve fuel efficiency because they decrease the body weight of vehicles. Furthermore, 1470-MPa-grade ultra-high-strength steel sheets that are produced via hot stamping technique (Senuma and Takemoto 2010; Zhou et al. 2014) are used for the frame parts of vehicles. Conversely, 780-MPa- and 980-MPa-grade high-strength steel sheets are typically processed by cold stamping. It is widely known that press formabilities deteriorate when tensile strength of conventional high-strength steel sheets exceeds 980 MPa, which has been regarded as a problem in cold stamping. It is also known that hydrogen embrittlement is a serious problem in ultra-high-strength steel sheets with tensile strength over 980 MPa. Thus, hydrogen embrittlement resistance of ultra-high-strength steel sheets have been investigated (Takagi et al. 2012, 2016).

From the formability viewpoint, the use of transformation-induced plasticity (TRIP) (Zackay et al. 1967) is a pathway to solve the problem in cold stamping. For example, TRIP-aided bainitic ferrite (TBF) (Caballero et al. 2008, 2013; Hojo et al. 2008, 2016, 2017, 2018; Peet and Hojo 2016; Song et al. 2003; Sugimoto et al. 2002, 2004; Yoshikawa et al. 2012), for which the matrix is bainitic ferrite, is expected to be used as the cold stamping ultra-high-strength steel with tensile strength of over 980 MPa because the TBF steel exhibits excellent press formabilities (Caballero et al. 2008; Sugimoto et al. 2002, 2004), impact property (Caballero et al. 2013; Hojo et al. 2016), and fatigue property (Song et al. 2003; Yoshikawa et al. 2012) associated with the TRIP of retained austenite. Furthermore, superior hydrogen embrittlement resistance (Hojo et al. 2008, 2017, 2018; Peet and Hojo 2016) is also reported due to the high hydrogen absorption capacity of retained austenite. Therefore, hydrogen embrittlement properties of the TBF steel attracted attention in terms of exploring practically available high-performance high-strength steels.

Understanding pre-strain effect before hydrogen uptake is a key to simulate the press forming and cold stamping for automobile parts. However, there is a paucity of investigations of hydrogen embrittlement property of pre-strained TBF steels. In the study, the effect of pre-straining on hydrogen embrittlement resistance of a TBF steel was investigated towards application as cold stamping ultra-high-strength steel sheets.

2 Experimental procedure

Cold-rolled steel sheet with a chemical composition of 0.4C-0.49Si-1.51Mn-1.02Al-0.05Nb-0.2Mo-0.0015O-0.0019N (mass %) was used in the study. The martensite-transformation-start-temperature (\(M_{\mathrm {S}}\)) (Tamura 1970) was estimated as \(376\, ^\circ \hbox {C}\). The cold-rolled steel sheet was annealed at \(915\, ^\circ \hbox {C}\) for 1200 s followed by austempering treatment at \(425\, ^\circ \hbox {C}\) for 500 s to obtain bainitic ferrite matrix with an approximately 10% interlath and blocky retained austenite as shown in Fig. 1. The carbon content in the retained austenite was estimated as 1.20 mass% from a lattice constant (Dyson and Holmes 1970).

Fig. 1
figure 1

a Inverse pole figure (IPF) of RD direction and b phase maps of the as-heat-treated TBF steel. \(\upalpha _{\mathrm {bf}}\) in (a) and \(\upgamma _{\mathrm {R}}\) in (b) denote bainitic ferrite and retained austenite, respectively. The electron backscatter diffraction measurement is performed at a beam step size of 0.17 nm. Specifically, RD, TD, and ND represent rolling, transverse, and normal directions, respectively. Red and blue regions in (b) represent fcc (retained austenite) and bcc (bainitic ferrite) phases, respectively

Full size image

Variation in the volume fraction of retained austenite during tensile testing was measured via the energy dispersion method using synchrotron white X-rays and Ge semiconductor detector at BL14B1 in SPring-8. The X-rays were shaped with a height of \(50\, \upmu \hbox {m}\) and a width of \(300\, \upmu \hbox {m}\) via slits set at an incident side, and X-rays penetrated through the sample were limited by a collimator of 50–200 \(\upmu \hbox {m}\) with a 500-\(\upmu \hbox {m}\) slit. The diffraction angle of the detector was set at \(10^\circ \). From the energy spectrum obtained by the X-ray diffraction analysis, the peaks corresponding to \(200_{{\upalpha }}\), \(211_{{\upalpha }}\), \(220_{{\upalpha }}\), \(200_{{\upgamma }}\), \(220_{{\upgamma }}\) and \(311_{{\upgamma }}\) were approximated using a Gaussian function, and an integrated intensity was obtained from Gaussian curve fit to those peaks. The volume fraction of retained austenite was estimated by the average of ratio of the integrated intensities of \(\upalpha \) and \(\upgamma \) peaks. A tensile test of the specimen shown in Fig. 2 was performed at a crosshead speed of 0.03 mm/min using a home-made small tensile testing machine.

Additionally, dislocation patterns were observed via transmission electron microscopy (TEM) at an acceleration voltage of 200 kV. The specimens for TEM were prepared by the Ion Slicer (EM-09100IS produced by JEOL). Fracture surface and specimen surface were observed via scanning electron microscopy at an acceleration voltage of 15 kV.

Tensile pre-straining was applied to an as-heat-treated specimen with dimensions of 50 mm in gauge length, 12.5 mm in width, and 1.2 mm in thickness at a crosshead speed of 1 mm/min at \(25\, ^\circ \hbox {C}\). Subsequently, specimens with different pre-strains were charged with hydrogen. Tensile tests were performed at the same tensile test condition as pre-straining. Hydrogen embrittlement property was evaluated by total elongation, \(\textit{TEl}_{\mathrm {f}}\). Total elongation in the study is defined as follows:

$$\begin{aligned} \textit{TEl}_{\mathrm {f }}= \varepsilon _{\mathrm {pre}} + \varepsilon _{\mathrm {m}} \end{aligned}$$
(1)

where \(\varepsilon _{\mathrm {pre}}\) and \(\varepsilon _{\mathrm {m}}\) denote elongations during pre-straining and after hydrogen charging, respectively.

Hydrogen was introduced via a cathode charging method using a 3 wt% NaCl and \(3\hbox { g/L }\hbox {NH}_{\mathrm {4}}\) SCN aqueous solution at current densities corresponding to 1 and \(10\hbox { A/m}^{\mathrm {2}}\) at \(25\, ^\circ \hbox {C}\) for 48 h. A platinum wire was used as an anode. Hydrogen concentration (mass ppm, hereafter ppm) was measured via thermal desorption spectrometry (TDS) using quadrupole mass spectrometry. The TDS measurements were conducted after tensile testing. Specimens were heated from room temperature to \(800\, ^\circ \hbox {C}\) at a heating rate of \(100\, ^\circ \hbox {C}\hbox {/h}\). Diffusible hydrogen was defined as hydrogen desorbed below \(300\, ^\circ \hbox {C}\). Specimens after the tensile tests were placed in liquid nitrogen to prevent hydrogen desorption prior to TDS measurements.

The average size of retained austenite was defined as diameters for blocky retained austenite and length for filmy retained austenite, which were measured from phase map obtained by an electron backscatter diffraction measurement with a beam step size of 170 nm at an acceleration voltage of 15 kV. On the other hand, the dimple size was defined as average length of major and minor axes after elliptic approximation of its shape, which was measured from scanning electron micrographs of fracture surfaces.

3 Results and discussion

3.1 Microstructure and tensile properties

First, we present basic mechanical and microstructural characteristics of the present TBF steel without hydrogen charging. Table 1 lists the tensile properties and retained austenite characteristics of the as-heat-treated TBF steel. Figure 2 shows the transformation behavior of retained austenite during tensile testing. The volume fraction of retained austenite decreased with increases in the plastic strain. Approximately, 4 vol% austenite still remained even after 23% plastic straining that corresponded to uniform elongation. The same experiment was carried out for another specimen, and we confirmed the reproducibility of the changing trend of austenite fraction with strain. Simultaneously, dislocation density increases with increases in the strain as shown in Fig. 3.

Fig. 2
figure 2

Variation in volume fraction of retained austenite (\(f_{{\upgamma }}\)) as a function of plastic strain (\(\varepsilon _{\mathrm {p}}\))

Full size image
Table 1 Tensile properties and retained austenite characteristics of TBF steel
Full size table
Fig. 3
figure 3

TEM micrographs of dark field image of specimens strained to a 0%, b 6%, c 10%, and d 15% without hydrogen charging

Full size image

3.2 Hydrogen embrittlement properties

Figure 4 shows nominal stress–strain curves of the TBF steel after pre-straining to 0, 6, and 12% with hydrogen charging at current densities corresponding to 1 and \(10\hbox { A/m}^{\mathrm {2}}\). The dotted lines denote nominal stress–strain curves without pre-straining and hydrogen charging. The hydrogen charging significantly decreased fracture strain. When the pre-strain was 12%, plastic deformation barely occurred in the hydrogen charged specimen. The diffusible hydrogen contents after fracture as measured by TDS are also shown in Fig. 4.

Fig. 4
figure 4

Nominal stress (\(\sigma _{\mathrm {N}}\))–nominal strain (\(\varepsilon _{\mathrm {N}}\)) curves of TBF steel pre-strained by (a, d) 0, (b, e) 6 and (c, f) 12% with hydrogen charging at current densities corresponding to (a–c) 1 and (d–f) \(10\hbox { A/m}^{\mathrm {2}}\). Hydrogen contents in the hydrogen charged specimens as measured after fracture are shown in the lower right

Full size image

Figure 5 shows total elongation \((\textit{TEl}_{\mathrm {f}})\) as a function of pre-strain (\(\varepsilon _{\mathrm {pre}}\)). Hydrogen charging degraded \(\textit{TEl}_{\mathrm {f}}\), irrespective of pre-strain. It should be noted that the pre-straining to 3–10% reduced the degree of hydrogen-induced degradation of the \(\textit{ TEl}_{\mathrm {f}}\) when the specimens were charged with hydrogen at \(1\hbox { A/m}^{\mathrm {2}}\). The specimens charged with hydrogen at \(\hbox {10 A/m}^{\mathrm {2}}\) also exhibited a similar trend in the pre-strain range of 6–10%. A factor that triggered the pre-strain effect can be a decrease in volume fraction of retained austenite via pre-deformation-induced martensitic transformation as shown in Fig. 2. The martensitic transformation during pre-straining avoided hydrogen supersaturation in the fresh martensite that occurred when hydrogen-charged austenite transformed to martensite (Hojo et al. 2019; Koyama et al. 2019). Additionally, the martensitic transformation during pre-straining decreased the probability of austenite/bainitic ferrite interface that acted as both preferential hydrogen localization site (Chan et al. 1991) and cracking site. Conversely, the advantageous effect of pre-strain on \(\textit{TEl}_{\mathrm {f}}\) disappeared when the pre-strain exceeded 12%. To explain the disappearance of the advantageous pre-strain effect, we here note that the pre-strained specimens showed higher yield strength than that of the specimen without pre-strain, due to work hardening (Fig. 4). Hydrogen-related cracking can be assisted by the work hardening associated with dislocation multiplication, deformation-induced martensitic transformation and pile-up of dislocations at lath, packet, block and prior austenite boundaries. The dislocation multiplication increased weak hydrogen trap site density (Choo and Lee 1982), thereby increasing diffusible hydrogen content after pre-straining and subsequent hydrogen charging as shown in Fig. 4. When the number of dislocations interacting with hydrogen increases, hydrogen accumulation at martensite boundaries or dislocation arrays can be promoted by dislocation motion and subsequent pile-up, which assists boundary cracking or quasi-cleavage fracture (Nagao et al. 2012). These dislocation-driven phenomena potentially increased hydrogen embrittlement susceptibility when the properties were compared at an identical current density.

Fig. 5
figure 5

Variations in total elongation \((\textit{TEl}_{\mathrm {f}})\) as a function of pre-strain \((\varepsilon _{\mathrm {pre}}\)), in which \(\textit{TEl}_{\mathrm {f}}\) is defined as the sum of elongations during pre-straining (\(\varepsilon _{\mathrm {pre}}\)) and after hydrogen charging (\(\varepsilon _{\mathrm {m}}\)), \((\textit{TEl}_{\mathrm {f }}= \varepsilon _{\mathrm {pre}} + \varepsilon _{\mathrm {m}})\)

Full size image
Fig. 6
figure 6

Fracture surface of the specimen without pre-straining and hydrogen charging

Full size image
Fig. 7
figure 7

Fracture surfaces of the specimens pre-strained to (a, d) 0%, (b, e) 6% and (c, f) 12% with hydrogen charging at current densities of (a–c) 1 and (d–f) \(10\hbox { A/m}^{\mathrm {2}}\)

Full size image

3.3 Behavior of hydrogen-assisted fracture

The fracture surface of the TBF steel without hydrogen charging was fully covered with dimples (Fig. 6). Conversely, the fracture surface of the pre-strained and hydrogen charged specimens also exhibited dimples although their sizes exceeded that without hydrogen charging. Furthermore, the dimple size increased due to increases in the current density from 1 to \(10\hbox { A/m}^{\mathrm {2}}\) as shown in Fig. 7. Generally, a dimple size of hydrogenated steels is reported as smaller than that without hydrogen (Marchi et al. 2008) although the present result revealed an opposite trend. To interpret the fracture behavior, microstructural cracking behavior constitutes a key as discussed below.

Fig. 8
figure 8

Specimen surfaces of the gauge section with strain of 6% a without and b with hydrogen charging

Full size image

Figure 8 shows surface of the stretched part of tensile specimens after 6% straining without and with pre-hydrogen-charging at a current density of \(1\hbox { A/m}^{\mathrm {2}}\). Hydrogen charging was conducted before 6% straining. Evidently, the pre-hydrogen-charging increased probability of deformation-induced cracks on the specimen surface. It was reported that hydrogen-assisted cracking occurred at transformed martensite itself and/or matrix/transformed martensite interfaces and that resultant fracture occurred in multi-phase TRIP-aided steels (Laureys et al. 2016; Ronevich et al. 2012; Wang et al. 2015). Thus, it is expected that the hydrogen-charged TBF steel fractured in an early stage of the tensile test given that crack initiation and propagation at matrix/martensite interfaces were accelerated by the existence of transformed martensite in the TBF steel with hydrogen in a manner similar to that in previous studies (Hojo et al. 2018; Ronevich et al. 2012; Zhu et al. 2016). An important aspect of the observation in Fig. 8 corresponds to the presence of “multiple” cracks, which indicated that crack propagation once stopped (otherwise, the first crack initiation causes fracture without second crack initiation). This indicated that the crack coalescence process can act as a significant factor that causes the final failure of the hydrogen-charged TBF steel. In this context, crack initiation probability was also a crucial factor that affected crack growth via coalescence. Specifically, when a subcrack or potential crack initiation site (i.e., retained austenite in this case) exists in front of the main crack, the crack growth can be accelerated. Thus, the pre-strain effect that reduced fraction of retained austenite directly contributed to resisting hydrogen-related crack growth because the martensite transformed from austenite acted as the preferential crack initiation site in the presence of hydrogen as previously mentioned. Additionally, it should be noted that the coarse dimple size shown in Fig. 7 was approximately \(14 \,\upmu \hbox {m}\), which was significantly larger than the average dimple size of the uncharged specimen (\(3\, \upmu \hbox {m}\)). Also, the coarse dimple size was three times larger than the size of retained austenite before the test (shown in Fig. 1). These facts implied that the coarse dimples arose from the coalescence of a few cracks that formed from martensite transformed from retained austenite. This also supported the importance of pre-strain effect, which reduced crack initiation probability on resistance to the hydrogen-assisted fracture.

4 Conclusions

In the study, the tensile properties of pre-strained and hydrogen-charged TBF steel were investigated to evaluate the hydrogen embrittlement resistance of press-formed TBF steel. The main results are shown as follows.

  1. (1)

    The total elongation at fracture of the TBF steel decreased by hydrogen charging when the pre-strain corresponded to 0–15%. The hydrogen-induced degradation of elongation appeared to be significant with increases in the hydrogen content.

  2. (2)

    With respect to the total elongation at fracture, the 3–10% pre-straining improved the hydrogen embrittlement resistance when compared to that without pre-straining. However, the advantageous effect of pre-strain disappeared when the pre-strain was over 12%.

  3. (3)

    The advantageous effect of pre-strain (3–10%) on hydrogen embrittlement resistance was attributed to the transformation of retained austenite prior to hydrogen charging, which suppressed crack initiation and propagation.

  4. (4)

    The disappearance of the advantageous pre-strain effect at 12–15% pre-strain is potentially caused by the increase in the absorbed hydrogen due to pre-straining.