1. INTRODUCTION

Engineering practice is constantly faced with the problems of hydrogen-induced cracking (HIC), hydrogen embrittlement, stress-corrosion cracking, and hydrogen damage. This is explained by the fact that hydrogen is widely spread in nature, either in a gaseous state or in the form of chemical compounds, water being the most widespread. For example, the oil and gas industry deals with these phenomena in all types of corrosion [1].

As a rule, an increase in the strength of metals and alloys is accompanied by a decrease in plasticity. Unlike other alloy components with the maximum permissible concentration of hundredths and thousandths of a percent, hydrogen begins to influence the properties of some metals at the mass concentration of one hundred thousandths of a percent.

Therefore, there is a special sensitivity of modern high-strength alloys with iron, aluminum, magnesium or nickel matrices to low concentrations of hydrogen. For Bessemer steels, the maximum permissible mass concentration of hydrogen is about 4 ppm, while, for modern ultrahigh-strength steels, it is about 30 times less.

Even if such low hydrogen concentrations are achieved in the production of alloys, the subsequent processing and operation of metal parts will increase the concentration of hydrogen and its significant influence on the mechanical properties of metals and alloys, which are strength, ductility, fracture toughness, and crack resistance [2].

The problem of the interaction of metals with hydrogen used as an energy source in various hydrogen energy projects is worthy of notice. Transportation of mixtures of natural gas and hydrogen [3–7] leads to a decrease in the ductility, fracture toughness, and crack resistance of steel walls of pipelines [8–11], to say nothing of the transportation of pure hydrogen gas [9, 12].

Thus, the influence of hydrogen on the properties of materials cannot be left disregarded without serious consequences. Solving engineering problems requires that the effect of hydrogen on the strength of materials be evaluated at the design stage of structures and during their operation.

Tests on hydrogen-charged notched corset metal specimens for strength, fatigue strength, ductility, and fracture toughness are often used in engineering practice and in research. Such tests allow the quality control of metals and the selection of metals that are resistant to hydrogen embrittlement and hydrogen-induced cracking.

Hydrogen charging is a standard procedure for HIC testing of steels for pipelines and other elements of machines [13–16]. In this case, use is usually made of nongaseous methods of hydrogen saturation [14, 16], which always lead to the formation of a skin layer for the time recommended by the saturation standards.

High inhomogeneity of distribution of hydrogen concentrations after the standard saturation procedure was reported in [17–20]. This phenomenon was referred to as the skin effect [17, 19]. Due to this effect, the hydrogen concentration in the surface layer with the approximately one grain thickness (10–100 μm) can exceed the concentration inside the specimen by tens of times. Modeling of this phenomenon has so far been carried out without taking into account the influence of the skin effect on the fracture of metal specimens [21].

The skin layer is sometimes explained by the very slow diffusion of hydrogen [22, 23]. However, previously we discussed in detail the experimental results that do not confirm this assumption [19]. In any case, a large increase in the hydrogen concentration in the boundary layer of metals is of fundamental importance to the test results.

The effect of the skin layer on the results of mechanical tests on hydrogen-charged specimens also has not been investigated. It follows from practice that standard methods for hydrogen saturation lead to a strong decrease in plasticity (relative to residual elongation) and strength of the entire specimen. At the same time, fracture often has a dual pattern: the fracture surface of the specimen has areas of both hydrogen brittleness and ductile fracture [24, 25].

Correct consideration for the influence of the skin layer is very important when modeling hydrogen embrittlement. This is due to the need of understanding the leading mechanisms of fracture and predicting the destruction of machine parts operating in aggressive media.

Crack resistance is critically reduced by hydrogen; therefore, the two most popular and well-known calculation models of hydrogen brittleness (hydrogen-enhanced decohesion (HEDE) [26] and hydrogen-enhanced localized plasticity (HELP) [27, 28]) are based on the approaches of crack theory and consider the nucleation and sink of dislocations at the crack tip as the physical mechanism of fracture. The initially developed models significantly differed from each other. Thus, the HEDE model considered brittle fracture as a result of the development of hydrogen embrittlement without plastic deformation, while, on the contrary, the HELP model interpreted the effect of hydrogen dissolved in a metal as the action of a chemical constituent, which reduces the energy required for dislocation nucleation. By the HELP mechanism, hydrogen is concentrated under the action of internal stresses at the crack tip, where metal “softening” or localized plasticity occurs.

These models were verified in numerous experiments [29–35] (see also reviews [24, 25, 36, 37]). The bulk of the experimental results were obtained on specimens artificially charged with hydrogen in the electrolyte solution, as a rule, using cathodic charging.

The investigations showed that the fracture surfaces of experimental specimens have not only areas of hydrogen brittleness but also zones of ordinary fracture without pronounced hydrogen signs. The idea of the hybrid HELP + HEDE mechanism was proposed for explaining this dual fracture pattern [38–41] (a review is done in [24]).

The HELP and HEDE models can be formally combined within the Abaqus package. The HELP method can be used to describe plastic flow at low local hydrogen concentrations, and the HEDE method can be employed to analyze the crack propagation from the groove or notch in the specimen at local hydrogen concentrations above the switching threshold of the models. This was done for the two-dimensional case in [42]. The combined model was applied to determine the fatigue crack growth rate with the parameters given in advance. The authors pointed to the existing problems of determination of a large number of parameters and stated only a qualitative agreement between the simulation results and the experimental findings.

The combined model with additional consideration for the void formation under loading was used to simulate the crack growth in a notched viscoelastic material in [43]. In this case, the transport of hydrogen is modeled by the simplified method [44], and the cohesive law is independent of hydrogen. The problem of crack growth in high-strength chromium steel 690 around an inclusion was studied in [45–47]. The HELP model in the form of [44] is supplemented by the HEDE cohesive law.

In all of the above works, the authors assume the uniform hydrogen concentration in metal specimens after hydrogen charging. All changes in the hydrogen concentration are associated with its redistribution under an external load or during the crack growth. As a rule, a specific value of the initial hydrogen concentration results from fitting the model parameters to the experimental data. Therefore, it varies within a wide range from 0.0005 [48] to 1.5 ppm in various papers [49].

By comparing the initial HELP and HEDE approaches, we can conclude that the HEDE mechanism corresponds to the concept of brittle fracture, which is observed in steels at high hydrogen concentrations [42]. Therefore, the use of the HEDE mechanism in describing the consequences of the skin effect for the crack growth should provide an adequate model for fracture of specimens artificially saturated with hydrogen in electrolyte solutions or at cathodic charging.

2. ORIANI DECOHESION MODEL OF BRITTLE FRACTURE

The process of hydrogen transport inside a solid is traditionally considered within the HEDE model as a diffusion process. It is described within Fick’s law with the additional term that takes into account the thermodynamic (or chemical) potential V [50]:

$${{\partial C} \over {\partial t}} = \nabla \cdot \left[ {D(T,\,{\bf{r}})\left( {\nabla C + {{C\nabla V} \over {RT}}} \right)} \right].$$
((1))

Here C is the hydrogen concentration, D is the diffusion coefficient, R is the gas constant, T is the absolute temperature, r is the position vector, t is the time, and \(\nabla \) is the Hamilton nabla operator. Taking into account the action of mechanical stresses, this equation can be rewritten as

$${{\partial C} \over {\partial t}} = D[T]{\nabla ^2}C - D[T]{{{V_{\rm{H}}}\nabla C \cdot \nabla \sigma } \over {RT}} - D[T]{{C{V_{\rm{H}}}{\nabla ^2}\sigma } \over {RT}},$$
((2))

where σ is the primary stress (spherical part of the stress tensor), and VH is the partial molar volume of hydrogen.

Then, the fracture criterion associated with the magnitude of crack opening comes into operation within the HEDE mechanism. According to Gorsky’s law [51], hydrogen is concentrated in the region of the maximum stress tensor (i.e. spherical part of the stress tensor) and reduces the adhesion of the crack faces which leads to decohesion [26].

These changes in cohesive forces between grains are described through the parameter of coverage of the free surface of a crack with hydrogen atoms θ, which can be represented as [52]

$$\theta = {C \over {C + \exp \;[ - {{\Delta {g_{\rm{H}}}} \mathord{\left/ {\vphantom {{\Delta {g_{\rm{H}}}} {(RT)}}} \right. \kern-\nulldelimiterspace} {(RT)}}]}},$$
((3))

where ΔgH is the Gibbs free energy difference between hydrogen adsorbed inside the crystal lattice and at the crystal boundary (taken from the experiment). Formula (3) was obtained by Serebrinsky [52] by comparing the parameters of hydrogen trapping in the metal based on the McLean ratio [53].

A value of the parameter θ determines the change in the specific energy of the free surface γ(θ), which depends on the surface sorption of hydrogen [52]:

$$\gamma (\theta ) = (1 - 1.0467\theta + 0.1687{\theta ^2})\gamma (0).$$
((4))

This dependence was obtained by approximating the graph for the case of deformation of a pure Fe single-crystal saturated with hydrogen in the (110) crystallographic direction.

From the energy identity 2γ(θ) = σzc(θ)δc(θ), where σzc(θ) is the maximum cohesive stress normal to the crack edges, and δc(θ) is the maximum possible crack-opening displacement without breaking bonds (decohesion), the law of hydrogen-induced degradation has the form [52]

$${\sigma _z}(\theta ) = (1 - 1.0467\theta + 0.1687{\theta ^2}){\sigma _z}(0)$$
((5))

under the assumption of the weak dependence of the δc(θ) value on the parameter θ.

In this paper, we investigate the applicability of the HEDE model to the description of hydrogen-induced fracture of a corset specimen with consideration for the skin effect.

3. INITIAL AND BOUNDARY CONDITIONS, MODEL PARAMETERS, AND CALCULATION PROCEDURE

We simulate uniaxial loading of a cylindrical specimen (Fig. 1) with the diameter d = 10 mm, which has a stress concentrator in the form of a semicircular groove with the radius r = 0.5 mm.

Fig. 1.
figure 1

Cylindrical specimen.

Full size image

The problem is solved in the two-dimensional axisymmetric formulation. By symmetry, only a quarter of the cylinder is considered and a rectilinear structured grid consisting of triangles is generated. Thus, we model a sector of the longitudinal section of the specimen (Fig. 2), with the Oy axis along the left side of the computational domain, and the Ox axis along the lower side. For a smoother distribution of the calculated values and greater accuracy of calculations, partitioning into finite volumes of two types (with different directions of the diagonal of a structural element) is carried out.

Fig. 2.
figure 2

Finite volume grid, boundary conditions (color online).

Full size image

Specimens under study are made of high-strength steel PSB1080 with the following physical properties: ultimate strength σB = 1498 MPa, yield stress σy = 1276 MPa, mass density ρ = 7800 kg/m3, and shear modulus G = 79.3 GPa. In addition, the diffusion coefficient D = 2.5 × 10–11 m2/s, the partial molar volume of hydrogen VH = 2 × 10–6 m3/mol [54], the Gibbs free energy difference for hydrogen adsorbed inside the crystal lattice and at the crystal boundary ΔgH = 30 kJ/mol [54], and the absolute temperature T = 298 K.

The initial distribution of hydrogen over the specimen is found as follows: the background uniform concentration \({\tilde c_0}\) = 0.1 ppm over the entire domain and the concentration in the surface layer with the one element thickness dH = 20 μm, c0 = 10 ppm. The initial concentration characterizes the experimentally observed hydrogen content in the hydrogen-charged specimens, which results from the skin effect (the enlarged fragment in Fig. 2).

By symmetry, displacements along Ox vanish for the left side and displacements along Oy vanish for the lower side. The tensile stress σ = 450 MPa is applied to the upper side of the domain and acts along axis Oy.

We investigate the uncoupled problem by the iterative method, as suggested by the original Oriani model. Based on the stress-strain characteristics of a body, the diffusion problem of hydrogen redistribution over the specimen is solved. Then, the change in cohesive stresses (and consequently in the crack configuration) is calculated in terms of the hydrogen content in the domain and particularly near the crack tip. The hydrogen distribution has no direct effect on the stress state of a specimen.

The calculation procedure consists of two successive stages: (1) solving the mechanical problem of the stress-strain state of a specimen, and (2) analyzing the diffusion problem with the simultaneous calculation of cohesive stresses σz.

At the first stage, we obtain the numerical solution of the boundary-value problem of linear elasticity by means of the relaxation method proposed by Wilkins [55, 56]. It relies on the fact that the stationary problem is solved as the nonstationary problem, which is brought to the stationary mode.

The stress-strain state of the specimen obtained at this stage is transferred to the next stage. It includes the solution of the system of the foregoing equations (2), (3), (5):

((6))

The numerical solution of Eq. (2) is also carried out based on the relaxation method by Wilkins [55, 56]. The stress σz(0) is taken to be 4.5σy according to [57].

The criterion of fracture, i.e. breaking of bonds between the atomic planes in the material, is fulfilled when the elastic stress obtained from the solution of the static problem exceeds the cohesive stress. As soon as this condition is fulfilled, the displacement constraint along axis Oy is removed at the grid node. This node is able to break away from the bottom face or, on account of symmetry, from the opposite node and to move upward under the action of tensile loads.

To monitor parameter changes during the problem solution and to reduce simulation errors, we use a well-tried approach [58–60]. To implement the approach, we developed a program code written in the Microsoft Visual Studio environment in C++ (VS), which allows obtaining a numerical solution to problems on the stress-strain state of a hydrogen-charged body within the finite volume method. Thus, we program the crack initiation and its propagation to a grid space. Thereafter a static problem is solved for the next crack propagation step, by repeating the calculation procedure.

4. SIMULATION RESULTS

The bond at the first grid node is broken instantly at the very first step of time integration due to high stresses acting near the stress concentrator and a significant hydrogen content in the surface layer. The field of distribution of the elastic stress component acting along the vertical axis near the groove at this time instant is shown in Fig. 3.

Fig. 3.
figure 3

Field of elastic stresses σy at the decohesion moment at the first node (color online).

Full size image

Maximum tensile stresses are seen to arise near the concentrator. The stress value for the first node along the crack propagation line is 1370.88 MPa.

Changes in the hydrogen concentration at this stage are insignificant. It decreases to C = 9.99967 ppm at the second node, and increases from the background value C0 = 0.1 ppm to C = 1.0028 ppm at the third node.

A similar situation occurs for the second node along the crack propagation line. At the initial stages, the hydrogen content in the surface layer is so high that a crack initiates almost instantly by the HEDE mechanism.

The calculations for the third node show that the solution of the diffusion problem takes some time for hydrogen redistribution until it arrives at the actual concentrator, i.e. the supposed place of debonding between the atomic planes of the material. The calculated time of this process is t = 9.17 s, and the critical hydrogen concentration is C = 0.386 ppm. The distribution of the elastic stress component at the moment of fulfillment of the fracture criterion at the third node is shown in Fig. 4.

Fig. 4.
figure 4

Field of elastic stresses σy at the decohesion moment at the third node (color online).

Full size image

The tendency towards increasing the time interval between the fulfillment of the fracture criterion for two successive nodes persists at subsequent integration steps. The corresponding dependence is shown in Fig. 5.

Fig. 5.
figure 5

Time between the fulfillment of the fracture criterion at the adjacent nodes.

Full size image

Starting from the third node, diffusion takes more and more time for hydrogen redistribution and accumulation at the potential fracture site of the material. At the seventh node, which corresponds to the distance 0.551 mm from the groove edge, this time interval (from the moment the sixth node was detached) amounts to 122.85 s, and the critical concentration is C = 0.213 ppm. The resulting elastic stress field is shown in Fig. 6.

Fig. 6.
figure 6

Field of elastic stresses σy at the decohesion moment at the seventh node (color online).

Full size image

Thereafter, as seen from Fig. 5, the time between the fulfillment of the fracture criterion at the adjacent nodes begins to decrease gradually. This is due to an increase in the level of stresses acting near the concentrator. Due to a lower hydrogen content in the region, cohesive stresses become lower than elastic ones, which leads to the propagation of a crack to a grid space according to the established algorithm.

This process is observed up to the thirteenth node, which corresponds to the distance 1.102 mm from the specimen surface. The field of elastic stress distribution at this time instant is shown in Fig. 7. In this case, the critical hydrogen concentration, at which the fracture criterion is fulfilled, is C = 0.113 ppm.

Fig. 7.
figure 7

Field of elastic stresses σy when a crack is halfway through the specimen thickness (color online).

Full size image

Thereafter elastic stresses acting along the lower side of the domain become so high that the background concentration in the material is sufficient to reduce the level of cohesive stresses (according to Serebrinsky’s formula (5)). Therefore, when solving the diffusion problem, the fracture criterion starts to fulfill already at the first time step and the hydrogen content at the node decreases progressively. For example, it is C = 0.103 ppm at the fourteenth node (1.194 mm from the groove surface) and C = 0.1005 ppm at the fifteenth node (1.286 mm from the edge).

Figure 8 shows the hydrogen concentration distribution along the specimen radius at different time instants: from the initial state to the instant when a crack passes a quarter of the specimen thickness.

Fig. 8.
figure 8

Hydrogen concentration versus distance along the specimen radius (color online).

Full size image

Under the action of mechanical stresses, hydrogen is gradually distributed and transferred from the specimen surface to the material interior. However, one can see that it does not diffuse deeper than 1.25 mm from the surface. Further fracture of the specimen occurs already at the initial background hydrogen concentration due to high stresses acting in the region.

5. DISCUSSION

By using the HEDE model with consideration for the skin effect, we obtained the results on brittle fracture in hydrogen-charged corset specimens. The considered parameters of the specimens and loading conditions are common for hydrogen brittleness tests [61–65].

The results obtained using the HEDE model were compared with the experimental data only in terms of the incubation time of crack growth. A universal means of fitting to the experimental data is the initial hydrogen concentration, which is considered by all authors to be uniform. Such fitting gives its wide scatter in different papers. Sometimes it differs by several orders of magnitude from the experimental average hydrogen concentration. We took experimentally measured values of concentration after 96 h of hydrogen charging of specimens [17–20].

We found that the incubation time of the macrocrack growth is determined not by the time of increase of hydrogen concentration at the crack tip from the uniform one, but by the diffusion time of the increased hydrogen concentration from the skin layer to the crack tip. A microcrack in the skin layer is formed immediately after the application of a mechanical load. This also agrees with the numerous experimental data on the formation of microscopic quasi-cleavage cracks on the surface of hydrogen-charged specimens under deformation.

From a mechanical point of view (excluding hydrogen), the HEDE model is linear and less accurate than modern approaches that take into account the size and configuration of grains, plasticity, and other nonlinear effects, including the reciprocal influence in the crack ensemble. At the same time, the model is generally recognized in modeling of hydrogen brittleness and allows us to investigate the skin effect in the simplest approximation, without any additional assumptions.

In our modeling, we used hydrogen diffusion Eq. (2). It ignores the presence of hydrogen traps [66] inside the material. All hydrogen is thought to be in the perfect crystal lattice. On the one hand, this is undoubtedly a simplifying assumption; on the other hand, it allows one to estimate the net effect of the skin layer formed during hydrogen charging of corset specimens. At the same time, such an assumption is physically possible; we simply assume a defect-free (without trap sites) homogeneous internal structure in the specimen.

There is another factor that justifies the made assumptions. According to the Oriani trap model for hydrogen transport in metals [66], the binding energy of trapped hydrogen is much higher than the diffusion activation energy, and the volume concentration of trapped hydrogen is only a fraction of the concentration of hydrogen diffusing in the crystal lattice. Otherwise the Oriani formula relating trapped hydrogen to lattice one does not work. This is because it is unclear from where hydrogen can be obtained with a sharp change in the volume concentration of trapping sites (such a change occurs, for example, during plastic deformation and this is one of the problems of the HELP model). Oriani’s assumptions make the effect of trapped hydrogen insignificant in dynamic processes since it has a smaller concentration and redistributes much more slowly than mobile hydrogen. Naturally, Oriani himself and his followers proceeded from the fact that the processes of hydrogen redistribution are slow and the entire system at each step must be in thermodynamic equilibrium. This raises the question of whether these assumptions are adequate to conditions of strength tests on notched and grooved corset specimens [38, 40, 67–70].

We found out that the experimentally discovered skin effect leads to a dual mechanism of fracture. We can schematically represent the following distribution of the areas of influence of hydrogen over the cross section of the specimen (Fig. 9).

Fig. 9.
figure 9

Schematic of the regions of propagation of hydrogen and conventional cracks at the fracture surface.

Full size image

We understand that our calculation of crack propagation is rather simplified. As a rule, the HEDE model is used to calculate the moment of growth of a hydrogen-induced crack. Theoretically, the maximum mechanical stresses at the tip of a crack of a certain length are unbounded. Therefore, stress values obtained by the finite volume method for corner nodes can only be considered as a kind of approximation. The use of the HEDE model to describe the crack propagation in a metal free of hydrogen is also the very first rough approximation.

In our opinion, the most important result is the description of the influence of the skin effect in hydrogen-charged specimens on the HEDE mechanism. We found that a crack is quickly initiated, but it has a microscopic length for a long time since its growth is limited by hydrogen redistribution due to diffusion. This diffusion motion occurs simultaneously with a decrease in the hydrogen concentration around the crack and stops its growth by the brittle mechanism. Moreover, the mechanism of hydrogen-induced, brittle, fracture acts much deeper than the initial thickness of the hydrogen-saturated skin layer. At the edges of the fracture surface, brittle fracture is associated with an increased hydrogen concentration. Areas of hydrogen brittleness should be observed in this part of the specimen. The middle part exhibits usual fracture by the decohesion mechanism. This duality was often found in experiments [63, 71]. Moreover, an annular crack similar to that in Fig. 9 with the characteristic width about 1 mm was obtained on grooved corset cylindrical steel specimens [63].

The dual pattern of fracture is usually explained by the two main mechanisms of hydrogen-induced degradation of metals: the HEDE and HELP mechanisms. It should be noted that the HEDE mechanism was initially considered as a mechanism of pure brittle fracture, but later, by formally combining the HEDE model with the elastic-plastic material model, this rigorous approach began to “smear” [72–74].

The HELP + HEDE model of hydrogen brittleness has been recently used to describe the duality of hydrogen-induced fracture observed in practice. Thus, fracture is described either by the HELP mechanism which provides hydrogen-induced plastic fracture or by the brittle HEDE mechanism, depending on the local concentration of hydrogen at the crack tip [67, 75, 76].

According to [24, 38, 41, 77–79], the HELP mechanism works at the hydrogen concentration below the threshold CH0, whereas the HEDE mechanism activates at the CHcr concentration. These assumptions were used to make approximations of the experimental data and to plot the distribution of hydrogen over the cross section of a specimen based on the fracture pattern. However, this hybrid concept does not deal with direct measurements of the distribution of hydrogen concentration. It focuses only on the microstructure of the fracture surface.

Almost all publications on the HELP and HEDE models use the assumption of the initially uniform hydrogen concentration. The presence of such inconsistencies in the known approaches allows us to say that the nonuniform hydrogen distribution in hydrogen-charged specimens is the true reason for the dual pattern of fracture observed in practice.

When verifying the models, many authors take the hydrogen concentration values calculated by Taha and Sofronis [80] as the reference data. These values indicate that the hydrogen concentration at the crack tip exceeds manifold the average values. This excess is obtained due to the formation of a large number of hydrogen sites inside dislocations during plastic deformation. The calculations performed by the authors of the HELP model in [44] show that significant changes in the mechanical properties of a metal occur at local relative mass hydrogen concentrations of about 10–2, which is an unattainable concentration for most metals. The calculation of local plasticity in the theoretical analysis of a crack with a spherical tip showed that the local hydrogen concentration at the crack tip is about 100 times higher than the average one [69]. Given the average mass concentration of about 10–6, local concentrations do not exceed 10–4. Thus, the assumption about the operation of the HELP mechanism at low average hydrogen concentrations does not correspond to the verification results. At low average hydrogen concentrations, external mechanical stresses cannot induce such local hydrogen accumulation that can trigger the physical mechanisms of hydrogen-enhanced localized plasticity.

This is especially important for industrial tests for hydrogen-induced cracking of metals based on the methods of hydrogen charging of specimens resulting in the skin effect. It is necessary to additionally investigate whether this effect occurs when oil, gas, hydrogen or other aggressive mixtures are pumped through pipelines and how it affects mechanic stresses in pipe walls.

6. CONCLUSIONS

We carried out special finite volume modeling of fracture of a hydrogen-charged corset cylindrical steel specimen with a notch. The HEDE model was used as a model of hydrogen embrittlement. The experimentally observed skin effect due to hydrogen charging was taken into account.

The simulation results demonstrated that fracture begins from the specimen surface as brittle fracture induced by hydrogen, then, due to the “lag” of diffusion fluxes of hydrogen, the crack grows by itself at normal background values of the hydrogen concentration.

This results in a dual pattern of fracture. The areas of both hydrogen embrittlement and normal fracture are observed on the fracture surface of the specimen.

The nonuniform distribution of hydrogen concentration can be the main source of the dual pattern of fracture, which is currently explained by the simultaneous operation of the HELP and HEDE mechanisms at the top of the main crack.

The skin effect in hydrogen-charged specimens has a strong influence on the fracture of metal specimens in spite of the very shallow depth.

Artificial charging with hydrogen can differ significantly in the distribution of hydrogen concentrations from natural saturation with hydrogen during operation, which must be taken into account in experiments since the skin effect strongly affects the results of mechanical tests.

From an engineering point of view, the obtained results mean that standard tests for hydrogen-induced cracking must be changed since the HIC due to the skin effect may not correspond to the HIC with a more uniform distribution of hydrogen inside the metal.