A review of the influence of hydrogen on the mechanical properties of DP, TRIP, and TWIP advanced high-strength steels for auto construction

2.1 Importance

AHSS were created for the automotive industry due to escalating concerns of fuel consumption and emissions of greenhouse gases and the increasing focus on safety (Bhattacharya, 2005; Bracke, Verbeken, Kestens, & Penning, 2009; WorldAutoSteel, 2014; Zhu, Ma, & Wang, 2007).

A range of strength, toughness, and ductility values has been achieved by (i) selecting appropriate AHSS compositions and microstructures, (ii) controlling the heat treatments, and (iii) applying various strengthening methods. The strength of AHSS is often between 500 and 1500 MPa.

The principal difference between AHSS and conventional steels is their microstructure. Conventional steels are typically ferritic. In contrast, AHSS have a microstructure containing a phase other than ferrite, such as martensite, bainite, austenite, or retained austenite. In addition, typical AHSS have a combination of strength and ductility better than those of conventional steels, as illustrated in Figure 1. AHSS include DP, complex-phase (CP), ferritic-bainitic (FB), martensitic (MS), TRIP, hot-formed (HF), and TWIP steels. TWIP steels are different from other steels in having an austenite matrix that ensures quite different HE properties because of the much lower hydrogen diffusivities.

Figure 1:

Comparison of ductility and strength of AHSS and conventional steels used for automobile construction.

IF, Interstitial free steel; BH, bake hardenable steel; CMn, carbon manganese steel; HSLA, high-strength low-alloy steel; FB, FB steel; 1st GEN AHSS, first-generation AHSS; DP, DP steel; TRIP, TRIP steel; CP, CP steel; MS, MS steel; 2nd GEN AHSS, second-generation AHSS; TWIP, TWIP steel; 3rd GEN AHSS, third-generation AHSS; Q&P, Q&P steel.

The ultra-light steel autobody (ULSAB) advanced vehicle concept indicates that using ~85% AHSS for an automotive body would achieve ~25% weight reduction compared to an average base model and without any increase of the manufacturing cost. AHSS have been widely used by the autoindustry since their first production (AutoSteel, 2014) because of their high strength/weight ratio and good energy absorption.

2.2 DP steels

DP steel is a first-generation AHSS. DP steel has a ferrite and martensite microstructure. DP steel development for the auto industry began in the 1980s (Galán, Samek, Verleysen, Verbeken, & Houbaert, 2012), and the demand for DP steel continues to be strong. DP steels are widely used due to their good combination of strength and ductility and their relatively ease of production.

2.2.1 Microstructure

DP steels have a two-phase microstructure. A hard MS second phase, in the form of islands, is located along the grain boundaries of the soft ferrite matrix. DP steels are commonly produced by cold rolling followed by continuous annealing and quenching (Granbom, 2010; Oliver, Jones, & Fourlaris, 2007). The annealing is in the intercritical region, which is the two-phase field of austenite and ferrite. On quenching, the austenite transforms to martensite. A higher annealing temperature produces a higher volume fraction of austenite, which transforms to martensite during quenching, and a higher strength, because the strength of DP steels is usually proportional to the volume fraction of martensite (Khan, Kuntz, Biro, & Zhou, 2008).

Figure 2 shows a typical microstructure of DP steel. The darker areas are the ferrite matrix. The embossed martensite islands appear light-colored and are located along the grain boundaries. There is less martensite than ferrite.

Figure 2:

SEM image showing a typical microstructure of a DP steel.

The darker, smooth areas are ferrite (F). The light-colored, embossed islands are martensite (M).

2.3 TRIP steels

TRIP steel is a first-generation AHSS. TRIP steel has a multiphase microstructure of ferrite, retained austenite, bainite, and possibly martensite. TRIP steels have attracted much attention since their first development in the 1990s for the automobile industry (Galán et al., 2012) because of a good combination of strength and formability.

2.3.1 Microstructure

TRIP steel has a microstructure consisting of typically three phases: a ferrite matrix, islands of retained austenite, and dispersed bainite regions. Martensite is also commonly present (Oliver et al., 2007).

There are two processing routes for TRIP steel: hot rolling and cold rolling.

Hot rolling starts from the fully austenitic region. After finishing rolling, the sheet goes through a run-out table and controlled cooling. During cooling, ferrite forms, the carbon concentration in the remained austenite increases due to the carbon solubility lower in ferrite than in austenite, and pearlite formation is avoided. During the following coiling, the sheet is coiled in the bainitic region and the austenite transforms to bainite. The transformation from austenite to bainite further enriches small islands of austenite with carbon, making this untransformed austenite stable at room temperature.

In contrast, cold-rolled TRIP steel is transformed from an initial microstructure consisting of ferrite and austenite in two stages: intercritical annealing and isothermal bainitic transformation. The steel is (i) heated into the ferrite and austenite intercritical region to form a two-phase microstructure of ferrite and austenite and (ii) rapidly cooled to avoid formation of intercritical ferrite or pearlite and quenched to a lower temperature, the isothermal bainitic transformation region, where there occurs the partial transformation of austenite to bainite, and the enrichment of the remaining austenite with carbon, resulting in the remaining austenite being stable at room temperature. The higher carbon, silicon, or aluminum content of TRIP steel plays a significant role of the retention of austenite and austenite stabilization (WorldAutoSteel, 2014).

Hot rolling can be used to produce steel sheets with a minimum gauge of 3 mm. In comparison, cold rolling can produce thinner gauge steel sheets. In addition, cold rolling produces steel with smaller grain size because ferrite recrystallization occurs during annealing, which decreases the grain size (Huang, Poole, & Militzer, 2004).

Figure 3 presents an optical micrograph of a typical TRIP steel etched using 2% nital. The ferrite appears yellow and is marked with “F”. The bainite appears brown. The retained austenite appears as the small white blocks dispersed in the ferrite and the bainite.

Figure 3:

Optical micrograph showing a typical microstructure of a TRIP steel.

F, ferrite; B, bainite; RA, retained austenite.

2.4 TWIP steels

TWIP steel is a second-generation AHSS. TWIP steel exhibits a combination of high strength and good formability and thus has attracted a lot of attention in recent decades in the autoindustry.

2.4.1 Microstructure

TWIP steel has a microstructure consisting entirely of austenite at room temperature, caused by the high manganese content (typically more than 15%), as shown in Figure 4. TWIP steel also typically contains alloying elements such as a relatively high content of C, Si, and Al. C assists in the stabilization of austenite at room temperature and strengthens the austenite by solid solution strengthening. Si also increases the strength by solid solution strengthening (Chen, Zhao, & Qin, 2013). The high Al content increases the stacking fault energy (SFE) of the austenite and stabilizes the austenite against transformation to ε-martensite. Al also helps strengthen the austenite by solid solution strengthening (Wan, Chen, & Hsu, 2001).

Figure 4:

SEM image showing the typical microstructure of a TWIP steel.

TWIP steels can be produced by the same manufacturing methods as the other grades of steels and are compatible with processes, such as rolling, pressing, and continuous casting. However, technical difficulties can occur in the production of TWIP steel, such as (i) the high manganese partial pressure during melting and casting, (ii) cracking during hot rolling resulting from the tendency to form strong oxide scales, and (iii) the requirement of stronger rolling equipment due to their high strength during rolling (Kliber, Kursa, Drozd, Hajduchová, & Pešlová, 2012).

3.1 Introduction

To understand the influence of hydrogen on the mechanical properties of AHSS for auto construction, it is necessary to understand (i) hydrogen evolution at the steel surface, (ii) hydrogen adsorption, absorption, and transportation through the steel bulk, and (iii) the mechanisms of embrittlement by hydrogen. This includes (i) how hydrogen is formed on the steel surface, (ii) how much hydrogen enters into the steels, (iii) how far the hydrogen can diffuse in a certain period, and (iv) how the hydrogen interacts with the defects of the steels.

3.2 Hydrogen evolution and entry

Hydrogen can be introduced into steels by two main methods: (i) exposure to hydrogen gas and (ii) exposure to electrolytic hydrogen produced either by cathodic polarization or by a corrosion reaction.

In an hydrogen gaseous environment, the interaction of hydrogen with the steel surface involves the steps of physisorption, chemisorption, and adsorption (Liu & Atrens, 2013). Physisorption is the result of the van der Waals force between a surface and an absorbent. Physisorption is reversible and generally occurs quickly. Physisorption has a low adsorption energy, and it is relatively easy to reach equilibrium at room temperature. For chemisorption, a chemical reaction occurs between the surface atoms and absorbent molecules. Chemisorption dominates the hydrogen uptake at higher temperatures. Chemisorption is limited to a monolayer because short-range chemical forces are involved. Chemisorption is usually slow and either slowly reversible or irreversible. The physisorbed hydrogen usually transfers into chemisorbed hydrogen before the hydrogen enters into the metal. The final step of the gas-solid interaction is adsorption, which involves the incorporation of the chemisorption products into the metal lattice. The amount of hydrogen entering into the steel is controlled by the hydrogen pressure and temperature (Perng & Wu, 2003).

Another source of hydrogen is cathodic polarization or corrosion. The hydrogen evolution reaction (HER) can proceed in acid, neutral, or alkaline conditions through the following steps (Lasia & Grégoire, 1995):

where M represents the metal surface and MH_ads represents the adsorbed hydrogen on the metal surface. Reactions (1) and (2) form atomic hydrogen adsorbed on the surface. The adsorbed hydrogen can be desorbed by the electrochemical [Heyrovsky reaction, (3) or (4)] or chemical [Tafel reaction, (5)] desorption reaction to form molecular hydrogen and leave the steel surface. Alternatively, part of the adsorbed hydrogen enters into the metal through the hydrogen absorption reaction [reaction (6)].

Two models have been proposed (Bockris, McBreen, & Nanis, 1965; Crolet & Bonis, 2001; Flitt & Bockris, 1981) for the hydrogen entrance into the metal. Model A considers that the adsorbed atomic hydrogen enters into the metal substrate. The hydrogen absorption reaction and hydrogen recombination processes are competing. The passage of atomic hydrogen into metal depends on the surface coverage (θ) and the number of available sites for the hydrogen to occupy. The absorption of hydrogen leads to the accumulation to a concentration (C₀) under the surface. As the surface coverage (θ) is determined by the charging conditions (surface coverage increases with decreasingly negative applied potential and current) and temperature, the subsurface concentration of hydrogen (C₀) depends on the charging conditions (applied potential and current), temperature, and state of the surface, such as presence of poisons or oxides, and the adsorption of species other than hydrogen on the surface.

In the other model (model B), a discharged hydrogen proton proceeds directly through the interface into the metal substrate without passing through the intermediate adsorbed state; thus, the hydrogen absorption reaction is competing with the hydrogen evolution processes for the same discharged protons. The transfer of a discharged hydrogen proton H⁺ could occur from water as well as by deprotonation of other complexes adsorbed on the surface, such as that H₂S becomes HS^-.

3.3 Hydrogen diffusion

Diffusion is the movement of a substance from a high concentration to a low concentration. The driving force for the diffusion of hydrogen is a gradient of chemical potential typically due to a variation in concentration C (mol cm^-3). The diffusion flux of hydrogen J (mol cm^-2 s^-1) can be represented with Fick’s first law:

where D (cm² s^-1) is the diffusion coefficient. The diffusion flux is proportional to the concentration gradient. Eq. (7) applies to an isotropic medium. The diffusion coefficient D is exponentially dependent on temperature T as given by

where Q (kJ mol^-1) is the activation energy, D₀ (cm² s^-1) is the constant pre-exponential term, and R (J mol^-1 K^-1) is the gas constant.

To understand hydrogen transportation in the bulk steel, it is important to know the hydrogen diffusion coefficient D (apparent and intrinsic) and the subsurface hydrogen concentration C₀, which allows the evaluation of the amount and diffusion speed of hydrogen in the steels. These two crucial parameters can be measured using permeation experiments.

Devanathan and Stachurski (1962) established a sensitive electrochemical technique that permitted the recording of the instantaneous permeation rate of electrolytic hydrogen through palladium membranes using a simple double-cell arrangement. Figure 5 presents a schematic drawing of an electrolytic permeation apparatus (Atrens, Mezzanotte, Fiore, & Genshaw, 1980; Liu & Atrens, 2015; Liu, Atrens, Shi, Verbeken, & Atrens, 2014).

Figure 5:

Schematic of the permeability cell.

The two cells are separated by a thin metal membrane electrode of the test material. One side (typically the left side) of the membrane acts as the cathodic side, and the opposite side acts as the anodic side. The cathodic polarization of one side and anodic polarization of the opposite side ensure that the coverage of hydrogen on the membrane is maintained at a certain fixed concentration on one side, and on the opposite side, the hydrogen concentration is zero. As viewed from the hydrogen evolution and entry reactions, hydrated protons are reduced from the aqueous solution under cathodic polarization at one side of the cell mostly by combining to form hydrogen gas bubbling away from the electrode surface, and a small fraction absorbed at the membrane surface diffuses through the membrane to the opposite side, which is under anodic polarization. The anodic polarization oxidizes the emerging hydrogen atoms and removes them from the membrane. Thus, an increase of current as a function of time is recorded on the anode side. This current increases and reaches a steady value. The recorded current is a direct measure of the instantaneous permeation rate of hydrogen. The sensitivity of this method reveals details in the permeation. Subsequently, much research (Addach, Berçot, Rezrazi, & Takadoum, 2009; Atrens et al., 1980; Devanathan & Stachurski, 1964; Liu et al., 2014; Manolatos, Duret-Thual, Le Coze, Jerome, & Bollinger, 1995a; Manolatos, Jerome, Duret-Thual, & Le Coze, 1995b; Manolatos, Jerome, & Galland, 1995c; Wang et al., 2013; Zakroczymski, 2006) on hydrogen diffusion and permeability has been conducted based on the method of Devanathan and Stachurski.

When using the technique of Devanathan and Stachurski to study the diffusion and permeation of steels, a palladium coating is typically applied on the anodic side (Atrens et al., 1980; Devanathan & Stachurski, 1964; Liu et al., 2014; Manolatos & Jerome, 1996; Zakroczymski, 2006). Without a palladium coating, a passive layer could be formed on the detection side and acts as a barrier to hydrogen permeation (Addach et al., 2009; Manolatos et al., 1995a,b,c). Atrens et al. (1980) introduced a palladium layer on the detection side of their Ni specimen to avoid Ni oxidation during their study of hydrogen diffusion and permeability in Ni. Devanathan and Stachurski (1964) also adopted a palladium coating on the detection side to reduce the possibility of passivation when they studied the hydrogen diffusion in iron. In contrast, the studies of Manolatos et al. (1995a,b) did not use palladium coating on the hydrogen exit side. They found that a passive layer was formed and was a barrier to hydrogen permeation depending on the time and the applied potential on the exit side. They further suggested that this barrier might cause hydrogen accumulation and thus result in unreliable hydrogen diffusion parameters. Moreover, they proposed that these phenomena did not occur with a palladium layer on the detection side, as the oxidation of hydrogen occurs fast and completely at a palladium-coated surface. Similarly, Addach et al. (2009) studied hydrogen permeation in iron without a palladium layer on the detection side and also found a lower hydrogen permeation flux with a longer passivation time. They attributed this difference of the permeation curves to the passive layer formed on the detection side of the specimen acting as a barrier to the hydrogen permeation.

Permeation experiments allow the determination of the diffusion coefficient D (apparent and intrinsic) and the subsurface hydrogen concentration C_H via the following methods.

3.4 Hydrogen diffusion measurements

Atrens et al. (1980) studied hydrogen diffusion and permeability in annealed Ni from 25°C to 90°C. In a typical permeability experiment, after a potential was applied to the input side of the permeability membrane, the current increased from zero to a steady-state value. Various potentials were applied in succession after steady state had been attained at the prior potential. Depending on whether the potential was increased or decreased, rise or decay transients were obtained, respectively. Individual rise and decay transients and successive rise transients were recorded at each temperature. The analysis of these transients using the aforementioned successive transient method gave values of the diffusion coefficient D. Figure 6 presents a plot of D versus T^-1 of the authors’ data compared to the data designated KGB of Katz, Guinan, and Borg (1971). The dashed line designated KGB represents D values determined from outgassing of high purity (99.999%) Ni single crystals at higher temperatures. The solid line was for permeation experiments with rise transients. There was a small but significant decrease in D from the KGB line to the D_rise data and to the D_decay data. These differences can be explained in terms of differences in H trapping. Due to the use of high purity Ni, the trapping effect was small.

Figure 6:

Variation of the diffusivity D with T^-1 (Atrens et al. 1980).

Liu et al. (2014) determined the relationship between electrolytic charging conditions and hydrogen fugacity of low interstitial steel (see Figure 7) using the permeation technique based on the method of Devanathan and Stachurski. Palladium was plated on the detection side of the specimen to prevent oxidation. Then, 0.1 m NaOH solution was used in the hydrogen detection cell, whereas the charging cell contained either acidified (pH 2) 0.1 m Na₂SO₄ solution or 0.1 m NaOH solution. The successive transient method was used for evaluating the diffusion coefficient D. A series of successive partial rise and decay transients were recorded in both acid and alkaline charging solutions. The average D values obtained from acid and alkaline charging solutions were 6.4±0.7×10^-5 and 4.4±0.4×10^-5 cm² s^-1, respectively.

Figure 7:

Hydrogen fugacity versus overpotential for a low interstitial steel from the work of Liu et al. (2014) charged in (i) 0.1 m NaOH (squares and circles) and (ii) acidified (pH 2) 0.1 m Na₂SO₄ solution (triangles).

Begić Hadžipašić, Malina, and Malina (2011) studied the influence of microstructure on hydrogen diffusion and embrittlement of multiphase structural steels with increased plasticity and strength. The diffusion of hydrogen was studied using Devanathan and Stachurski’s method, with 2 mol L^-1 H₂SO₄ as the charging solution and 1 mol L^-1 NaOH on the anodic side. The anodic side of the metal membrane was coated with nickel to ensure that the membrane was passivated and corrosion resistant. The effective diffusion coefficient D_eff values calculated by the time-lag method for DP and TRIP steels were 7.5×10^-7 and 5.7×10^-7 cm² s^-1, respectively. The effective diffusion coefficient D_eff referred to the diffusion coefficient obtained with the influence of trapping effects and thus was lower than the diffusion coefficient obtained from low interstitial steels by Liu et al. (2014).

3.5 Hydrogen trapping

4.1 Introduction

HE, first described by Johnson (1874), is the designation of the phenomenon in which hydrogen decreases the mechanical properties of a metal. This degradation can range from (i) subcritical crack growth leading to fracture at a low applied stress to (ii) some decrease in ductility with no decrease in yield stress, no decrease in the tensile strength, and no subcritical crack growth. HE is of concern for high-strength steels (Hardie, Charles, & Lopez, 2006). Hydrogen may be unintentionally introduced into the steel by finishing and forming processes (Lovicu et al., 2012), such as electroplating, forging, and welding, and by corrosion and cathodic protection (Robinson & Kilgallon, 1994). The degree of HE is dependent on many variables such as hydrogen concentration, temperature, level and type of applied or residual stress, microstructure, and surface condition of the steel (Lovicu et al., 2012; Michalska, Chmiela, Łabanowski, & Simka, 2014). Susceptibility to HE has been found to increase successively for lower bainite, quenched and tempered martensite or bainite, pearlite or spheroidized structures, and untempered martensite. In addition, steels with finer grain size and carbide size are more resistant to HE (Hirth, 1980; Lovicu et al., 2012).

Many mechanisms of HE have been proposed. The three most likely mechanisms of HE in steel, which is a non-hydride-forming material, are outlined below.

4.2 Hydrogen-enhanced decohesion (HEDE)

The decohesion theory proposed by Troiano (1960) and developed by Oriani (1970) and Oriani and Josephic (1974) is based on the idea that hydrogen accumulated within the lattice reduces the interatomic cohesive forces. The reduction of interatomic cohesive forces results from the increased interatomic distance due to the electron transfer from the 1s band of hydrogen to the 3d and 4s bands of iron.

Oriani and Josephic (1974) postulated that the highly elastically stressed region at the crack front lowered sufficiently the chemical potential of dissolved hydrogen, which attained a concentration several orders of magnitude larger than in normal lattice sites and lowered the cohesive energy. Cracks propagated when the local crack tip tensile stress exceeded the atomic cohesive energy. This fracture initiated at a distance ahead of the crack tip where the tensile stress was a maximum.

The HEDE mechanism is supported by the facts (Oriani, 1987) that (i) a large concentration of hydrogen should accumulate at crack tips where there are high stresses and (ii) atomistic simulations reveal that hydrogen can reduce atomic cohesion. The HEDE mechanism considers that there is a critical concentration of hydrogen, which causes brittle fracture. HEDE could also cause intergranular fracture, in which a high concentration of hydrogen accumulates at grain boundaries and thus reaches the critical concentration for brittle fracture.

4.3 Hydrogen-enhanced localized plasticity (HELP)

This mechanism, first formulated by Beachem (1972), proposes that hydrogen enhances dislocation motion so that the localized plastic deformation is large enough to cause subcritical crack growth with macroscopically brittle characteristics. The presence of hydrogen in solid solution increases dislocation mobility and creates localized high deformation regions. The increase of dislocation mobility is attributed to the reduction in the interactions between dislocations or between dislocations and other obstacles. The dislocations thus move closer to each other and to obstacles, forming more compact pile-ups and less ductile zones. These dislocation pile-up zones are surrounded by high deformation regions, and the applied stress is concentrated on these zones that macroscopically occupy a small portion of the cross-section. Failure occurs when the tensile stress in these zones is higher than the ultimate tensile strength of the material. This fracture process leads to cracking by microvoid coalescence along preferred crystallographic glide planes (Abraham & Altstetter, 1995; Liang, Ahn, Sofronis, Dodds, & Bammann, 2008).

Robertson (2001) conducted deformation studies in a hydrogen environment in situ in a transmission electron microscope (TEM) equipped with an environmental cell to elucidate the mechanisms of HE. The HELP mechanism was supported by observations revealing an increased number of dislocations in a pile-up, and decreased stacking fault energy, as well as the increased crack propagation rate caused by solute hydrogen.

4.4 Adsorption-induced dislocation emission (AIDE)

The AIDE mechanism was developed by Lynch (1988, 2012). The AIDE model involves both dislocation nucleation and subsequent movement away from the crack tip. The nucleation of dislocations is facilitated by the adsorbed hydrogen at the surface of the crack tip. During the nucleation, the adsorbed hydrogen weakens the interatomic bond, facilitating the simultaneous formation of a dislocation core and a surface step by the breaking and reforming of interatomic bonds. Once the nucleation is accomplished, dislocations can readily move away from the crack tip under the applied stress, contributing to the crack growth.

In the AIDE model, in addition to dislocation emission, crack growth involves the nucleation and growth of microvoids at the crack tip. The nucleation and growth of voids occurs because the stresses for dislocation emission are so high that some dislocation activity occurs ahead of the crack. Although the void formation can contribute to the crack growth, the crack growth primarily occurs by the dislocation emission from crack tips.

4.5 Other mechanisms

Other mechanisms, such as the internal pressure theory (Zapffe & Sims, 1941) and hydride-induced embrittlement (Petch & Stables, 1952; Westlake, 1969), have been proposed. There are also suggestions (Gangloff, 2003; Koyama, Tasan, Akiyama, Tsuzaki, & Raabe, 2014) of mixed mechanisms.

5.1 Mechanical degradation

The influence of hydrogen on the mechanical properties is often studied using slow tensile tests, which enable the measurements of the yield strength, ultimate tensile strength, total elongation, and reduction in cross-section. A smooth cylindrical, a notched cylindrical, or a plane sample is subjected to a slow, continuously increasing tensile strain in air or in various hydrogen environments until failure. Two widely used methods are the SSRT (Koyama et al., 2012a,b; Lovicu et al., 2012; Ryu et al., 2012; So et al., 2009; Zhu, Li, Zhao, Wang, & Jin, 2014) and the linearly increasing stress test (LIST; Atrens, Brosnan, Ramamurthy, Oehlert, & Smith, 1993; Gamboa & Atrens, 2005; Liu, Irwanto, & Atrens, 2013; Ramamurthy, Lau, & Atrens, 2011; Ramamurthy & Atrens, 2010; Venezuela, Liu, Zhang, Zhou, & Atrens, 2015; Villalba & Atrens, 2008a,b, 2009; Winzer, Atrens, Dietzel, Song, & Kainer, 2008).

The LIST allows the direct measurement of the stress at which hydrogen-induced cracking begins. For SSRT, the strain is increased until fracture. Embrittlement can be characterized using (i) the threshold stress for the initiation of subcritical hydrogen cracking, (ii) the change of ductility, and (iii) the HE index. The change of ductility is measured either as a total elongation (e) or as the reduction in area at fracture (R_A%).

where l_f (mm) is the final gauge length, l_i (mm) is the initial gauge length, A_i (mm²) is the initial area, and A_f (mm²) is the final fracture area.

The HE index I can be evaluated using (Begić Hadžipašić et al., 2011; Loidl, Kolk, Veith, & Göbel, 2011)

where R_A,air is the reduction in area in air and R_A,H is the reduction in area in the hydrogen charging environment. The index can range from 0 to 100%, where 0% indicates no HE and 100% indicates zero ductility in hydrogen.

Loidl et al. (2011) studied the HE of automotive AHSS. SSRTs were conducted in helium and a gaseous hydrogen environment to investigate the effect of hydrogen on the mechanical properties of various AHSS, including DP 1000 and DP 1200. The changes of strength, total elongation, and HE index were used to characterize the effect of hydrogen on the steels. For the DP steels, neither the yield strength nor the ultimate tensile strength changed significantly in the presence of hydrogen, meaning that there was no noteworthy effect of hydrogen on the strength of the steels. However, there was a considerable loss of ductility for the DP steels. The DP 1000 and DP 1200 experienced decreases of total elongation from 12% to 7% and from 9% to 5%, respectively, resulting in HE indices of 40% and 47%, respectively, which were higher than those of CP and MS steels with similar strengths, tested under the same conditions. However, the authors claimed that the SSRT was just one possible method to judge HE in steels. They suggested that other methods, such as HDF with static loading, also need to be investigated to gain a more comprehensive understanding of the behavior of different steel grades under the influence of hydrogen.

Begić Hadžipašić et al. (2011) obtained similar results in their investigation of the influence of microstructure on HE of DP and TRIP steels. Tensile testing was carried out to evaluate the changes of mechanical properties of hydrogen-charged specimens. The hydrogen charging was conducted by cathodic polarization at -700 mV_SCE for 4 h at 19±2°C. The tensile tests were started within 30 min after the cathodic polarization. Hydrogen did not significantly influence the yield strength and ultimate tensile strength but decreased total elongation and reduction of area. The reduction of area of DP steel decreased from 64% to 26%, resulting in an HE index of 59%, indicating that the DP steels had low resistance to HE, which they ascribed to the presence of martensite.

Davies (1981) also studied the influence of hydrogen on DP steels using two DP steels with tensile strength of 690 MPa. Hydrogen charging did not significantly decrease the tensile strength, but the total elongation decreased considerably from 23% to 6% and from 21% to 5%, respectively. Scanning electron microscopy (SEM) observations and delayed fracture tests reinforced the conclusion that the DP steels were susceptible to HE, and this susceptibility was due to the presence of high-carbon, high-strength martensite islands in the microstructure.

Koyama et al. (2014) studied the HE of DP steels. Neither the yield nor the tensile strength changed significantly after hydrogen charging, whereas the ductility decreased by 52%, indicating that the DP steel was influenced by HE.

Depover, Pérez Escobar, Wallaert, Zermout, and Verbeken (2014) studied the effect of hydrogen charging on the mechanical properties of several AHSS, including DP steel, by tensile tests using two different speeds. The ductility loss was 54% for DP steel tested at the faster speed and 73% at the lower test speed. Similar results were obtained when they investigated the role of hydrogen diffusion on HE susceptibility of DP steels (Depover, Wallaert, & Verbeken, 2016). They determined the hydrogen uptake as a function of electrochemical precharging time and the corresponding influence of different charging time on the HE resistance. They found an increased loss of ductility and brittle features with longer precharging time, up to 50% loss of ductility for samples saturated with hydrogen. They further proposed that the HE sensitivity increased with decreasing test speed, providing hydrogen with enough time to diffuse to the stressed region ahead of the crack tip and facilitate the crack propagation.

5.2 Fractography

Davies (1981) observed a fracture change during HE from ductile dimpling to transgranular cleavage in their study of the HE of DP steels. The steel failed by ductile shear with considerable local necking without hydrogen. However, after hydrogen charging, the steel failed in a brittle manner, which was at right angles to the tensile axis, with no necking. Furthermore, Davies indicated that the fracture involved cleavage of both martensite and ferrite phases in the DP steels, and these cleavage cracks probably initiated in the stronger martensite or at the interfaces between martensite and ferrite and then propagated through the softer ferrite. The crack initiation might result from stress concentration caused by the different strain hardening rates of ferrite and martensite. The presence of hydrogen-assisted crack initiation was caused either by lowering the lattice cohesive energy resulting in transgranular fractures or by changing the ferrite slip character resulting in increasing stress concentration at the head of a dislocation pile up or by a combination of both.

Sun, Gu, and Chen (1989) also found that cracks initiated along the interfaces between martensite and ferrite and along the martensite lath boundaries when they investigated the hydrogen influence on martensite and ferrite DP steels. They cathodically charged the specimens with hydrogen in a 5% sulfuric acid solution containing a small amount of arsenic trioxide at different current densities for 30 min. They found that the martensite lath boundaries widened with hydrogen charging. When the steel was cathodically charged at a high current density, a microcrack was nucleated along the boundary of lath martensite. The widening also occurred at the interfaces between martensite and ferrite. The interfaces widened and changed from a smooth form into a tortuous form with increasing cathodic charging current density, resulting in a significant local stress at these interfaces, and crack initiation.

Koyama et al. (2014) studied the hydrogen influence on DP steels using a high-resolution SEM-based damage quantification technique, postmortem electron channeling contrast imaging, electron backscatter diffraction (EBSD) analyses, and in situ deformation experiments. They found that the cracks were initiated in martensite and propagated along martensite and ferrite interfaces or through ferrite. Furthermore, in the crack incubation period, hydrogen reduced the critical strain for decohesion in martensite by the HEDE mechanism. When this critical strain was reached, decohesion-associated martensite cracking initiated. Upon further straining, hydrogen-facilitated cracks propagated along martensite and ferrite interfaces or through ferrite, associated with HEDE and HELP mechanisms, respectively.

Depover et al. (2014, 2016) found a change of fracture surface from ductile dimples to brittle transgranular cleavage after hydrogen charging. They observed a clear distinction between a brittle hydrogen-affected region and a ductile nonaffected zone. In their study (Depover et al., 2016), hydrogen charging started at the moment the tensile test started, so that the hydrogen diffusion distance and hydrogen effect could be studied at various cross-head displacement speed. At cross-head displacement speeds of 5 and 0.5 mm min^-1, a brittle zone was observed on both sides of the sample, whereas the fracture surface at the center was still ductile. At the lower test speed, a larger hydrogen-influenced zone was observed. However, at 0.05 mm min^-1, the distinction between the hydrogen-affected zone and the nonaffected zone became less obvious, with even some brittle features in the central region. They further claimed that their DP steel contained an Mn-rich segregation line in the center, providing more MnS precipitates and thus attributing to hydrogen-induced cracking, as also reported by others (Pérez Escobar, Miñambres, Duprez, Verbeken, & Verhaege, 2011). In addition, they suggested that the more significantly increased dislocation densities at the slowest speed attributed to an increase in hydrogen diffusion coefficient, so that hydrogen might be able to diffuse to the central part of the sample and embrittle the segregation line.

5.3 Microstructure

Martensite has been identified as the cause of HE in DP steels (Begić Hadžipašić et al., 2011; Davies, 1981; Sun et al., 1989) as cracks initiated in the martensite or at the interface between martensite and ferrite.

The influence of the martensite content on HE in DP steels (Davies, 1983) was studied using a series of specimens quenched into brine from a salt bath at different temperatures, to produce ferritic and martensite structures, containing from 5% to 45% martensite, situated along the ferrite grain boundaries. With increasing martensite content, the ferrite grain size decreased, and the tensile strength linearly increased. However, the susceptibility to HE did not linearly increase with increasing tensile strength and martensite content. The ratio of uniform elongation after hydrogen charging to that without hydrogen was taken as a measure of HE. This ratio was plotted as a function of the martensite content for all the tested DP steels. For all the five experimental steels, the embrittlement could be divided into three regions: (i) there was no HE in the first region up to 10% martensite, (ii) there was increasing HE in the second region as the martensite content increased from 10% to about 30%, and (iii) the embrittlement was constant for a martensite content >30%.

This change of HE with martensite content was attributed to the variation in ferrite and martensite morphology. The grain size of the ferrite decreased and the size of martensite islands increased in the steels with martensite content up to 10%, which did not encounter HE. The size of martensite islands was not large enough to enable the cracks formed in the martensite to grow through the large volume of soft ferrite and cause failure. When the martensite content was in the range of 10%–30%, the decreased ferrite grain size and increased size of the martensite islands along the ferrite grain boundaries allowed an easy path for crack propagation; thus, the susceptibility to HE increased. For steels containing more than 30% martensite, the regions of martensite almost linked together forming a continuous network, resulting in a constant degree of HE.

Davies (1981) studied the influence of martensite tempering at temperatures from 200°C to 500°C for DP steels that suffered from HE caused by the martensite. He postulated that tempering to soften high-carbon martensite should reduce HE. The total elongation of hydrogen-charged specimens increased with tempering temperature and at 500°C approached that of the specimen without hydrogen charging. At the same time, the yield strength increased with increasing tempering temperature, whereas the tensile strength decreased up to the tempering temperature of 350°C. Thus, the tensile strength was unchanged by hydrogen charging for tempering temperatures of 350°C and above.

Inclusions may influence hydrogen-induced crack initiation (Begić Hadžipašić et al., 2011; Pérez Escobar et al., 2011). Begić Hadžipašić et al. (2011) used SEM to show that their DP steel contained many globular and elongated inclusions identified by energy-dispersive spectroscopy (EDS) to be (Ca,Al)-oxides, (Al,Mn)-oxisulfides, and (Al,Mn)-oxicarbides. These inclusions acted as irreversible traps contributing to the susceptibility to HE of the DP steel. Elongated inclusions were also identified by Pérez Escobar et al. (2011). They found elongated MnS inclusions in the middle of the crack for their DP 600 steel and suggested that these particles play a role in the initiation of hydrogen-induced cracks.

In conclusion, hydrogen in DP steels may not decrease strength but may reduce ductility and induce more brittle features on the fracture surface. The hydrogen-associated cracks might initiate in the martensite or at the interface between martensite and ferrite and then propagate along the interfaces or through ferrite. The cracking mechanism was suggested to be HEDE and HELP. Inclusions might also play a role in the crack initiation. Hydrogen resistance could be improved by tempering.

6.1 Mechanical degradation

The hard martensite is the microstructure most susceptible to HE (Hirth, 1980). Furthermore, hydrogen has a higher diffusivity in martensite than in austenite but a higher solubility in austenite. Mine, Horita, and Murakami (2009) indicated that, when austenite transforms to martensite, the excess hydrogen is released from the austenite and can rapidly diffuse through the martensite. Because the crack propagation needs continuous hydrogen transport and accumulation at the crack tip, the crack advance was controlled by the hydrogen transport (Oriani & Josephic, 1974). Thus, the increased diffusivity of hydrogen assisted crack propagation. Consequently, there is concern that TRIP steels, which have TRIP, are susceptible to HE.

Ronevich, Speer, and Matlock (2010) studied the HE of commercial TRIP steels and showed that increasing hydrogen content caused an increase of ductility loss without significant change in the deformation behavior.

Lovicu et al. (2012) also observed similar changes in stress-strain curves in SSRTs on TRIP 800 with different hydrogen contents. There was a significant increased ductility loss with increased hydrogen content. Furthermore, a critical hydrogen concentration was needed to cause HE. The critical hydrogen concentration of TRIP 800 steel was about 2.5 wt ppm, which caused an HE index of about 27%. Combined with fractographic analysis, it was concluded that the TRIP 800 steel had some susceptibility to HE, attributed to the high hydrogen uptake of the austenite and the tendency of this austenite to transform into embrittled martensite. Similarly, Ryu et al. (2012) showed a decrease in ductility with increasing hydrogen content for TRIP steels.

In a comparative study of HE of different multiphase ferritic steels (Loidl et al., 2011), TRIP 700 showed the highest tendency for HE. SSRTs indicated that the total elongation of TRIP 700 decreased from 30% to 10% resulting in an HE index of 66%, which was higher than other CP, DP, and MS steels tested under the same conditions. In the TRIP steels, the deformation from austenite to martensite was the critical factor for HE, which was different from the mechanisms of other steels.

6.2 Fractography

TRIP steels experience a significant degradation of mechanical properties with increasing hydrogen content. Furthermore, the fracture surfaces exhibit a change from ductile dimples to brittle features, with an increasing brittle percentage with increasing hydrogen content (Loidl et al., 2011; Lovicu et al., 2012; Sojka et al., 2011; Zhu et al., 2014).

Lovicu et al. (2012) observed that fracture surfaces became increasingly brittle with increasing hydrogen content. The uncharged specimen exhibited a completely ductile fracture. With increasing hydrogen content, above the critical concentration of 2.5 wt ppm, brittle features took up an increasing area of the fracture surface.

Sojka et al. (2011) investigated the influence of hydrogen on the fracture characteristics of two TRIP 800 C-Mn-Si steels at different hydrogen charging current densities of 15 and 30 mA cm^-2 and with hydrogen absorption promoters. Charging lasted for 8 h at ambient temperature. The fractographic analysis showed that, without hydrogen charging, the failure characteristics of both steels were ductile failure characterized by a dimple morphology. After hydrogen charging, the fracture surfaces showed a change in the failure mechanism. For the lower charging current density of 15 mA cm^-2 without a hydrogen absorption promoter, the fracture surfaces consisted of a mixture of transgranular ductile and transgranular cleavage fracture. For the higher current density of 30 mA cm^-2, or with a hydrogen absorption promoter in the solution, transgranular cleavage fracture became dominant for the fracture surfaces.

Lovicu et al. (2012) indicated that the cracks initiated near the regions of hard particles based on the transformation from austenite to martensite. In TRIP steels, the hydrogen behavior was influenced by the presence of austenite. Due to the high solubility and low diffusivity of hydrogen in austenite, austenite was enriched in hydrogen with a higher concentration than the other phases. When the transformation from austenite to martensite occurred, because martensite has the highest susceptibility to HE, the hydrogen-enriched transformed martensite cracked immediately, causing failure of the specimen. During this deformation, the unstable austenite was made less stable by the presence of hydrogen, which induced further martensite transformation and failure. This crack initiation mechanism was supported by other researches. Imlau, Bleck, and Zaefferer (2009) suggested that the crack-forming region was highly correlated with the martensite transformation regions. Ronevich et al. (2010) found that some cracks initiated from hard martensite islands and propagated into ferrite grains. Laureys et al. (2016) highlighted similar crack initiation mechanisms in their study of hydrogen-induced cracking in TRIP steel using EBSD. They found that cracks initiated mostly in or between two MS regions by decohesion and then propagated in the ferrite or the martensite and the matrix interface. Zhu, Li, Zhao, and Jin (2013) also used EBSD to study hydrogen-induced cracking in TRIP steels and revealed that cracks initiated in martensite, which was freshly transformed from the unstable austenite, and propagated into the adjacent ferrite.

6.3 Microstructure

The microstructure, alloying elements, deformation, and manufacture processes can affect the influence of hydrogen on steels.

Many researches (Loidl et al., 2011; Lovicu et al., 2012; McCoy & Gerberich, 1973; Ronevich et al., 2010, Ronevich, De Cooman, Speer, De Moor, and Matlock 2012a) attribute HE of TRIP steels to the transformation of austenite into martensite on straining. McCoy and Gerberich (1973) stated that the TRIP steel was relatively immune to HE in the austenitic state but, when the austenite transformed into martensite induced by straining, the transformed martensite was brittle and led to premature failures of TRIP steels.

Ryu et al. (2012) studied the effect of deformation on the hydrogen influence on TRIP steels with respect to hydrogen trapping and effusion, characterized using TDS. Plastic deformation caused the transformation from austenite to martensite and led to an alternation of the inherited trapped hydrogen. The austenite in TRIP steels had a relatively larger solubility for hydrogen than that of generic traps in ferrite and martensite, so the austenite could be regarded as a more potent trapping site than dislocations or grain boundaries in ferrite. In this case, the deformation-induced MS transformation led to the reduction of trap binding energy, leaving the inherited hydrogen with enhanced mobility, likely resulting in a more damaging effect on mechanical properties than in steels without the presence of deformation-induced transformation from austenite to martensite.

Ronevich et al. (2012a) studied the effect of dislocations on the distribution on HE of TRIP steels. The TRIP steels were prestrained 5% at the different temperatures of 253, 296, and 375 K to introduce various amounts of transformed martensite and then cathodically charged to a hydrogen content of 1–2 ppm. TDS indicated that the dislocations in TRIP steels were the main trapping sites for hydrogen, and the density of these trapping sites increased during deformation. Furthermore, dislocations in TRIP steels prestrained 5% at the lower temperatures of 253 and 296 K were clustering adjacent to transformed martensite, whereas, in TRIP steels prestrained 5% at higher temperatures of 375 K, dislocations were less localized and distributed more uniformly, dispersing hydrogen and providing better resistance to HE.

Hojo, Kobayashi, Kajiyama, and Sugimoto (2010) investigated the influence of the alloying elements Al, Nb, and Mo on HE of ultra-high-strength TRIP steels developed for automotive structural application. The addition of Al, Nb, and Mo increased the amount of stable retained austenite, mainly resulting from the addition of aluminum. The total hydrogen concentration in the steels after charging increased with increasing volume fraction of retained austenite, because hydrogen was preferentially trapped in the retained austenite or at the interface between retained austenite and matrix. The normal trapping at grain boundaries, at dislocations, and at carbide-matrix interfaces, which made it easier for crack propagation, was suppressed, resulting in less effect of hydrogen on the properties of those steels with Al, Nb, and Mo addition.

Zhu et al. (2013) explored the effect of a cryogenic tempering (CT) treatment on the HE of TRIP steels using SSRTs, EBSD observations of microstructure evolution and crack initiation, and SEM examination of fractured surfaces. They found that, with the CT treatment, the TRIP steel exhibited better resistance to HE, characterized by lower loss of total elongation and less brittle “flat” features compared to the untreated counterparts. The reduced susceptibility to HE after the CT treatments was attributed to the CT treatment contributing to the enhanced stability of the retained austenite and the reduced amount of fresh untempered martensite after straining, thus lowering the local stress intensity and hydrogen concentration and reducing the possibility of occurrence of crack initiation caused by hydrogen. Furthermore, from a comparison between specimens with CT treatment and without CT treatment, the specimens with CT treatment had a better dislocation distribution avoiding dislocations clustering adjacent to retained austenite. This dislocation distribution may lower the hydrogen concentration in local regions near retained austenite, which were sensitive to HE. This dispersion of dislocations, caused by the CT treatment, may be another reason for the higher resistance to HE.

Sojka et al. (2011) attributed the poor resistance of their steels to HE partially to the nonmetallic inclusions in the TRIP steels, around which high hydrogen content was expected. These nonmetallic inclusions were preferential initiation sites for HE. A similar concept relating to inclusions for the easier crack initiation in TRIP steels was also stated by Begić Hadžipašić et al. (2011).

In conclusion, TRIP steel is influenced by hydrogen, in that the ductility is reduced, whereas the strength is not affected. Different critical hydrogen concentrations were proposed. The various hydrogen concentrations were due to the different microstructure of the TRIP steels. The hydrogen influence on TRIP steels was attributed to the transformation of retained austenite to martensite during straining. The retained austenite, which has a higher solubility of hydrogen, acted as a detrimental hydrogen source, releasing hydrogen to the freshly formed hard martensite, which was susceptible to HE. The susceptibility of TRIP steels to HE can be improved by a CT treatment.

7.1 Mechanical degradation

The research results are not consistent with regard to HDF. There are reports that the TWIP steel is prone to delayed fracture after forming attributed to HE (Chin et al., 2011; Chun et al., 2012). There are also reports stating that some grades of TWIP steels are immune to HDF (Chin et al., 2011; So et al., 2009). Furthermore, it was also reported that, by adding Al as an alloying element, the HDF susceptibility can be decreased or even avoided (Park, Jeong, Jung, Lee, & Lee, 2012). Investigations concerning mechanical degradation after introducing hydrogen are also not consistent.

Some researches have observed negligible mechanical degradation using tensile tests with hydrogen precharged specimens (Jung, Lee, Park, Kim, & Jin, 2008; Ronevich, Kim, Speer, & Matlock, 2012b; Ronevich et al., 2010; So et al., 2009). Similarly, So et al. (2009) found there was little difference of the mechanical properties of Fe-18Mn-1.5Al-0.6C TWIP steel charged with different hydrogen contents and those without hydrogen charging. Similarly, Ronevich et al. (2010, 2012a) studied the influence on TWIP steels of different hydrogen contents introduced by cathodic charging the steels at the same current density for different durations before the tensile tests. The tensile properties, including yield stress, ultimate tensile stress, strain to failure, and strain hardening behavior, of the TWIP steels without hydrogen, and with different hydrogen contents, were essentially the same. It was concluded that the TWIP steels in their studies were not affected by hydrogen, attributed to the low hydrogen diffusion coefficient in austenite, introducing basically no hydrogen into the bulk steel and leading to the steels unaffected by hydrogen.

Suh (2014) critically reviewed hydrogen-induced fracture in TWIP steels and concluded that, because of the low diffusion coefficient of hydrogen in austenite, there was limited penetration of hydrogen into the bulk of the steels, resulting in the little change of the mechanical properties of TWIP steels.

In contrast, Koyama et al. (2012a,b), Koyama, Akiyama, and Tsuzaki (2012c) found a significant decrease of mechanical properties, caused by the presence of hydrogen, of the Fe-18Mn-0.6C and Fe-18Mn-1.2C TWIP steels. Tensile tests measured the change of mechanical properties. The specimens were cathodically charged in a 3% NaCl aqueous solution containing 3 g/l NH₄SCN at current density of 10 A m^-2 during the tensile tests to introduce hydrogen. Koyama et al. (2012b) found that their hydrogen charged Fe-18Mn-0.6C TWIP steels had values of average total elongation and ultimate tensile strength of 32% and 1010 MPa, respectively, compared to 70% and 1200 MPa without hydrogen. For Fe-18Mn-1.2C TWIP steel (Koyama et al., 2012c), there was a similar reduction of elongation and ultimate tensile strength by hydrogen, which was about 20% reduction of ultimate tensile strength and about 40% reduced elongation from about 80% to 42%.

7.2 Fractography

Researchers who measured a negligible decrease of mechanical properties caused by hydrogen observed a fully ductile fracture surface morphology (Ronevich et al., 2010, 2012b; So et al., 2009), whereas those who measured a significant decrease of mechanical properties after hydrogen charging observed brittle fracture surfaces (Chun et al., 2012; Koyama et al., 2012a,b,c).

For those TWIP steels that experienced negligible degradation of mechanical properties with hydrogen charging, the fracture surface after the SSRTs of specimens both without and with hydrogen charging contents showed fully ductile fractures, characterized by dimples, resulting from microvoid nucleation and coalescence.

In contrast, Koyama et al. (2012b) reported that the hydrogen introduced into the TWIP steels influenced the characteristics of the fracture surface along with the total elongation. The fracture surfaces of the charged specimens exhibited intergranular features in the parts near the surface of the specimen, whereas, in the central part of the specimen, there were ductile dimples that resembled those of the uncharged specimen. The intergranular fractures were only located near the surface. This was attributed to the hydrogen concentration difference from the surface to the interior. Near the surface, the hydrogen concentration was high, and this high concentration of hydrogen induced the intergranular fracture. However, the hydrogen concentration in the interior was low and not high enough to induce intergranular fracture. They attributed the causes of the intergranular fracture to the reduction of grain boundary cohesion, deformation twinning, and martensite transformation, influenced by hydrogen. In another of their research concerning hydrogen in TWIP steels (Koyama et al., 2012c), there were also brittle features in the vicinity of the surface and ductile dimples in the interior. However, the brittle features were different, as the fracture mode was not all intergranular but also included some transgranular fracture. This transgranular fracture propagated along the primary deformation twin boundaries and parallel to the secondary twinning plane. The intergranular fracture resulted from the interception of grain boundaries at primary deformation twin boundaries. The following twin-twin interaction, in which the primary deformation twins were intercepted by the secondary deformation twin boundaries, caused the transgranular fracture.

As mentioned previously, crack initiation and propagation were observed on even coherent Σ3 twin boundaries. This was in contradiction to some studies (Chun et al., 2012; Oudriss et al., 2014), suggesting that coherent Σ3 twin boundaries as CSL boundaries were crack arrest sites, as introduced in Section 3.3.3. Koyama, Akiyama, and Tsuzaki (2013a) further explained this contradiction in four aspects: (i) very high stress concentration existed at the tip of a moving deformation twin contributed to crack initiation; (ii) the dislocation-twin interaction influenced the coherency of the twin boundaries; (iii) the change in the position of interstitial carbon atoms from an octahedral to a tetrahedral site lead to lattice distortions in the deformation twins, producing a local elastic strain; and (iv) a large amount of lattice defects, such as dislocations and nanotwins, existed in and around deformation twins, acting as reversible hydrogen trapping sites.

7.3 SFE

During austenite plastic deformation, the following three mechanisms could occur: martensite transformation, twinning, and dislocation glide. The SFE is a critical factor that determines which transformation occurs. The SFE depends on the temperature and the chemical composition of the TWIP steel, particularly concentrations of manganese, aluminum, and silicon. The effect of SFE on the deformation mechanism (Remy & Pineau, 1977) is that twinning occurs when the SFE is within the range of 18–50 mJ m^-2. If the SFE is lower, the martensite transformation is more likely. If the SFE is higher, dislocation glide is the only possible mechanism (De Cooman et al., 2012). However, different ranges have been reported for the value of SFE for mechanical twinning, such as 18–45 mJ m^-2 (Curtze & Kuokkala, 2010), 20–40 mJ m^-2 (Lee, 2012), and 12–35 mJ m^-2 (Allain, Chateau, Bouaziz, Migot, & Guelton, 2004). These differences result from the difficulty in experimentally determining the value of SFE (Lee, 2012).

It has been reported that hydrogen, under cathodic charging conditions, can cause the austenite phase of austenitic stainless steels to transform to α′- or ε-martensite (Mine et al., 2009; Rozenak & Bergman, 2006; Yang, Qiao, Chiovelli, & Luo, 1999; Yang & Luo, 2000). These studies indicate that hydrogen can decrease austenite stability. Furthermore, Yang et al. (1999) found that there was surface cracking only in specimens, in which there was hydrogen-induced MS transformation, indicating that hydrogen could reduce austenite stability, induce austenite to transform into martensite, and cause mechanical property degradation. This could be explained based on (i) the assumption that the hydrogen lowers the SFE of austenite and (ii) the increased stress gradients. Pontini and Hermida (1997) measured the reduction of SFE caused by hydrogen in an AISI 304 steel by X-ray diffraction (XRD) with the aid of TEM and found that the SFE was reduced about 40% by 274 ppm hydrogen.

For TWIP steels, Ryu, Kim, Lee, Suh, and Bhadeshia (2013) found a reduction of the SFE in Fe-0.6C-18Mn TWIP steel charged with hydrogen. This reduction of SFE led to the transformation from austenite to ε-martensite, which was susceptible to HE, resulting in transgranular cracking parallel to interfaces between ε-martensite and austenite. Similarly, Neu (2013) reported that a decrease of SFE could cause the transformation of α′-martensite. McCoy (1973) stated that, with decreasing Mn content in Fe-(15-25)Mn-0.3C steels, α′-martensite was formed, resulting in decreased HE resistance.

One possible explanation for the reduction of SFE is the decrease in energy associated with the formation of H-H pairs during faulting (Moro, Obiol, Roviglione, Hermida, & Juan, 1998).

7.4 Grain boundaries

Because there are reports of significant degradation of mechanical properties of TWIP steels by hydrogen (Koyama et al., 2012b,c; Ryu et al., 2013), it is important to investigate the influence of hydrogen content on embrittlement behavior on TWIP steels.

Koyama et al. (2012a) studied the effect of hydrogen content on the HE in Fe-18Mn-0.6C TWIP steels through tensile tests with hydrogen charging at different current densities to introduce various hydrogen contents. The hydrogen concentration in the steels increased significantly with increasing hydrogen charging current density. There was a critical diffusible hydrogen content, which divided the effect of hydrogen on embrittlement behavior. At a relatively low diffusible hydrogen content, 0.33 wt ppm corresponding to a charging current density of 1 A m^-2, there was no significant change of tensile properties compared to those without hydrogen charging, and the fracture surface was ductile and consisted of dimples, which was similar to the fractures without hydrogen. For the samples with higher hydrogen contents, corresponding to charging current densities within the range of 3–10 A m^-2, the true stress and elongation decreased significantly with increased charging current density, premature fracture occurred, and there was intergranular fracture. The strain hardening behavior was not affected by hydrogen charging even when HE occurred. This lack of change in the strain hardening behavior was assumed to indicate that dislocation slip, MS transformation, and deformation twinning was not affected by hydrogen, as the behavior of those deformation mechanisms would be changed if the strain hardening behavior altered. The possible cause of HE behavior under the relatively high charging current densities was attributed to the reduction of the cohesion of grain boundaries, not twin boundaries, which explained the lack of change in strain hardening behavior, the occurrence of premature fracture, and the intergranular fracture. This lack of change in strain hardening behavior seems to contradict the references (Pontini & Hermida, 1997; Ryu et al., 2013), which indicated a change of SFE by hydrogen, as mentioned in Section 7.3. In the study of Ryu et al. (2013), the transformation of ε-martensite contributed to HE and was attributed to the reduced SFE by hydrogen. In contrast, in this study of the effect of hydrogen content on HE in TWIP steels, HE occurred without the transformation of ε-martensite and without hydrogen affecting the strain hardening behavior of the steel, indicating that the main reason for HE in this case was the reduction in the cohesion of grain boundaries, which could lead to the hydrogen-assisted cracking without changing stress-strain behavior of the steel, instead of hydrogen-dislocation interactions (Koyama, Akiyama, Tsuzaki, & Raabe, 2013b).

7.5 Al concentration

As mentioned in Sections 2.4 and 7.3, aluminum influences the SFE, which is a relevant parameter for determining the deformation mechanism in austenitic steels, and the deformation mechanism in TWIP steels, resulting in a change of resistance of TWIP steels to HE. As a consequence, it is important to investigate the effect of aluminum on HE in TWIP steels.

Chin et al. (2011) found that the addition of aluminum in TWIP steels decreased the tensile elongation, and the deformation twinning was more homogenous and less intense than those in TWIP steels without aluminum. This led to a lower stress concentration during cup forming than that without aluminum, which exhibited crack formation.

Ryu et al. (2013) investigated the effect of aluminum on hydrogen-induced embrittlement in TWIP steels. The addition of aluminum reduced the loss in both elongation and ultimate tensile strength. The fracture surface had regions near the surface of brittle features that were both intergranular and transgranular, and those in the center were ductile with dimples. The depth of the brittle zones increased with the increasing hydrogen charging current density but was reduced by the addition of aluminum, consistent with the results that the mechanical properties of TWIP steel with aluminum were less sensitive to hydrogen. This effect of aluminum was attributed to the fact that aluminum could increase the SFE of austenite, reducing the possibility of mechanical twinning and suppressing the transformation of ε-martensite during deformation, both of which would contribute to the hydrogen trapping and promote transgranular fracture and thus lead to a lower susceptibility of the steel to HE.

Park et al. (2012) investigated the effect of different aluminum contents on hydrogen-induced embrittlement in TWIP steels using SSRTs with hydrogen charging in aqueous solution and obtained results that were consistent with those of Ryu et al. (2013). Park et al. found that the addition of aluminum retarded the loss of mechanical properties of TWIP steels, and the area of the brittle zone in the vicinity of the surface was decreased with the increased aluminum content. For the 2Al TWIP steel, which contained higher amount of aluminum than the others, there was almost no brittle fracture after the tensile test with hydrogen charging. A different explanation of the aluminum effect on HE was given that during hydrogen charging in aqueous solution an aluminum oxide layer was formed on the surface and hindered the absorption of hydrogen into the steel. Thus, the amount of diffusible hydrogen inside the sample, which could cause hydrogen-induced embrittlement, was reduced and the resistance to HE was improved.

Consistent results showing that the HE susceptibility decreased with increasing aluminum content was also provided by Koyama et al. (2013a). Moreover, they further proposed that the strain rate of predeformed steels also affected the HE susceptibility in a way that the HE susceptibility decreased with increasing strain rate. They explained the improved resistance to HE from the viewpoint of strengthening by strain aging. It was proposed that the addition of aluminum contributed to the suppression of static strain aging under loading, and the increased Al content and strain rate led to the suppression of dynamic strain aging during predeformation. Because dynamic and static strain aging influenced the HE, the increased aluminum content and strain rate could affect the susceptibility to HE.

7.6 Influence of Ti

Park, Jo, Kang, Lee, and Lee (2014) also investigated the influence of Ti on the HE of TWIP steel. There was a significant reduction in elongation and quasi-cleavage brittle features in the hydrogen-charged sample containing Ti, indicating the low resistance of this steel to HE. Moreover, without Ti, the sample charged with hydrogen exhibited no reduction in elongation and fully ductile fracture. The reduction in total elongation was attributed to the transition from ductile fracture to brittle fracture at the hydrogen-concentrated edge part of the samples. There were TiN particles near the quasi-cleavage brittle features, revealing that the TiN particles acted as strong hydrogen traps and contributing to the low resistance of Ti-bearing TWIP steels to HE.

7.7 Summary

In summary, the results of the influence of hydrogen on TWIP steels were not consistent from the researches as introduced above. Some researchers found no significant influence of hydrogen on TWIP steels and fully ductile fracture features, and they attributed the lack of influence to the low diffusivity of hydrogen in austenite. In contrast, some other researchers found significant loss of ductility and strength due to the presence of hydrogen and some brittle features. They attributed the hydrogen influence to the reduction of cohesion between grain boundaries, deformation twinning, and ε-martensite transformation, which resulted from the decreased SFE caused by hydrogen. In addition, aluminum alloying and increased strain rate were found to decrease HE susceptibility, whereas titanium alloying was found to be detrimental.