Techniques for investigation of hydrogen embrittlement of advanced high strength steels

2.1 Grades of AHSS

There are four main groups of grades of AHSS: DP, complex phase (CP), martensitic (MART) and TRIP. Mechanical properties of selected AHSS grades are shown in Table 1. Detailed description of each grade follows.

2.2 Dual phase steels

DP steels have duplex microstructure consisting of soft ferrite matrix and 10%–40% of hard martensite islands (Kuziak et al., 2008; Demeri, 2012). DP steels derive their strength from the martensite phase and their ductility from the continuous ferrite phase. Therefore, the size, proportion and distribution of these phases are the most important features influencing mechanical properties of DP steels (Kim & Thomas, 1981; Bag et al., 1999; Park et al., 2014). For instance, Kuang et al. (2009) found that deficiency in martensite volume resulted in decreasing tensile strength and increasing yield strength, while martensite becoming the dominant phase led to both yield and tensile strength increase and an obvious drop in elongation.

Mn, Si, Cr, Mo, V, Ti and Nb are most common alloying elements in DP steels. Almost all these elements stabilize austenite. Silicon is effective in improving the balance of tensile strength and elongation (Hironaka et al., 2010). Manganese strengthens ferrite by solid solution strengthening and decreases the yield to tensile strength ratio (Fonstein et al., 2007). Chromium can contribute to a microstructure refinement or lead to a coarser grain structure depending on its content (da Silva et al., 2016). Molybdenum and chromium retard pearlite and bainite transformations (Podder et al., 2007; Girina et al., 2015). Niobium refines the microstructure and promotes ferrite transformation from austenite (Song et al., 2014; Girina et al., 2015). Titanium and vanadium increase the strength by precipitation and refine the microstructure (Kamikawa et al., 2015; Bellavoine et al., 2017).

The dual-phase microstructure is obtained by heating the material into the inter-critical region producing a mixture of ferrite and austenite followed by a very rapid cooling to transform austenite to martensite (Aksoy & Esin, 1988). A tempering at low temperatures during galvanizing or paint curing may be also used (Gündüz, 2009).

2.3 Transformation-induced plasticity steels

TRIP steels show a better combination of strength and formability in comparison with DP steels. The microstructure of TRIP steels consists of ferrite, bainite and retained austenite and minor amounts of carbides and martensite (De Cooman, 2004). During plastic deformation, soft retained austenite is transformed into martensite. The retained austenite is a key phase which provides good formability before and high strength after the transformation of austenite to martensite. This TRIP effect depends on the amount of retained austenite and its stability in view of deformation induced transformation (Matsumura et al., 1987; Jacques et al., 2001; Bhattacharyya et al., 2011).

Composition of TRIP steels has a significant influence on retained austenite stability and final strength. A higher carbon content in comparison with DP steels enables stabilization of the retained austenite at ambient temperature. Mn is added to achieve required hardenability and to stabilize austenite at room temperature. Si enhances the volume fraction and stability of the retained austenite. Al is also an austenite stabilizer and helps to avoid problems with worse weldability and coatability caused by high Si content. Ti, Ni and V may be added to increase strength (Bhattacharyya et al., 2011; Demeri, 2012). Both Al and Si suppress carbide precipitation during bainite formation.

Processing of TRIP steels includes inter-critical annealing in the austenite-ferrite region, fast cooling to temperature below bainite start temperature, isothermal holding to produce some bainite and final cooling to room temperature (De Cooman, 2004).

2.4 Complex phase steels

CP steels consist of a very fine microstructure of ferrite-bainite matrix with small volume fractions of hard phases of martensite, retained austenite, and pearlite (Demeri, 2012; Dias et al., 2014). Besides the elements found in DP and TRIP steels, CP steels have small quantities of Ti, Ni and V to form fine strengthening precipitates and to achieve very fine microstructure. This microstructure ensures good mechanical properties regarding strength and ductility due to presence of hard phases, low hardness phases and fine precipitates. In comparison to DP steels, CP steels display significantly higher yield strength values at equal tensile strength levels (Hofmann et al., 2009). Processing of CP steel is essentially similar to that of TRIP steels but with less stringent cooling practice during the last stage as the presence of retained austenite is not required in the CP steel microstructure (Kuziak et al., 2008).

2.5 Martensitic steels

Martensitic steels have ultimate tensile strength up to 1700 MPa being thus a group of materials with the highest tensile strength level among multiphase steels. The microstructure of MART steels consists of a MART matrix containing small amounts of ferrite and bainite (Demeri, 2012). Carbon strengthens martensite and increases hardenability. Mn, B, Cr, Mo, Si, V and Ni are added to increase hardenability (Mohrbacher, 2015). MART steels are produced by rapid quenching from the austenite phase to produce martensite and subsequent tempering to improve ductility (Kuziak et al., 2008).

3.1 Source of hydrogen

There are two types of HE, internal and environmental. The former type is related to hydrogen absorption during steelmaking or processing, while the latter one is a result of hydrogen absorption from external sources.

Steel can pick hydrogen up during iron and steelmaking as a consequence of water dissociation on the surface. Hydrogen absorption occurs during scale removal by acid pickling or during electrolytic cleaning, where hydrogen bubbles are formed at a cathodic surface of steel in order to remove an oil film, and a part of hydrogen is absorbed by the metal. It also occurs during annealing in an atmosphere containing hydrogen gas to prevent steel oxidation and during Zn electroplating. Casanova et al. (1997) assumed a barrier effect of zinc coatings for hydrogen absorption, but results of their measurements showed that zinc can act as a reservoir of hydrogen which can further diffuse to steel under suitable conditions.

The content of hydrogen can be reduced after production steps listed above by specific heat treatment. However, absorption of hydrogen may also occur during steel service life as a result of cathodic corrosion reaction, and HE as a consequence of corrosion may eventually cause an unexpected failure.

3.2 Mechanism of hydrogen entry

Hydrogen can be introduced into steel from a gas phase or from electrolytic solution. There are two main steps of hydrogen evolution and entry into metal: adsorption and absorption. Adsorption process is caused by excess of free energy on the metal surface and effort to diminish this free energy by attaching foreign particles.

In the case of hydrogen-containing gaseous environments, physical and chemical adsorption takes place in the context of interaction between hydrogen species and the steel surface. Physical adsorption involves weak molecular or Van der Waals forces, has a low adsorption energy and can easily occur at room temperatures. Chemical adsorption prevails at higher temperatures. It is based on attaching hydrogen atoms by chemical bonds and hence represents stronger interaction with higher adsorption energy. Hydrogen attachment to the surface is followed by incorporating a part of the hydrogen into metal lattice called absorption or bulk occlusion.

Hydrogen entry from aqueous environments includes primarily adsorption of hydrogen atoms by losing electrons to become protons which interact with the metal lattice. This Volmer reaction (Eq. 1) competes with Heyrovsky (Eq. 2) and Tafel (Eq. 3) desorption reactions (Lasia & Grégoire, 1995), where hydrogen molecules form and under suitable conditions can leave the surface as gas. The recombination process can be inhibited and hydrogen entry into metal is thus promoted by recombination poisons such as arsenic, antimony, sulfide, thiourea, etc., which block active centers of the metal by adsorbing on them (Eaves et al., 2012).

whereas a part of adsorbed hydrogen protons recombines into molecules and desorbs out, a part of the adsorbed hydrogen enters into the metal by the reaction (Eq. 4).

It should also be mentioned that solubility of hydrogen from gas phase is proportional to the square root of hydrogen pressure according to Sieverts law, whereas solubility of hydrogen from aqueous solution is proportional to the square root of current density (Perng & Wu, 2003).

Hydrogen entry as a result of atmospheric corrosion reactions occurs when pH at the metal/corrosion product interface and corrosion potential reach low values (Tsuru et al., 2005). When pH of water layer on the steel is neutral, oxygen reduction is the cathodic reaction and hydrogen evolution is negligible. However, when corrosion products are formed on the steel surface, decrease in pH makes hydrogen evolution possible. Decrease in pH can also be expected when pollutants such as SO₂ are present in the atmosphere (Tsuru et al., 2005). Formation of proton may be caused by three processes: (i) hydrolysis of ferrous ions (Fe²⁺) (Eq. 5), (ii) oxidation of ferrous ions and ferrous hydroxide (Eqs. 6 and 7) and (iii) hydrolysis of SO₂ (Eq. 8) and formation of ferrous sulphate (Eq. 9) (Nishimura et al., 2004). Hydrogen entry depends on the composition of corrosion products, which is linked to changes in the acidity (Akiyama et al., 2011a).

3.3 Hydrogen diffusion and trapping

Adsorption and absorption of hydrogen is followed by diffusion and trapping processes. The diffusion flux of hydrogen J (mol m⁻² s⁻¹) can be expressed by Fick’s first law:

where D (m² s⁻¹) is the diffusion coefficient and ∂C/∂x is the concentration gradient. Diffusion coefficient greatly depends on the system and temperature as Arrhenius equation describes.

where D₀ (m² s⁻¹) is the temperature-independent constant, Q (J mol⁻¹) is the activation energy of diffusion, R (8.314 J mol⁻¹ K⁻¹) is the gas constant and T (K) is absolute temperature.

Diffusion coefficient can be determined only experimentally by using thermal desorption spectroscopy (TDS) or electrochemical permeation methods as the most common techniques. These methods are described in Section 5.

Rehrl et al. (2014a) measured diffusion coefficient of microalloyed and standard AHSS by using the permeation method. It was 1.2⋅10⁻⁶ cm² s⁻¹ for DP1200, 1⋅10⁻⁶ cm² s⁻¹ for CP1200, 0.9⋅10⁻⁶ cm² s⁻¹ for CP1400 and from 11.5⋅10⁻⁷ to 19.4⋅10⁻⁷ cm² s⁻¹ for TM1200, where the last material is a tempered martensite (TM) steel with tensile strength 1200 MPa. They assumed that diffusion coefficient decreased with increasing complexity of microstructure and reversible trapping sites. They also found that precipitates of microalloying elements did not significantly influence hydrogen diffusion.

Besides hydrogen atoms located in interstitial sites, a part of hydrogen atoms can be immobilized in the metal structure (Stopher et al., 2016). This phenomenon is described as hydrogen trapping. Hydrogen can be trapped by microstructure features such as voids, dislocations, grain boundaries, carbide interfaces and impurities. These hydrogen traps increase hydrogen solubility and decrease the hydrogen diffusion coefficient.

Reversible and irreversible traps are often distinguished. When a hydrogen atom is trapped by an irreversible trapping site it becomes non-diffusible because the trap activation energy is higher than binding energy. For reversible trapping sites, the trap activation energy is lower and hydrogen atoms can leave these traps and influence HE susceptibility. Carbide interfaces and incoherent precipitates are examples for irreversible traps, while dislocations, coherent precipitates and twin boundaries are examples for reversible ones. Although such a classification is somewhat artificial since the reversibility of any trap would depend on the difference between binding and trap activation energies, i.e. temperature, it proved useful for description of steel properties in view of HE.

Pressouyre (1980) stated that irreversible traps should be always considered as sinks with beneficial effect on HE resistance and reversible traps could act as sinks or as injurious hydrogen sources depending on initial state of the material (hydrogen pre-charged versus not pre-charged specimen) and type of hydrogen transport (transport by dislocations versus interstitial diffusion). The most detrimental is the situation when a material is pre-charged and hydrogen moves with dislocations. All reversible traps act then as hydrogen sources, more hydrogen is transported and critical hydrogen concentration is achieved rapidly. In the case of not pre-charged specimen with normal diffusion of hydrogen, reversible traps act as sinks leading to less hydrogen penetration. Oriani (1970) assumed that trapped hydrogen is in equilibrium with hydrogen in normal lattice sites. Based on that, the author suggested that solid-solid interfaces are more important for the trapping than dislocations in non-cold-worked steels.

Trapping sites can be characterized by the hydrogen binding energy, E_b, and the hydrogen trap density, N_T. Dong et al. (2009) proposed that the hydrogen trap density, N_T (sites cm⁻³), could be estimated by the equation

where D_L (cm² s⁻¹) is the lattice diffusion coefficient of hydrogen, D_eff (cm² s⁻¹) is the effective diffusion coefficient of hydrogen in the presence of traps, N_L (sites cm⁻³) is the density of the interstitial sites in the steel and E_b (kJ mol⁻¹) is the hydrogen trap binding energy.

An alternative method for determining the hydrogen trap density applying the electrochemical permeation technique (EPT) was proposed by Zakroczymski (2006). The area under the experimental desorption permeation curve corresponds to the total amount of diffusible and reversibly trapped hydrogen, whereas the area under the theoretical permeation curve calculated using lattice diffusion coefficient D_L represents only diffusible hydrogen. Thus, the reversible trap density can be evaluated from the areas difference. Liu and Atrens (2015) suggested equation for calculating N_T^* (sites cm⁻¹) for the simplest case with only one type of hydrogen traps:

where c (A s cm⁻²) is the difference between areas of permeation curves and L (cm) is the thickness of the specimen. Each trap was assumed to hold one hydrogen atom and 1 A s (6.24⋅10¹⁸ e) was considered as 6.24⋅10¹⁸ trapping sites.

In a later work, Liu et al. (2016b) compared hydrogen trap density values for DP AHSS from the Dong method and from the permeation curve method. Total trap density evaluated by the permeation curve model was slightly smaller than that evaluated by the Dong method. It was because only one type of reversible traps was considered when calculating the hydrogen trap binding energy and the values for α-Fe were used for the density of the interstitial sites according to the Dong method. These approximations could lead to incorrect trap density values. Reversible trap density from complete decays was about 2⋅10¹⁸ sites cm⁻³.

Winzer et al. (2016) evaluated hydrogen trapping characteristics for a CP steel by fitting the permeation curves with theoretical curves calculated according to equations based on the McNabb and Foster diffusion-with-trapping model (McNabb & Foster, 1983). Average density of traps for CP1200 was 4.36⋅10¹⁸ cm⁻³.

3.4 Hydrogen embrittlement mechanism

HE in steels is currently discussed in terms of three mechanisms: hydrogen-enhanced decohesion (HEDE), hydrogen-enhanced plasticity (HELP) and adsorption-induced dislocation emission (AIDE). Other mechanisms such as the surface energy model, internal pressure model, corrosion-enhanced plasticity (for passive metals), hydrogen rich phases model or hydride-induced embrittlement (for hydride-forming materials such as Zr, Nb and V) have also been proposed.

3.5 Role of AHSS microstructure

Hydrogen distribution in the steel microstructure may significantly influence HE micro-mechanisms. AHSS grades contain ferritic, MART, bainite and retained austenite phases. These phases have different structure and thus different hydrogen diffusion rate and different hydrogen solubility. Ferrite has a high hydrogen diffusion rate but low solubility due to open lattice structure. On the contrary, austenite exhibits a lower diffusion coefficient and higher solubility of hydrogen due to close-packed lattice. The diffusion coefficient of martensite is between those of ferrite and austenite because it is a closer-packed phase than ferrite. Even phase ratios may have a great influence on hydrogen permeability. Davies (1983) compared DP steels with different martensite contents and found out that there was no HE up to 10% volume fraction of martensite. The susceptibility to HE increased with the increasing martensite content from 10 to 30%, and it was independent of the martensite content above 30% of martensite. Rehrl et al. (2014a) investigated CP, DP and TM steels and concluded that the TM steel had the highest diffusion coefficient, while the CP steel had the lowest one. Hadžipašić et al. (2011a) stated that TRIP steels had greater susceptibility to HE compared to DP steels because of a favorable microstructure with residual austenite instead of martensite and less inclusions of the latter materials. Yang et al. (2016) also confirmed that retained austenite retarded diffusion and increased the solubility of hydrogen in AHSS. Therefore, there is a need to study the hydrogen content and distribution as a function of AHSS type and phase ratio.

4.1 Introduction

Characterization and evaluation of HE susceptibility of different steel grades is usually carried out with the help of fracture tests. A slow strain rate test (SSRT) and a constant load test (CLT) are appropriate in most cases. Hydrogen-free and hydrogen-charged specimens are subjected to fracture tests, and comparison of the obtained results reveals the influence of hydrogen on yield strength, tensile strength, elongation and fracture character. Difference in stress-strain curves of uncharged and charged specimens is obvious in Figure 1 illustrating the possible results of SSRT. Decrease in yield strength, elongation at fracture and tensile strength at failure can be observed for hydrogen charged specimens.

Figure 1:

Stress-strain curves of uncharged and electrochemically hydrogen charged AHSS specimens.

Available techniques and their advantages and disadvantages are listed below.

4.2 Constant load test

In CLT, the stress applied to pre-charged specimens with uniform hydrogen content causes hydrogen diffusion and accumulation in regions with the highest stress levels (Hagihara, 2012). Specimens subjected to cathodic charging are often kept for a certain period of time at ambient temperature to get homogeneous hydrogen distribution.

This method allows to estimate the critical hydrogen content leading to a failure, H_c, which is defined as the maximum hydrogen content in intact specimens under each applied stress level (Chida et al., 2016). Takagi et al. (2005) used CLT to obtain basic HE susceptibility data for quenched and tempered AHSS. They conducted cathodic pre-charging in NaCl water solution with NH₄SCN followed by Cd coating for prevention of hydrogen degassing and the hydrogen content homogenizing treatment. The hydrogen content after fracture was determined by thermal desorption analysis. Chida et al. (2016) investigated HE susceptibility of a low-alloy high-strength steel in a similar manner but using a Zn coating instead of the Cd one. It was shown that fracture stress decreased with increasing hydrogen content. Scharf et al. (2016) used CLT with in situ hydrogen charging for determination of HE resistance of pre-damaged zinc electrogalvanized DP1000 specimens. Hydrogen charging was conducted using a “corrosion cup” with NaCl solution, and the constant load was applied by using a proof ring. The effect of material pre-damaging by punching, laser cutting and milling on HE was investigated. Only the specimens with shear cut edges had microcracks; the specimens with milled edges showed no failure.

4.3 Slow strain rate test

SSRT is conducted under an extremely low strain rate usually of the order 10⁻⁶ s⁻¹, which is about four orders of magnitude lower than that used in standard tensile tests. The applied stress is increased gradually until fracture. It promotes stress-induced diffusion of hydrogen resulting in hydrogen accumulation according to the stress concentration in a specimen. Rehrl et al. (2014b) confirmed that the strain rate significantly influences deterioration of mechanical properties of AHSS in the presence of hydrogen. No deterioration of mechanical properties by hydrogen was observed at a high strain rate of 20 s⁻¹. At a slow strain rate of 10⁻⁵ s⁻¹, a decrease of tensile strength and elongation at fracture by about 25% and 90%, respectively, was noted comparing hydrogen-charged and hydrogen-free specimens. Thus, AHSS shows strain-rate sensitivity when containing hydrogen. Due to the possibility of hydrogen diffusion during SSRT, the relationship between the hydrogen content and fracture stress evaluated by SSRT should coincide with results obtained by means of CLT (Chida et al., 2016). However, SSRT significantly reduces the test period and provides less scatter in results in comparison with CLT. The SSRT technique also avoids the problem of specifying a time of test, because it always ends in material failure. The test can be performed on smooth or notched specimens. The susceptibility to cracking is evaluated from the stress-strain curves, and the test is usually followed by a fractographic analysis by scanning electron microscopy (SEM) and evaluation of hydrogen content by thermal desorption.

Wang et al. (2005, 2006, 2007) studied HE susceptibility of AHSS by means of SSRT. Smooth and notched specimens after cathodic hydrogen charging, Cd plating and hydrogen content homogenization treatment were used. They confirmed experimentally and by using numerical calculations that the fracture stress decreased with increasing diffusible content of hydrogen. Xu et al. (2012) used in situ hydrogen charging SSRT for evaluation of HE susceptibility of a Cr-Mo steel. They determined the safe hydrogen concentration to 0.064 ppm for temper embrittled steel and claimed that in situ hydrogen loading was an effective charging method.

4.4 Conventional strain rate test

In the case of a conventional strain rate test (CSRT), the hydrogen content in a specimen corresponds to the accumulated hydrogen concentration in specimens during CLT and SSRT. The test is performed under a high conventional strain rate resulting in negligible hydrogen diffusion. Thus, the hydrogen content at the crack site is equivalent to the average hydrogen content in the specimen. The principal advantage of this technique is a very short test time. On the other hand, highly uniform hydrogen charging prior to the test is required. Chida et al. (2016) studied susceptibility to HE of low alloy steels by using SSRT, CLT and CSRT. SSRT and CLT gave a similar relationship between the diffusible hydrogen content and nominal fracture stress because both of these methods allow for hydrogen diffusion during the tests. CSRT showed a lower susceptibility to HE even at the same local hydrogen concentration probably due to limited interactions between hydrogen and dislocations.

4.5 Stepwise load test

Takagi et al. (2005) proposed another evaluation method for HE susceptibility. In the stepwise load test, phases of increasing stress and stress holding phases are repeated until specimen fracture. The hydrogen distribution almost reaches equilibrium during the holding time. The authors used initial loading of 702 MPa during 12 h as low stress conditions and then applied stress increments of 14 MPa with the holding time of 2 h. The advantage of this technique is independence from dimensions of a specimen and from stress conditions. Thus, results of the test can be used for components with different shapes, size and stress concentration factor.

4.6 Comparison of the techniques and applicability for testing in atmospheric conditions

Advantages and shortcomings of the four techniques described above are shown in Table 2.

Since HE as a consequence of atmospheric corrosion is of great interest currently, there is a need to consider if techniques for tensile testing described above are appropriate for experiments under atmospheric conditions.

Two types of fracture tests have already been successfully used for HE susceptibility studies under atmospheric corrosion. Akiyama et al. (2013) conducted SSRT after a cyclic corrosion test and outdoor exposure of MART steels. They measured changes in notch tensile strength of steels as a function of an increasing number of corrosion cycles and diffusible hydrogen content. They obtained a power law relationship between notch tensile strength and hydrogen content. Using SSRT after outdoor exposures at two sites in Japan allowed taking into account the influence of actual environmental conditions. Akiyama et al. (2011b) also performed CLT after exposing steel specimens under similar conditions. When the hydrogen content from the environment was higher than the critical hydrogen content for delayed fracture, a fracture occurred, and it did not depend on the relative humidity in CLT chamber. However, a number of corrosion cycles influenced the delayed fracture. Relative humidity leading to the delayed fracture was lower when a number of corrosion cycles was higher, suggesting certain influence of exposure duration on the susceptibility to delayed fracture.

CSRT and stepwise load tests were not yet used in atmospheric HE research. In the case of CSRT, a uniform hydrogen pre-charging is required, and hydrogen content after pre-charging should correspond to the accumulated hydrogen during SSRT. These requirements cannot be implemented under field exposure or by using a cyclic corrosion test. Stepwise load testing could be an interesting option. Equilibrium hydrogen distribution can be reached during holding phases simulating conditions with load and unload periods more realistically than SSRT.

6.1 Electrochemical permeation technique

A widely used method for evaluating diffusion of hydrogen through metals is the EPT based on the Devanathan and Stachurski work (1962). The apparatus shown schematically in Figure 2 involves two electrochemical cells separated by a thin metal specimen which acts as a working electrode. A current is applied in the first cell, and the specimen is cathodically polarized to promote reduction of hydrogen protons from solution. A part of evolved hydrogen leaves the specimen surface as gas bubbles, but a fraction of hydrogen atoms is absorbed by the specimen, diffuses through the metal and is oxidized on the opposite surface. Hydrogen diffusion through the metal may be assessed by the change in oxidation current between the anodically polarized specimen and a counterelectrode in the second cell. The anodic side of the specimen is often covered with a palladium coating to prevent oxidation and formation of a passive layer which may act as a barrier to hydrogen permeation (Manolatos, 1995). EPT can be used to measure the hydrogen diffusion coefficient, sub-surface hydrogen concentration and steady state permeation rate. According to the ISO 17801 standard (EN ISO, 2014), the diffusion coefficient can be calculated from permeation curves using the time-lag method (Eq. 14) or breakthrough time method (Eq. 15) in agreement with Fick’s second law.

Figure 2:

Electrochemical permeation technique setup.

In Eq. (14), L (cm) is the specimen thickness and t_l (s) is the elapsed time to achieve the value J(t)/J_ss = 0.63 where J(t) in mol m⁻² s⁻¹ is the time-dependent hydrogen permeation flux and J_ss is the hydrogen permeation flux at the steady state.

Using the breakthrough time method, the diffusion coefficient is evaluated according to

where t_b in seconds is the time measured by extrapolating the linear portion of the rising permeation current transient or the time elapsed from the start of the test to first detection of hydrogen on the opposite side of the specimen.

Liu et al. (2016b) utilized EPT for DP and quenched and tempered AHSS. They evaluated the reversible hydrogen traps density using the Oriani-Dong model (Oriani, 1970; Dong et al., 2009) and the permeation curve method and concluded that both methods are useful. Complete transients and partial transients were measured, and significant difference in density of reversible traps from complete decay and partial decays was observed, because a value from complete decay involved all reversible traps, while values from partial decays responded to reversible traps under particular charging conditions. Values of trap density were quite similar for the studied steels.

Hadžipašić et al. (2011a) compared diffusion parameters for DP and TRIP steels obtained from EPT. Reversible traps density was lower, and effective diffusion coefficient was higher in the DP compared with the TRIP. It was also concluded that the TRIP contained less irreversible traps than the DP based on a steeper permeation curve for the former material.

In their other paper, Hadžipašić et al. (2011b) studied changes in diffusion parameters from the first permeation transient to the subsequent ones for a DP steel. A steeper curve for the first transient indicated a small number of irreversible traps, while changes in diffusion parameters as a function of a number of transients suggested domination of reversible trapping sites in the structure.

Trapping effects in TRIP steels during different transients were observed by Sojka et al. (2016). The lowest diffusion coefficient was established for the first buildup transient due to trapping by both irreversible and reversible traps. During the decay, transient hydrogen detrapping influenced the diffusion coefficient, the value of which was between the first and the second buildup transient values. Fitting experimental curves with theoretical one gave the best results for the second buildup transient.

Pang et al. (2014) compared diffusion coefficients for DP and commercial interstitial-free steels. They observed a negative effect of high charging current density over 0.1 mA cm⁻². Increasing amount of hydrogen was released as gas, instead of entering the steel with increasing current density.

EPT can be applicable for testing in atmospheric conditions. Conventional EPT is used with two electrochemical cells. Akiyama et al. (2011a) applied a cell only at the exit side of a specimen for hydrogen detection, whereas the entry side was exposed in a corrosion chamber to investigate the effect of atmospheric conditions on the hydrogen entry behavior. Hydrogen permeation current density slightly increased at 70%–80% relative humidity (RH) and remarkably increased when the humidity reached 98% RH. This observation showed an enhancing effect of corrosion on hydrogen entry. Omura (2012) used a similar setup under actual atmospheric exposure. Hydrogen permeation coefficient was affected by daily temperature-humidity cycles, season and exposure site. Experiments in a corrosion chamber showed that hydrogen permeation coefficient was dependent on temperature, relative humidity and amount of sea salt on the specimen surface. Obtained results were interpreted based on the effect of the composition and thickness of an electrolyte film on steel surface, on corrosion rate and hydrogen permeation.

6.2 Thermal desorption spectroscopy

The TDS technique involves measurement of the hydrogen desorption flux under controlled temperature ramping conditions. Gas chromatography or mass spectroscopy is used for hydrogen quantification. The method can provide information about density and binding energy of trap sites. It is also used to measure critical hydrogen content in AHSS required for HE fracture. TDS can also be used to study hydrogen solubility and diffusivity after hydrogen pre-charging (Yagodzinskyy et al., 2011).

Kushida (2003) used TDS for monitoring the content of absorbed hydrogen in an AHSS during atmospheric exposure after weathering. The maximum hydrogen content was measured after 6 months of exposure, whereas the hydrogen permeation rate reached maximum immediately after start of weathering. The author suggested that the delay between the maxima of the permeation rate and hydrogen content was due to a slow diffusion process. He also noted that hydrogen entry was influenced by initial rust formation.

Akiyama et al. (2010, 2011a) applied TDS for AHSS subjected to a cyclic corrosion test to follow hydrogen absorbed under simulated atmospheric corrosion conditions. It was revealed that the diffusible hydrogen content increased in time, although the corrosion rate of the specimens was constant. That confirms the observations of Kushida that formation of corrosion products can enhance hydrogen entry.

Escobar et al. (2012) investigated hydrogen content in TRIP and DP AHSS after hydrogen charging. They also calculated activation energies of hydrogen traps for all peaks. The activation energies for low temperature peaks were pretty similar. In the case of the TRIP steel, a high temperature peak was also observed, which was assumed to occur owing to hydrogen released from retained austenite. Based on this observation, the authors concluded that retained austenite could incorporate hydrogen trapping sites.

Takai and Watanuki (2003) studied an AHSS by TDS and SSRT and showed that while low temperature peaks corresponded to harmful trapping sites in terms of HE, trapping sites associated with high temperature peaks are innocuous due to non-diffusibility of the trapped hydrogen at the laboratory temperature.

Nagumo et al. (2001) investigated a MART AHSS with TDS and supposed that vacancy clusters as deformation-induced defects could be responsible for HE susceptibility rather than high concentration of hydrogen itself. Formation and growth of these clusters promoted by hydrogen could primarily lead to failures.

TDS is often used for determination of critical and environmental hydrogen contents. Akiyama et al. (2010, 2011a, 2013) and Kushida (2003) successfully applied this technique after actual and simulated atmospheric exposures.

6.3 Hydrogen microprint and silver decoration techniques

Hydrogen microprint technique (HMT) was developed by Ovejero-García (1985). It utilizes a reduction effect of hydrogen atoms released from a specimen surface. A specimen is coated with a photographic emulsion containing silver bromide crystals. When hydrogen atoms are released from the specimen surface, they reduce silver ions in the crystals to metallic silver. After a certain time, a photographic solution (Na₂S₂O₃) is used to eliminate unreacted AgBr crystals, and hydrogen exit sites can be visualized as silver particles. The particles can be observed by SEM and analyzed with an energy dispersive X-ray spectrometer (Ichitani & Kanno, 2003). Ichitani et al. (2003) showed that nickel plating of steel prior to the emulsion application can enhance the detection efficiency of HMT. Relative humidity showed to be the parameter influencing hydrogen detection efficiency with Ni plating and should be about 80%. Ronevich et al. (2012) revealed that using the image-before-etch method does not lead to artifacts compared with results obtained from the traditional image-after-etch method. In the first method, a specimen is hydrogen-charged, observed with SEM after the emulsion application, etched and re-imaged. In this case, there was a more uniform contact between the emulsion and steel surface due to a lower topography influence. This seems to be particularly important for fine-grain steel microstructures such as the observed TRIP steel. There was no significant difference in observations of hydrogen distribution in ferritic steel. Luppo et al. (1999) used this technique to study the hydrogen effect in austenitic stainless steel welds containing delta ferrite phase. It was shown that ferrite-austenite interfaces act as hydrogen-trapping sites. Momotani et al. (2016) studied hydrogen accumulation in low-carbon MART steel during tensile tests and revealed that hydrogen preferentially diffused to austenite grain boundaries during SSRT, whereas it was distributed uniformly in the lath martensite at higher strain rates. Ishikawa et al. (2016) also observed high concentration of hydrogen on the lath boundary in both as-quenched and tempered MART steel by HMT.

Schober and Dieker (1983) described another variant of the Ag crystal decoration technique. Potassium-silver dicyanide solution prepared from KCN and AgNO₃ solutions was used instead of silver bromide emulsion. The Ag precipitation occurs from a homogeneous solution, and the resolution of the technique is thus higher. Akiyama and Matsuoka (2015) carried out hydrogen charging on the opposite side in order to observe the distribution of diffusing hydrogen. It was selectively present on slip lines for austenitic stainless steel and almost homogeneous for MART steel. Koyama et al. (2013) observed hydrogen localization in a quasi-cleavage-cracked region after deformation of steel to the fracture with hydrogen charging followed by silver decoration procedure. Nagashima et al. (2017) described observations of preferential desorption of hydrogen from martensite using Ag decoration after hydrogen charging.

Both techniques are used for determination of the diffusible hydrogen distribution in specimens. Experiments using HMT and Ag decoration after atmospheric corrosion can be conducted; however, there is a danger of quick hydrogen discharging. Thus, the measurement would need to be rapid and storage under low temperature required.

6.4 Tritium autoradiography technique

The tritium autoradiography technique, TARG, is based on the same principle. Hydrogen (tritium) atoms can be detected as silver particles. The advantage of this technique is the detection of tritium radio isotope of hydrogen, which remains in the vicinity of the surface, whereas HMT detects only hydrogen released from the specimen. However, the technique gives no information on the fast diffusing hydrogen atoms (Hanada et al., 2005).

The specimen is charged in tritiated aqueous solution, etched, covered with photographic emulsion and stored in liquid nitrogen. Beta rays radiating from tritium are visualized by photographic development and fixing. Ag particles distribution is observed by using SEM or TEM (Saito et al., 1998). The experiment can take a long time due to low irradiation intensity in liquid nitrogen. It can take 4 days (Le & Wilde, 1983), 28 days (Watakabe et al., 2012) or even 50 days (Hanada et al., 2005).

TARG was used for studying the trapped hydrogen distribution in steels by a few authors. Le and Wilde (1983) were first who had conducted TARG on carbon steels and demonstrated a hydrogen segregation on inclusions after tritium charging. Hanada et al. (2005) observed local hydrogen accumulation in cementite precipitates and at boundaries between the precipitates and ferrite matrix. Otsuka et al. (2005) found that a type of inclusion (MnS, Al₂O₃ and Cr carbides) influences hydrogen distribution in the ferrite matrix around the inclusions. Aoki et al. (1994) studied hydrogen distribution as a function of deformation conditions and reported uniform hydrogen distribution at laboratory temperature deformation and inhomogeneous distribution for deformation at −80°C.

TARG has not been conducted to study hydrogen distribution under atmospheric corrosion conditions. It might be performed by using charging in tritiated NaCl solution.

6.5 Secondary ion mass spectroscopy

Secondary ion mass spectroscopy (SIMS) is an alternative method for determination of hydrogen distribution. The specimen surface is scanned with primary ions, and ejected secondary ions are detected. For instance, Cs and Bi ions can be used as primary ions. These ions are delivered by a liquid metal ion gun. Gun with sputtering ions, as Ar ions, may be used in order to remove surface contaminants from the studied region of a specimen (Sobol et al., 2016a). SIMS enables detection of hydrogen not only on the surface as HMT but also in deeper layers. However, the results are not analytically precise due to the low specific weight of hydrogen.

Sobol et al. (2016a) conducted SIMS measurements with a cell mounted in a SIMS chamber to perform in situ deuterium permeation experiments. The measurement showed much faster deuterium permeation in ferrite than in austenite. The advantage of this approach is the possibility to follow diffusion of deuterium or hydrogen through particular grains and phase boundaries, while ex situ experiments allow only for measuring deuterium or hydrogen accumulated in a specimen. Using of deuterium instead of hydrogen for SIMS can be useful to avoid the effect of background-originated hydrogen from moisture or compounds in a SIMS chamber on obtained results (Silverstein et al., 2017).

SIMS is capable of three-dimensional hydrogen mapping (Koyama et al., 2017). It has a better depth resolution in comparison with TARG and requires much shorter time (Brass et al., 1996). On the other hand, detection of hydrogen in steel with high hydrogen diffusivity and low hydrogen content may be difficult. Brass et al. (1996) recommended performing SIMS at very low temperatures in order to avoid loss of hydrogen after hydrogen charging (Brass et al., 1996). Cryogenic SIMS was successfully used by Nishimoto et al. (2015) to study hydrogen distribution in duplex stainless steel.

SIMS is often used for hydrogen detection in ferritic and stainless steels. Awane et al. (2014) reported that distribution of the hydrogen intensity changed in time due to hydrogen diffusion in austenitic stainless steel. The hydrogen content in a sample decreased in time. Hryniewicz et al. (2011) compared hydrogen concentrations in stainless steel specimens after different surface treatments and concluded that parameters of electrolytic polishing affect it. Hydrogen concentrations were lower in electrolytically polished specimens than in as-received and abrasive polished ones. Saintier et al. (2011) observed a high hydrogen content around the crack tip in austenitic stainless steel after hydrogen charging and a fatigue crack propagation test. Sobol et al. (2016b) studied duplex stainless steel by SIMS after deuterium electrochemical charging and observed accumulation of deuterium around cracks, while no external load had been applied. The authors assumed that localized strain was introduced into steel by deuterium. Silverstein et al. (2017) showed locally high deuterium concentrations in deformed regions of ferrite revealing high deuterium content around cracks.

SIMS could be useful for determination of hydrogen content and distribution in steels after atmospheric exposure.

6.6 Nanoindentation

Nanoindentation technique is a useful method to study the mechanism of HE and interactions between hydrogen and dislocations by local mechanical testing. Indents are made using triboindenter. Measurements are usually performed within one grain to avoid influence of different grain orientation. Sufficient distance from grain boundaries to indents is needed to avoid any influence of grain boundaries (Barnoush et al., 2012).

Load-displacement (L-D) curves are measured. Hardness and elastic modulus are calculated from L-D curves using Oliver-Pharr method described in the original paper (Oliver & Pharr, 1992). Three stages can be distinguished on the L-D curve: (i) elastic loading, (ii) pop-in stage and (iii) continuous elasto-plastic deformation. An example of L-D curve is in Figure 3. The second stage, pop-in, is a discrete jump of the displacement which appears at the onset of plasticity. Decrease in pop-in load with increasing hydrogen concentration was observed (Barnoush et al., 2012; Gaspard et al., 2014). Gaspard et al. assumed that reduction of the pop-in load is a manifestation of homogeneous dislocation nucleation (HDN) promoted by hydrogen. Hydrogen can reduce interatomic bonding energy (HEDE mechanism) and stacking fault energy inducing decrease of activation energy of HDN (Barnoush & Vehoff, 2010). Barnoush et al. (2012) noticed not only the effect of hydrogen on pop-in load reduction but also on the elastoplastic part of L-D curve linking it with evaluation of dislocation mobility through the measurement. Hong et al. (2017) analyzed the effect of hydrogen on hardness and elastic modulus during deformation and concluded that this effect changed with deformation stages. In the initial stage of plastic deformation, hydrogen increased hardness and decreased elastic modulus, but the effect became weaker during further plastic deformation. The explanation was found in change in the dislocation mobility and change in the slip systems character. Dislocation nucleation takes place during plastic deformation, but mobility of dislocations decreases with growing dislocation density reducing hydrogen effect on elastic modulus. Weakening the hydrogen effect on hardness was related to prevalence of cross-slip systems instead of planar-slip systems. It was reported earlier that an increase of the measured hardness might be due to hydrogen-enhanced slip planarity (Nibur et al., 2006).

Figure 3:

Load-displacement curve with three stages of deformation.

Nanoindentation measurements can be done on hydrogen charged and uncharged specimens, but nanoindentation with in situ hydrogen charging can be the better choice. Experimental setup for in situ nanoindentation is shown schematically in Figure 4. During in situ charging, hydrogen concentration becomes uniform close to the surface within a very short time, while hydrogen discharging can occur relatively quickly if precharged samples are used (Barnoush et al., 2012; Tomatsu et al., 2016). Kheradmand et al. (2016) showed that the hardening effect of hydrogen increases as the negative cathodic polarization increases. They linked this observation with higher concentration of hydrogen entering into steel due to more negative potential.

Figure 4:

In situ nanoindentation setup.

Nanoindentation results can be linked not only to the HEDE mechanism but also to HELP. Lee et al. (2014) showed transition from hardening to softening effects induced by hydrogen using indenters with different sharpness. The softening effect was observed during tests with sharper indenters. It was associated with hydrogen’s elastic shielding of dislocation-defect interactions leading to high dislocation mobility, which is one of the mechanisms for explaining HELP. The authors stated that dislocation density increases with indenter sharpness, and a hardening to softening transition can be thus observed.

Tomatsu et al. (2016) tested the influence of strength rate of nanoindentation on hardening and softening processes. Hydrogen-induced hardening was observed at a high strain rate, and hydrogen-induced softening was noticed at a slow strain rate. Hydrogen softening at the slow strain rate was explained by longer time for hydrogen accumulation around dislocations and manifestation of the hydrogen’s elastic shield effect or the hydrogen-enhanced cross-lip.

Nanoindentation was used for AHSS (Ariza et al., 2016; Zhang et al., 2016; Hashemi et al., 2017) but only to study mechanical properties of particular phases. Publications reporting on a link between hydrogen and nanoindentation for AHSS were not found. The technique could be theoretically used under atmospheric corrosion conditions. For instance, one side of a specimen may be exposed under atmospheric corrosion condition, and changes in hardness may be measured on the opposite side.

6.7 Scanning Kelvin Probe and Scanning Kelvin Probe Force Microscopy

It would be very valuable to find an answer to the technically relevant question on how much the process of HE is affected by the microstructure of AHSS. Because the previously described techniques are limited in sensitivity and spatial resolution, another technique is needed.

It was recently found out that the scanning Kelvin probe (SKP) method was able to provide highly relevant spatial information on the release of hydrogen from iron, steel, palladium and other materials. SKP is a non-contact, non-destructive device for spatially resolved determination of work function between a conducting specimen and a vibrating tip. It has been shown by Kelvin that there is a potential difference between two conducting materials in electrical contact. This contact potential difference depends on work functions (Fermi levels) of the materials, e.g. the Kelvin probe tip and specimen surface. Electrical contact between the probe and specimen makes the charge flow from the material with the lower work function to the one with higher work function. It leads to levelling of Fermi levels in the probe tip and the specimen resulting in a potential difference. The Kelvin probe vibrates above the specimen surface creating a vibrating capacitor. The distance between the probe and specimen changes inducing a change of capacitance and flow of the alternating current. In the conventional nulling technique, the backing voltage is applied until the alternating current is 0, and at this point the external potential is equal to the contact potential difference. The technique allows for scanning measurements at a constant height by maintaining constant capacity between the probe tip and specimen. It can be done by applying small harmonically oscillating ac bias in addition to the externally applied dc bias (Rohwerder & Turcu, 2007). Then, the measurement of the contact potential difference becomes independent of the capacity and distance between the tip and specimen.

It was shown (Rohwerder & Benndorf, 1994; Halas et al., 2004) that hydrogen may cause a change in the surface work function detectable by SKP. SKP can be used on Pd-coated steel where the concentration of hydrogen in a palladium layer influences surface potential of palladium according to the Nernst equation (Evers & Rohwerder, 2012). Its work function shows logarithmical dependence on the hydrogen amount absorbed in the metal. Concentrations as low as 0.01 ppm of atomic hydrogen dissolved in a metal are detectable. Furthermore, Pd acts as a very efficient drain for hydrogen, and even hydrogen from the trapping sites can be accumulated in a Pd layer; differences in the distribution of trap sites can be directly detected.

Alternatively, SKP measurements can be carried out on bare steel samples. In this case, hydrogen is detected by reducing the surface oxide film. Hydrogen reduces some part of Fe³⁺ species and lowers the contact potential difference because the ratio of Fe³⁺ and Fe²⁺ species determines work function of the specimen (Evers et al., 2013a; Williams et al., 2013). Bare steel samples are used for qualitative detection of hydrogen where lower contact potential difference spots appear due to hydrogen release and the contact potential difference of hydrogen-free surrounding areas remain unaffected.

It was revealed later that atomic force microscopy (AFM) in the SKP mode (scanning Kelvin probe force microscopy, SKPFM) can provide considerably better resolution for determination of hydrogen distribution. SKPFM is based on a two-pass measurement. In the first pass, the topography is acquired from mechanical excitation of the cantilever. In the second pass, the contact potential difference is measured from electrical excitation of the cantilever applying voltage to the tip. SKPFM maps the contact potential difference between the scanning tip and the surface with a high resolution of 1 mV. Thus, SKPFM allows for detection of properties of particular phases, and it can measure differences in the hydrogen permeation through different grains or along grain boundaries.

Li et al. (2012) observed hydrogen-induced changes of a contact potential difference map of ferrite and austenite in duplex stainless steel by means of SKPFM. Some low-potential areas were found at ferrite/austenite boundaries or inside the ferrite phases after cathodic hydrogen charging. These areas were considered as the pitting nucleation sites induced by hydrogen. They also observed time dependence of work function in austenite and ferrite after hydrogen charging corresponding with hydrogen behavior in these phases. Work function decreased after some time due to collection of hydrogen in the surface, then it was stable indicating predominance of aggregated over escaped hydrogen and after that increasing work function meant more escaped hydrogen than aggregated one.

Wang et al. (2013) studied contact potential difference maps before and after hydrogenation of maraging steel. They showed that the work function decreased due to the presence of hydrogen, and it varied with time due to hydrogen diffusion and release.

Senöz et al. (2011) incorporated a miniature cell for on-line hydrogen loading during SKP and SKPFM measurements on a Pd foil and thin sheet of duplex stainless steel. A specimen was fitted onto an opening at the top side of the cell. The bottom side of the specimen was in contact with electrolyte while the opposite side was scanned by SKP or SKPFM. The authors observed differences in permeation of austenite and ferrite phases. Due to the fact that a thin electrolyte layer on the specimen surface is necessary for SKPFM measurements (Cook et al., 2012), it can be interpreted as a need for wet atmosphere in the SKPFM chamber. Hence, it was proposed by Senöz et al. (2011) that measurements even on the dry surface could bring reasonable information about the presence of hydrogen in specimens because ultra-thin layer forms even under such conditions.

Schimo et al. (2015) also carried out in situ hydrogen loading of Pd foil specimens during SKP measurements with a miniature electrochemical cell. The cell allowed for hydrogen charging without surface blockage by evolving gas bubbles due to convection of the electrolyte.

Evers et al. (2013b) used in situ hydrogen loading during SKP measurements, in which the probe tip remained in one position, as well as conventional electrochemical hydrogen charging prior to SKP and SKPFM measurements. They investigated local corrosion spots on the exit side of a duplex stainless steel specimen with a zinc layer with an artificial defect on the opposite side and observed that the contrast in potential maps due to absorbed hydrogen smeared out with time after transferring from charging cell to the SKP or SKPFM chamber. Thus, the in situ charging seems to be more appropriate for studies of the hydrogen effect on steel properties.

However, the investigations described above did not focus on the hydrogen entry to AHSS as a consequence of atmospheric corrosion. It has been mentioned above that it might principally lead to unexpected failures and has to be studied carefully. There is a lack of investigation devoted to hydrogen caused by atmospheric corrosion in steels by means of techniques with high spatial resolution, apart from the studies of Nazarov et al. (2015, 2016) by SKP. In the earlier work (Nazarov et al., 2015), the authors studied hydrogenation of iron, carbon steel and zinc-coated steel under conditions of atmospheric corrosion. They applied a drop of NaCl solution on one side of a specimen, exposed it in humid air and then measured potential profiles on the opposite site by SKP. The contact potential difference decreased with increasing exposure time in humid air. In the case of the zinc-coated steel, atmospheric corrosion of zinc and cathodic polarization of steel in the coating defect caused formation of a round region with a relatively small potential drop. Using in situ hydrogen charging (Nazarov et al., 2016) during SKP measurements was successfully tested for an AHSS and pure iron. One side of the specimen was contaminated with NaCl droplet and exposed in the humid air, while another side was measured by SKP. Hydrogen generated under such corrosion conditions led to decreasing potential for the pure iron and the AHSS specimen s by 200 mV and 300 mV, respectively. Separation of anodic and cathodic electrochemical reaction areas and spreading of the cathodic area with time of exposure were observed.

While applicability of SKP for investigation of changes in contact potential differences caused by hydrogen absorbed due to atmospheric corrosion was confirmed by Nazarov et al. (2015, 2016), SKPFM could allow to conduct similar experiments with better resolution and probably to study the effect of microstructure constituents on changes in contact potential difference due to the hydrogen.

6.8 Comparison of the techniques

Advantages and disadvantages of the above discussed techniques are summarized in Table 3. Information on detection limits and spatial resolution are given in Table 4.