Investigating the influence of hydrogen on stress corrosion cracking of 2205 duplex stainless steel in sulfuric acid by electrochemical impedance spectroscopy

2.1 SSRTs

SCC susceptibility was evaluated based on SSRTs. The SSRT experimental specimen is shown in Figure 2.

Figure 2:

Size of the SSRT sample.

Each specimen was polished along the tensile direction to 1200-grit emery paper, degreased in acetone, and washed by distilled water. After that, all specimens were coated with silicone rubber to provide an identical, exposed surface area. Before the SSRT test, all specimens were divided into two groups. The first group of specimen was strained directly in the air, and the second group of specimen was precharged with a charging current density of −5 mA/cm² in 0.5 m H₂SO₄ for 24 h. For hydrogen charging, a three-electrode cell was used: the 2205 DSS electrode as working electrode (WE), a platinum grid as counter-electrode (CE), and a saturated calomel electrode (SCE) as reference electrode (RE). SSRT tests were carried out after cathodic charging. The latter group of specimens was completely immersed in aqueous 0.5 m H₂SO₄ during the whole test. All tests were carried out at a constant strain rate of 2×10⁻⁷ s⁻¹. Before test, aging treatment was carried out to reduce the residual stresses.

2.2 EIS tests

For the EIS test, all specimens were divided into three groups: (1) no pre-charging and no stress, (2) pre-charging for 24 h but no stress, and (3) hydrogen charging for 24 h and then take SSRT tests. All specimens were performed in 0.5 m H₂SO₄ made with distilled water at 20°C. The exposed coating area was 3.9 cm². For the first two groups, EIS tests were performed every 2 h until the passive film of the DSS remained stable. For the third group, EIS tests were performed regularly from the beginning up to the failure of the specimens every 2 h by Gamry Reference-600 electrochemical workstation. A platinum grid acts as CE and SCE acts as RE, as shown in Figure 3.

Figure 3:

Experimental set-up.

A sinusoidal potential wave with an amplitude of 10 mV in a frequency range of 10 MHz–100 kHz was applied to the samples.

2.3 Fracture morphologies tests after SSRT

All specimens were examined using JSM-6480 scanning electron microscope (SEM) and optical microscope after the SSRT tests. The purpose was to distinguish the difference of fracture type between hydrogen charged and non-hydrogen charged.

2.4 Hydrogen permeation test

Hydrogen permeation experiments were conducted at 20°C by D-S double cells. The steel was cut to a thickness of 1.0 mm using a wire-cut technique and then ground to a final thickness of 0.5 mm. The surfaces of both sides of the steel were polished to 1 μm then washed with isopropyl alcohol and rinsed by distilled water followed by air-drying. The RE (Hg/HgO) was positioned 8 mm away from the WE using a Lugin capillary tube, whereas the CE was 80 mm away from the WE. The experimental set-up was given in a previous paper (Zheng et al., 2016). The cathodic cell (charging side) was filled of 0.4 m hydrochloric acid and the anodic cell (oxidizing side) was filled of 0.2 m NaOH. The potential for passivation was 0.2 V (vs. Hg/HgO) and the cathodic charging current density was 5 mA/cm². Until the passive current density was less than 1.5 μA/cm², a corrosion solution (0.4 m hydrochloric acid with and without molybdate) was added into the charging side, and the charging current of −5 mA was used simultaneously.

3.1 SSRTs

The results of SSRTs in H₂SO₄ solution are shown in Figure 4. It can be seen that there is a little difference in the ultimate tensile strength at the two conditions. However, the elongation was reduced in H₂SO₄ solution compared to air. To investigate SCC susceptibility, reduction in area loss factor (F_R) and elongation loss factor (F_E) were calculated as follows (Liu et al., 2008):

Figure 4:

Stress-strain curves of 2205 DSS.

where R_E and E_E are reduction in area and elongation in H₂SO₄ solution, respectively, and R₀ and E₀ are reduction in area and elongation in air, respectively. The results are presented in Table 2. F_R and F_E are the ductility loss factors, and R_E and E_E of the steel in H₂SO₄ solution are lower than R₀ and E₀, which suggests that SCC occurs in H₂SO₄ solution. According to some researchers (Guo et al., 2011; Lv et al., 2016; Moradi et al., 2014), SCC resistance of the alloy 2205 DSS mainly depends on the residual stresses present in the material. Before test, aging treatment was carried out to reduce the residual stresses. Therefore, SCC in this paper is mainly because some of the hydrogen atoms absorbed by the metal surface during hydrogen charging permeate into the metal, resulting in HE (Chu et al., 2003).

3.2 EIS tests

Figure 5 shows the Nyquist curves of the unstressed 2205 DSS with no hydrogen charging for 50 h. All Nyquist curves present a similar appearance. The semicircle of Nyquist curves is smallest at 0 h. The diameter of Nyquist curves increases with increasing immersion time in H₂SO₄ solution until the immersion time reaches 8 h. The initial increase in the diameter of Nyquist curves is caused by the passivity ability and kinetics of film growth of the DSS surface. After 8 h, the diameter of Nyquist curves remains the same. This phenomenon shows that the density of the passive film reaches a maximum. It appears that film formation during immersion maintains the high polarization resistance for hydrogen-free DSS.

Figure 5:

Nyquist curves of 2205 DSS under unstressed and non-hydrogen-charging condition.

Figure 6 shows the Nyquist curves of the unstressed 2205 DSS under hydrogen-charging conditions for 160 h. Nyquist curves of the hydrogen-charged DSS show a similar trend with hydrogen-free DSS up to 8 h. However, when the immersion time increases beyond 8 h, the diameter of Nyquist curves has consistent growth until the immersion time reaches 98 h. Compared to the hydrogen-free DSS, the absolute value of the impedance of hydrogen-charged DSS is obviously lower. The reason for the above phenomenon is primarily that hydrogen charging causes the deterioration and instability of the passive film on the DSS surface. Thus, it would require more time to repair itself after hydrogen charging. To prove the passivation film instability after hydrogen charging, Raman spectroscopy was carried out in air for 24 h and charging for 24 h. The results are shown in Figure 7.

Figure 6:

Nyquist curves of 2205 DSS under unstressed and hydrogen-charging condition.

Figure 7:

Raman of 2205 DSS.

According to Yu et al. (1997), passivation film exhibited Raman shifts of 300, 480, and 700 cm⁻¹. The intensity of Raman shifts decreased after charging for 24 h under −5 mA/cm² in 0.5 m H₂SO₄. This indicated that hydrogen charging causes deterioration and instability of the passive film on the DSS surface.

Figure 8 shows the Nyquist curves of 2205 DSS under stressed and hydrogen-charging conditions from the beginning of the stretch to the fracture of DSS. The diameter of Nyquist curves increases with increasing immersion time until the DSS specimen breaks down. It shows a superficially similar trend with unstressed DSS. This is due to the passivation of the DSS surface by H₂SO₄ solution. However, cleavage fracture and secondary cracks on the fracture surface of the hydrogen-charged specimens are observed after the failure of DSS. This means that it not only reflects a passivation process.

Figure 8:

Nyquist curves of 2205 DSS under stressed and hydrogen-charging condition.

The comparison of Nyquist curves of specimens with and without stress measured at different times is shown in Figure 9A to D. At 2 h of the stretch, the diameter of Nyquist curves of the stressed specimen is larger than that without stress. This phenomenon shows the difference of surface states of two specimens at original time. However, the difference between the two diameters becomes unclear with increasing immersion time. When the immersion time reaches 10 h, the diameters of the two Nyquist curves are almost equal, as shown in Figure 9C. After that, the increasing speed of the diameter of the unstressed specimen is larger than that with stress. Figure 9D shows the Nyquist curves of the two conditions at 40 h. It can be clearly seen that the diameter of the unstressed specimen is distinctly larger than that of the specimen with stress. The experimental results show that the stress more or less inhibits the passivation of the surface. This is mainly because the addition of stress causes the initiation of the cracks on the specimen surface. However, the increasing trend of the impedance of the stressed specimen during SSRTs provides evidence that film passivation is the dominant factor. Although the cracks appeared during SSRTs, they would also soon be passivated.

Figure 9:

Nyquist curves of specimens with and without stress at different times: (A) 2 h, (B) 8 h, (C) 10 h, and (D) 40 h.

The impedance data for 2205 DSS with no stress and no hydrogen-charging conditions agreed with the equivalent circuit shown in Figure 10. The circuit consists of the following: R_sol is the solution resistance, R₁ is the passive layer resistance, and CPE₁ is the capacitance equivalent of the constant phase element, which consists of the capacity of the passive layer formed (C₁) and the phenomenological coefficient (n). The n value of a CPE indicates its meaning: n=1 (capacitance), n=0.5 (Warburg impedance), n=0 (resistance), and n=−1 (inductance) (Nam, et al., 2009).

Figure 10:

Equivalent circuit for 2205 DSS under unstressed and non-hydrogen-charging condition.

Table 3 shows the impedance data for 2205 DSS with no stress and no hydrogen-charging conditions. At this condition, n is consistently maintained at 0.93, which is the result of the deviation from the ideal dielectric behavior, arising due to the heterogeneity of the surface both laterally and within the depth of the oxide film. R_sol changes from 0.35 to 0.47 Ω cm², indicating no obvious change of solution during immersion time. In the first 8 h of immersion time, the decrease of C₁ and increase of R₁ show that the passive layer of DSS in solution becomes increasingly more stable. After 8 h, R₁ remains stable, reflecting the fact that the passive layer becomes stable.

The circuit model of 2205 DSS under unstressed and hydrogen-charging conditions is proposed and presented in Figure 11. In the model, R_sol is the solution resistance and CPE₁ is the capacitance equivalent of the constant phase element, which consists of the capacity of the passive layer formed (C₁) and phenomenological coefficient (n). R₁ is the passive layer resistance, C_dl is the double charge layer capacitance between the DSS surface and H₂SO₄ solution, and R_dl represents charge-transfer resistance.

Figure 11:

Equivalent circuit for 2205 DSS under unstressed and hydrogen-charging condition.

Table 4 shows the electrochemical impedance parameters obtained from the fitting of the EIS diagram. R_sol changes from 0.34 to 0.64 Ω cm², indicating no obvious change of solution during immersion time. The value of n is consistently maintained at 0.9. The closer to 1 the numerical value of n is, the greater is the property of the passive layer (Lavigne et al., 2014). The change of C₁ with time indicates that the passive layer is formed and its thickness increases. The value of R₁ increases with increasing immersion time first and then remains stable after the immersion time reaches 98 h. This reveals a process where the passive layer, which is deteriorated and broken down by cathodic hydrogen charge, forms during immersion time. The value of C_dl decreases, indicating that the area of metal exposed in the solution becomes smaller; this is due to the formation of the passive layer. The change in trend of R_dl is opposite of that of C_dl, which results from larger electrical resistance with increasing immersion time.

The maximum value of R₁′ (passive layer resistance under unstressed and hydrogen-charging conditions) is 1.619×10⁵ Ω cm² at 8 h. At the same time, the maximum value of R₁ (passive layer resistance under unstressed and hydrogen-charging conditions) is 5540 Ω cm² at 98 h, which is much smaller than R₁′. This means that the property of the passive layer, which is passivated in H₂SO₄ solution, after hydrogen charge is far poorer than without hydrogen charging, and it also requires a much longer time for the passive layer to become stable.

The circuit model of 2205 DSS under stressed and hydrogen-charging conditions for the first 6 h is the same as the unstressed and hydrogen-charged one, as shown in Figure 11. Therefore, the behaviors of 2205 DSS under two different conditions are the same for the first 6 h during passivation of the DSS surface. The electrochemical impedance parameters obtained from the fitting of the EIS diagram are shown in Table 5. Compared to R₁ and R₁′, it can be seen that the increasing rate of R₁ remains large and stable from 0 to 6 h, whereas the increasing rate of R₁′ from 2 to 6 h is less than R₁. This is due to the inhibition of the passivation of 2205 DSS by stress. The passive film formed in unstressed and hydrogen-charging conditions is easier. After 6 h, the circuit model in Figure 11 is no longer suitable for the stressed and hydrogen-charging conditions, and other circuit models are also not suitable for this condition. The reason for this is that cracks are generated after 6 h. The cracking process generates high instability, and the spectrum obtained from a long time measurement is averaged (Lou & Singh, 2011). Thus, data fitting using an equivalent circuit is not suitable for SCC studies. The transmission line model (TLM), as shown in Figure 12, was used to describe the whole SCC process during SSRTs (Liu et al., 2008; Lou & Singh, 2011; Nam et al., 2009). All the systems studied have a similar characteristic – the phase angle decreases at a certain point in time. This means that a single crack propagation process occurs in these systems. However, the phenomenon is not found in this system; the diameter of Nyquist curve and phase angle continued to increase during measuring time. In other words, when the crack is generated, it is not a single propagation process. Thus, the TLM is not suitable for this system. To evaluate the initiation and propagation of cracks, the phase shift (phase_stressed–phase_nonstressed) between two conditions is used (Bosch, 2005).

Figure 12:

TLM.

Generally, the analyzed range of frequencies in EIS should reflect the dynamics of the electrochemical process. The response from a higher-frequency perturbation, such as 100 and 1000 Hz, represents a faster electrochemical process, which is closer to a non-Faradaic progress (Bosch et al., 2001; Lou & Singh, 2011). Therefore, data clearly show that a frequency that is lower than 10 Hz reflects the electrochemical behavior during SSRT. It should also be noted that the slower-frequency analysis, such as 0.01 Hz, can generate higher instability during the dynamic cracking process. Therefore, the phase angle from a frequency between 0.1 and 10 Hz better represents the cracking behavior.

The phase differences are calculated for different frequencies from 0.1 to 10 Hz. Figure 13 shows the evolution of phase shift during SSRTs. A clear difference between the phases of the tests with and without stress can be noticed from about 5 h at 10 Hz. At 1 and 0.1 Hz, the difference is clearer and also starts from about 5 h. Thus, it shows that for the application of EIS to the monitoring of SCC under these circumstances a frequency between 10 and 0.1 Hz is the correct choice. It also shows that the phase shift at 1 and 0.1 Hz is much larger than those at 10 Hz. Thus, 1 and 0.1 Hz responses are more pronounced than the 10 Hz response. When it reaches 6 h, the phase shifts start to increase, suggesting the formation of SCC. At 24 h, the phase shift at 1 Hz reaches the largest value and then it starts to decrease until the DSS fractures. A similar phenomenon is observed at 0.1 Hz, where it also has both a rising trend at the beginning of the SSRT and a downward trend in the process of SSRT, meaning the development of the crack is not continued. Because of the passivation ability of DSS in H₂SO₄, when the crack forms during SSRTs, it may be passivated quickly. As a result of the stress on DSS, the passive layer at the crack will be broken, but it will soon be repassivated again. This is the reason why phase shifts at 0.1 and 1 Hz do not increase continuously.

Figure 13:

Evolution of phase shift during SSRT.

3.3 Hydrogen permeation tests

Figure 14 shows the hydrogen permeation current curves of 2205 DSS with charging current density of −5 mA/cm² in 0.5 m H₂SO₄ for 24 h. With the time increasing, the current density increases gradually, indicating that hydrogen has permeated from the charging side to the oxidizing side (Zheng et al., 2016). The permeated hydrogen amount was calculated by integral. About 7.05 mmol H atoms permeated through the specimen, meaning 7.05 mmol H atoms permeated into the SSRT specimen during the 24 h charging time. Because the specimen was solution treated to reduce the residual stresses, SCC in this paper is mainly because the hydrogen atoms absorbed by the metal surface during hydrogen charging permeate into the metal resulting in HE (Chu et al., 2003).

Figure 14:

Hydrogen permeation curve of 2205 DSS under charging current density of −5 mA/cm² in 0.5 m H₂SO₄ for 24 h.

3.4 Fracture morphologies after SSRT

SEM photomicrographs of the hydrogen-free specimen after failure in air and the specimen after failure in 0.5 m H₂SO₄ with hydrogen charging are shown in Figure 15. It can be seen that the fracture morphology is significantly different between the two specimens. The fracture surface obtained in air shows that severe necking is distributed around the fracture (Figure 15A), indicating a typical ductile fracture in the surface. This phenomenon is not observed in the other specimens. Instead of severe necking, there are many steps around the fracture of the hydrogen-charged specimen in 0.5 m H₂SO₄ (Figure 15C). From the macrophotographs, necking is small for hydrogen-charged specimens, and fracture characters are found in hydrogen charging. The elongation for the hydrogen charged and uncharged are 20.4% and 13.8%, respectively. These indicate that, after hydrogen charging, SCC susceptibility increases.

Figure 15:

SEM photomicrographs of the hydrogen-free specimen after failure in air and the specimen after failure in 0.5 m H₂SO₄ with hydrogen charge. (A, B) No-hydrogen charged in air; (C, D) hydrogen charged in 0.5 m H₂SO₄.

A closer view of the photomicrographs is given in Figure 15B to D. Some typical dimples resulting from ductile fracture can be seen on the fracture surface of the specimen in air and no crack is observed. On the contrary, the cleavage fracture and secondary cracks on the fracture surface of hydrogen-charged specimens in 0.5 m H₂SO₄ are observed, indicating a brittle fracture.

Figure 16A and B shows the secondary cracks in the failed specimen with hydrogen charging. Cracks propagate from the surface of the specimen to the internal structure. Zigzag microcracking is shown in Figure 16B, which is a characteristic of cracks by HE of DSS (Lou & Singh, 2011; Zucchi et al., 2006). The transgranular cracking morphology is also shown. Both phenomena show evidence that when the secondary cracks run up against ductile austenite they will be arrested by austenite or change direction.

Figure 16:

Secondary cracks in the failed specimen with hydrogen charge. (A, B) Cross section of fracture after hydrogen charged in 0.5 m H₂SO₄.

A schematic figure of the repassivation mechanism is shown in Figure 17. At first, the passive films are formed somewhere on the DSS surface. After that, the stress made the passive film and the steel substrate break down. As soon as the passive film and the steel substrate were broken down, repassivation was initiated due to H₂SO₄ solution. Thus, although the stress during SSRT ruptures the passive film and steel substrate, the rate of repassivation is higher than the rate of breakdown of the passive film and steel substrate.

Figure 17:

Schematic figure of repassivation mechanism.