The Effect of Hydrogen on Failure of Complex Phase Steel under Different Multiaxial Stress States

Abstract

The demand for advanced high-strength steel (AHSS) in the automotive industry has increased over the last few years. Nevertheless, it is known that AHSSs are susceptible to hydrogen embrittlement. Therefore, the influence of hydrogen on the localization and damage behavior of a CP1000 steel sheet was investigated in this work. The sheet metal was electrochemically charged to a hydrogen content of about 3 ppm (by weight). Tensile tests were performed at different nominal strain rates between 0.00004 s⁻¹ and 0.01 s⁻¹ to investigate the effects of strain rates on their susceptibility to hydrogen embrittlement. Nakajima tests were utilized to investigate the hydrogen effects on the steel’s formability under different stress states. Three different Nakajima specimen geometries were employed to represent a uniaxial stress state, a nearly plane strain stress state, and an equibiaxial stress state. Further, forming limits were evaluated with the standardized section line method. Hydrogen embrittlement, during tensile testing, occurred independent of the strain rate, unlike the Nakajima test results, which showed hydrogen effects that were strongly dependent on the stress state.

1. Introduction

To increase the efficiency, safety and performance of vehicles, advanced high-strength steels (AHSS) are particularly attractive in the automotive industry. Significant reduction in vehicle weight can be achieved by using AHSSs with a tensile strength of 1000 MPa and more. However, it is well known that AHSSs are especially susceptible to hydrogen embrittlement, which results in deteriorated critical mechanical properties, such as ductility, toughness and strength.

Hydrogen can enter metal structures by absorbing hydrogen from the atmosphere, during manufacturing processes, or through electrochemical reactions such as corrosion. The adsorption and dissociation processes of hydrogen on metal surfaces and the absorption into bulk material have been well studied [1,2,3]. Several theories and models of hydrogen degradation mechanisms have previously been proposed. The relevant theories which apply to hydrogen in steels are hydrogen-enhanced decohesion (HEDE) [4,5], hydrogen-enhanced localized plasticity (HELP) [6,7,8] and adsorption-induced dislocation-emission (AIDE) [9]. The interplay of these mechanisms is still an ongoing subject of research [10]. The HEDE and HELP mechanisms assume that hydrogen diffusion and solubility are influenced by hydrostatic stress states [11]. This stress-driven hydrogen diffusion also has a large corresponding influence on shaping processes where different stress states can occur. The influence of hydrogen on AHSS has been studied by several authors using uniaxial tensile tests [12,13,14,15,16]. Nevertheless, the influence of multiaxial stresses due to forming processes needs to be examined in more depth [17,18]. To address this knowledge gap, this work investigates the influence of nearly uniaxial tension, plane strain and equibiaxial tension on hydrogen-induced localization and failure. Nakajima tests were performed according to DIN EN ISO 12004-2 [19] to determine the stress state-dependent forming limit, which is defined by the beginning of localized necking [20].

2. Materials and Methods

2.1. Metallographic Methods

To acquire microstructure images, metallographic cross sections were prepared via grinding, polishing, and subsequent etching in a solution of nitric acid and alcohol (Nital etchant). Pictures of the microstructure were taken with a Nikon Eclipse ME optical microscope (Nikon Metrology GmbH, Düsseldorf, Germany). The fracture surfaces were examined with a Hitachi S-S400N Typ II (Hitachi High-Tech Corporation, Tokyo, Japan) scanning electron microscope (SEM) using an accelerating voltage of 25 kV and a secondary electron detector. The hardness was measured with a Wolpert small load hardness tester vtd12 (ITW Test & Measurement GmbH, Leinfelden-Echterdingen, Germany).

2.2. Material

In this study, an advanced high-strength HCT980C (CP1000) steel sheet with a sheet thickness of 1.5 mm was investigated. The material has been used in earlier studies to analyze strain rate-dependent shear behavior [21]. Its microstructure is shown in Figure 1a. Heibel et al. investigated the microstructural composition of a HCT980C steel sheet, showing a microstructure with 4.3% ferrite, 93% bainite (or tempered martensite), 0.9% martensite and 0.3% retained austenite [22]. The CP1000 steel studied in this work is similar to the one studied by Heibel et al. It must be noted that the microstructural composition may deviate strongly depending on the processing and sheet thickness. The hardness at various positions over the sheet’s thickness is given in Figure 1b. It is constant with a mean value of 337 ± 6 HV0.1. The chemical composition was measured via optical emission spectroscopy (OES) on the surface-near area of the sheet material, which is given in

Table 1. Chemical composition of the sheet material measured via OES given in wt.%.

Fe	C	Si	Mn	P	S	Cr	Mo	Ni
bal.	0.30	0.32	2.34	0.028	0.030	0.22	0.02	0.02
	Al	Co	Cu	Nb	Ti	V	W	Pb
	1.45	0.02	0.03	0.03	0.07	0.02	0.09	0.03

. There is a slightly higher C and S content compared to the material data sheet for HCT980C [23], which might be a result of surface effects.

2.3. Hydrogen Charging and Measuring

Cathodic electrolysis was used to charge the specimens with hydrogen. Therefore, the specimens were prepared by grinding the surface to remove any residues, followed by cleaning with ethanol and acetone. A Wenking HP 96 potentiostat (Bank Elektronik—Intelligent Controls GmbH, Pohlheim, Germany) was connected to the test piece as the working electrode, and a platinum-coated mesh electrode was used as the counter electrode. Both electrodes were placed in a small vessel filled with 0.1 M NaOH solution as the electrolyte; see Figure 2. Charging was carried out at room temperature, and the solution was rinsed with N₂ for at least 1 h before the charging procedure.

For the charging process, achieving a high hydrogen content was desirable while minimizing the damage done to the specimen surface by electrochemical processes. To achieve this balance, different charging conditions were tested. First, the constant cathodic current density was varied, ranging from 1 to 15 mA/cm². Second, electrolytes with and without 1 g/L CH₄N₂S were tested, and lastly, the charging time was varied from 1 h to 5 h. The optimal combination of charging conditions was identified for the following experiments and is given in

Table 2. Hydrogen charging conditions for Nakajima specimens and tensile test specimens.

Solution	Current Density [mA/cm²]	Time [h]
0.1 mol NaOH + 1 g/L CH4N2S	1	5

. To minimize any hydrogen effusion before testing, all experiments started within 15 min after charging and finished no later than 60 min after charging.

The ELTRA OH900 analyzer (ELTRA GmbH, Haan, Germany) measured the hydrogen content of the specimens by hot gas melt extraction. Therefore, small pieces with a weight of approximately 1 g were cut from the metal sheet or test specimen. The measurement for the non-charged CP1000 steel resulted in initial hydrogen concentrations of 1.35 wppm and 1.13 wppm. The measurements after hydrogen charging, under the charging conditions described in Table 2, with a wait time of one hour, resulted in a hydrogen content of 3.63 wppm. The hydrogen contents after the Nakajima tests were 3.18 wppm in the deformed area and 2.67 wppm in the clamping area.

2.4. Tensile Tests

Tensile tests were performed on smooth tensile specimens at multiple strain rates, according to ISO 26203-2 [24]; see Figure 3. The tensile orientation was chosen perpendicular to the rolling direction of the metal sheet, and the specimen extraction was performed by micro waterjet cutting. The tests were performed using a universal testing machine. The tensile tests were carried out at constant test speeds of 0.001 mm/s, 0.02 mm/s and 0.25 mm/s to investigate the influence of the test speed on hydrogen embrittlement. With the used specimen geometry in Figure 3 on the left, these test speeds led to nominal engineering strain rates of 0.00004 s⁻¹, 0.0008 s⁻¹ and 0.01 s⁻¹. The nominal engineering strain rate is defined in ISO 26203-2 [24] and is calculated by the adjusted test speed of the testing machine and the parallel length of the specimen [24,25]. The medium value for the time-dependent engineering strain rate between the beginning of yielding and up to the beginning of necking, defined as the characteristic strain rate, is mostly slightly lower than the nominal strain rate; see Figure 3 on the right. After the beginning of necking, the strain rate increases [25,26]. Up to the fracture, strain rate values of about 2–10 times the nominal engineering strain rate can occur, depending on the hardening value of the material [25,26,27,28]; see also Figure 3 on the right. Altogether, the nominal strain rate can be seen as an average strain rate value for the whole test.

2.5. Nakajima Tests

The Nakajima tests were performed to investigate hydrogen embrittlement on the formability under different stress states. The Nakajima test involves deforming sheet metal blanks of different geometries until a fracture occurs, using a hemispherical punch. By varying the specimen widths, different deep drawing and stretch forming conditions occur on the sheet metal’s surface [17,19,20,27,28]. In this work, three different Nakajima specimen geometries were chosen: an approximately uniaxial stress state (Figure 4a), a nearly plane strain stress state (Figure 4b) and an equibiaxial stress state (Figure 4c). Therefore, the specimen geometries will be named uniaxial Nakajima (UAN) specimen, plane strain Nakajima (PSN) specimen and equibiaxial Nakajima (EBN) specimen in the following. The Nakajima tests were performed using a servo-hydraulic testing machine with an integrated clamping device and a punch speed of 1.5 mm/s. These test conditions are in accordance with DIN EN ISO 12004-2 [19]. For the disk-shaped specimen, this punch speed leads to a strain-rate order of the local strain rate of about 0.01 s⁻¹ in the highly deformed zone and can increase up to the beginning of necking up to 0.08 s⁻¹ [27]. This strain rate range is comparable to the largest investigated nominal strain rate in the tensile tests. The experimental setup can be found in [27], and the testing procedure is described in [28] in more detail. Achieving failure in the center of the specimen surface in order to get a defined loading situation is critical. Therefore, a multilayer film of plasticized PVC and Teflon sheets was placed between the tup and the specimen to minimize the friction between the punching tup and the specimen.

The formability of the reference non-charged and hydrogen-precharged material was investigated. The forming limits were evaluated according to DIN EN ISO 12004-2 [19], using the three-dimensional digital image correlation DIC-software ARAMIS v6.3 (Carl Zeiss GOM Metrology GmbH, Braunschweig, Germany) [29]. The forming limit was determined to be the maximal allowed major and minor strain in a forming process, which avoids localized thinning and fracture. Therefore, it follows that the forming limit can be interpreted as a parameter representing the beginning of localized necking [30]. A characteristic stochastic black and white pattern was applied on the surfaces of the specimens. The deformation of the surface was recorded with two high-speed cameras with 25 frames per second. For each specimen geometry, at least three tests with non-charged reference specimens and three tests with hydrogen-precharged specimens were performed. The evaluation of the forming limit (following the section line method proposed in DIN EN ISO 12004-2 [19]) was performed by analyzing the major and minor strains in the last image before the fracture. First, from the specimen, three sections that were perpendicular to the fracture line were identified. The polynomial fits for the major and minor strains versus the position on the section were approximated automatically by the ARAMIS software; see Figure 5. The crossed strain values in the gray windows in the diagram, in Figure 5 on the right, were used for the polynomial fitting. The forming limits were then evaluated as the maximum values of the major and minor strains gained by the polynomial fits.

3. Results

The engineering stress–strain curves for the tensile tests are shown in Figure 6. The black curves represent the non-charged reference specimens, and the blue curves represent the hydrogen-precharged specimens. The mechanical properties are given in

Table 3. Mechanical properties determined during tensile tests with different crosshead speeds in air and after hydrogen charging.

Condition	Crosshead Speed [mm/s]	Rp0,2 [MPa]	Rm [MPa]	εf [%]	RA [%]
Non-charged	0.001	906	998	13.7	55.5
Non-charged	0.02	894	1002	13.7	53.0
Non-charged	0.25	912	1005	12.6	55.1
Precharged	0.001	939	1020	9.6	42.9
Precharged	0.02	940	1019	9.0	43.6
Precharged	0.25	942	1024	8.4	39.1

. As expected, the hydrogen-precharged specimens show significantly smaller elongation at fracture and moderately larger strength when compared to the non-charged reference specimens. The elongation at fracture of the non-charged reference specimen at the largest investigated strain rate of 0.01 s⁻¹ (test speed of 0.25 mm/s) shows a smaller elongation at fracture compared to the non-charged specimens at lower strain rates. This effect could be caused by adiabatic heating at elevated strain rates. Additionally, the hydrogen-precharged specimen at the strain rate of 0.01 s⁻¹ shows a smaller elongation at fracture when compared to the precharged specimens at lower strain rates. Nevertheless, within the investigated nominal strain rate range of 0.00004 s⁻¹ up to 0.01 s⁻¹, no significant influence of the strain rate on hydrogen embrittlement can be observed.

The hydrogen embrittlement can be quantified by the relative reduction in area (RRA) and the relative elongation to failure (RƐ_f), according to ASTM G129 [31]. RRA is defined as the ratio of reduction in an area with hydrogen (RA_H) to an area without hydrogen (RA), and RƐ_f is similarly defined as the ratio of elongation to fracture with hydrogen (Ɛ_f,H) to that without hydrogen (Ɛ_f).

$$\mathrm{RRA}=\frac{{\mathrm{RA}}_{\mathrm{H}}}{\mathrm{RA}}$$

(1)

$$\mathrm{R}{\mathrm{Ɛ}}_{\mathrm{f}}=\frac{{\mathrm{Ɛ}}_{\mathrm{f},\mathrm{H}}}{{\mathrm{Ɛ}}_{\mathrm{f}}}$$

(2)

For both measures, a value of 1 corresponds to no hydrogen effects. The results are shown in Figure 7. For the slow tests at 0.00004 s⁻¹, the hydrogen embrittlement index, RƐ_f, seems to be slightly larger compared to the tests at higher strain rates. However, with an increasing strain rate, no further influence of the strain rate on the hydrogen embrittlement index, RƐ_f, can be seen in the investigated range of strain rates. The RRA exhibits a little more scatter and the lowest value for the largest strain rate of 0.01 s⁻¹. Altogether, no significant trend with varying strain rates in the investigated strain rate range from 0.00004 s⁻¹ up to the forming relevant strain rate of 0.01 s⁻¹ is observed.

Since the tensile tests at the highest investigated nominal strain rate (0.01 s⁻¹, crosshead speed of 0.25 mm/s) displayed hydrogen embrittlement, Nakajima tests were performed at a forming relevant test speed of 1.5 mm/s, which leads to comparable strain rates in the localized zones [28]. The evaluation of the forming limits was performed according to the section line method, described in Figure 5 in Section 2.5, and the results are plotted in Figure 8 for the non-charged and precharged specimens.

For the UAN and PSN specimens, there were only very slight differences observed concerning the evaluated forming limits of the non-charged reference and the hydrogen-precharged specimens. Nevertheless, the measured maximal major strain for the hydrogen-charged specimens is reduced for both specimen geometries. The calculated forming limit is significantly reduced by hydrogen for the EBN specimens, and the non-charged reference specimen shows a comparable measured major and minor strain, as expected for an equibiaxial stress state. In contrast, the EBN hydrogen-precharged specimen’s measured maximal major strain is about 50% larger than the measured maximal minor strain. Therefore, it can be concluded that the stress state changes from an equibiaxial deformation into a plane strain deformation during the Nakajima testing of the precharged EBN specimen.

In Figure 9, the forming limits obtained by the section line method are plotted, showing the major versus minor strains. Even though differences for the local failure strain, evaluated by DIC, can be deduced from Figure 8, the computed forming limit, according to DIN EN ISO 12004-2 [19] and calculated as the major and minor strains, seems to be not significantly reduced by hydrogen, except for that in an equibiaxial stress state.

4. Discussion

Specimens were electrochemically charged for 5 h with a current density of 1 mA/cm², increasing the hydrogen content from about 1.2 wppm of the non-charged material to about 3 wppm. Lovicu et al. conducted notched SSRT on four different AHSSs with different hydrogen contents and evaluated the critical hydrogen contents, C_H,cr, for these materials, which were found to be between 1 wppm and 4 wppm [12]. The hydrogen measurements that were taken 1 h after electrochemical charging showed no significant effusion. By taking the hydrogen measurements within 1 h after charging, it was ensured that most of the hydrogen remained in the specimen, and a critical hydrogen content was reached.

Uniaxial tensile tests were performed at different crosshead speeds to investigate the effects of different strain rates. Even though strong effects of strain rate on hydrogen embrittlement for the range between 0.00004 s⁻¹ and 0.01 s⁻¹ are reported in the literature for different steels [32,33,34,35], no significant effects could be seen for the CP1000 steel investigated in this work. However, Depover et al. investigated the effects of hydrogen precharging on DP600 steel in high strain-rate tensile tests and found a reduction of elongation to failure for the static strain rates of 0.0167 s⁻¹ (HE = 72%) and 1.67 s⁻¹ (HE = 45%). Further testing at dynamic strain rates of 450 s⁻¹ (HE = 16%) and 900 s⁻¹ (HE = 11%) showed less hydrogen influence, but it was still observable [36]. Nevertheless, the strain rate during the Nakajima tests was about 0.01 s⁻¹, at which hydrogen embrittlement fully occurred during the tensile tests of this work. Therefore, when conducting the forming processes in this strain rate range, hydrogen embrittlement and its effects still must be taken into consideration.

Nakajima tests with three different stress states were performed in this work. The demonstration of the complete strain paths, which were plotted as major strains dependent on minor strains, confirms a change in the stress state from an equibiaxial deformation into the direction of a plane strain deformation for the different evaluated EBN specimens; see Figure 10. This deviation from the linear equibiaxial strain path into the direction of a plane strain deformation with nearly constant minor strains is defined by Hora as the beginning of localization [37]. Therefore, the stress state seems to affect hydrogen embrittlement. In this work, earlier localization of necking could only be seen for the equibiaxial specimen—not for the plane strain or the uniaxial specimen. This could be due to a higher hydrostatic stress state in the equibiaxial specimen (σ_H = 2/3 σ₁) compared to the uniaxial specimen (σ_H = 1/3 σ₁). An increase in the hydrostatic stress state increases hydrogen diffusion and solubility and, therefore, might enhance local dislocation mobility, according to the HELP theory.

Moreover, Gao et al. investigated the hydrogen effects on the forming limits of DP1180 steel with the Nakajima experiments, but the evaluation of the forming limits was not standardized using a time-dependent evaluation method. Similarly, for the UAN specimen, no effect of hydrogen on the forming limit was seen. However, for the PSN specimen, a drop in major strain for the hydrogen-charged specimen was observed. The results of the EBN specimen in this work showed a comparable hydrogen-induced reduction in the forming limit, but Gao et al. evaluated fracture without accounting for the localized necking in hydrogen-precharged specimens [17].

The fracture surfaces of the UAN non-charged reference specimen and the hydrogen-precharged specimen can be seen in Figure 11. Both specimens show macroscopic ductile failure, with the hydrogen-precharged specimen showing comparatively larger dimples all over the surface. However, at a higher magnification, a local brittle material failure can be observed on the ground of those dimples. The non-charged UAN shows a more uniform distribution of dimples. The PSN specimens also failed under macroscopically ductile conditions (Figure 12). The dimples appear shallower compared with the UAN specimen, but the hydrogen-precharged PSN specimen shows similar local brittle crack initiations at the base of the dimples (see red inserts in Figure 12d). In Figure 13, it can be seen that the non-charged EBN specimen shows a more homogeneous surface, with small dimples and ductile failure than the hydrogen-precharged specimen, where a large defect in the center of the fracture surface is observed. On the ground of the defect, a transgranular fracture occurred. It is noted, however, that for all strain states, differences in the appearance of the fracture surfaces of hydrogen-precharged specimens are seen compared to the non-charged reference samples.

In similar studies on AHSSs, dimples can be found on non-charged specimens, with increasing areas of transgranular fracture, cleavage or even intergranular fracture on the fracture surface after hydrogen charging [10,38]. Gao et al. produced similar results to what was observed in this work, where the hydrogen-charged specimens show ductile fracture with a transgranular quasi-cleavage in larger dimples [17]. All specimen types exhibited hydrogen embrittlement, leading to a hydrogen-induced reduction of the local strain at fracture.

5. Summary

This work investigated the strain rate and the stress-state dependency on hydrogen embrittlement of the advanced high-strength steel, CP1000.

(1)

The tensile tests on non-charged and electrochemically precharged specimens showed that hydrogen embrittlement is effectively independent of the strain rate in the investigated range from 0.00004 s⁻¹ up to a forming relevant strain rate of 0.01 s⁻¹.

(2)

The Nakajima tests showed that the influence of hydrogen on the beginning of localized necking strongly depends on the stress state. For the equibiaxial stress state, localization of necking starts earlier for the precharged specimens than for the non-charged specimens. Critically, the local failure strain was reduced for all the Nakajima specimens tested with hydrogen precharging.

(3)

The subsequent evaluation of fracture surfaces showed an observable increase in the occurrence of local brittle fracture initiation for the precharged specimens.