Passivity Breakdown and Crack Propagation Mechanisms of Lean Duplex (UNS S32001) Stainless Steel Reinforcement in High Alkaline Solution Under Stress Corrosion Cracking Free

Lean duplex stainless steel (LDSS) is a new type of DSS with a lower Ni content, which is compensated with other elements such as Mn or N as the austenite phase (γ phase) stabilizer in the duplex microstructure.¹  In addition, Mo content is also reduced to avoid the formation of secondary phases such as sigma (σ phase) or chi (χ phase) phases, which serve as anodic sites, enhancing the corrosion susceptibility, as well as inducing premature development of stress corrosion cracking (SCC).²  LDSS, such as 2304, 2101, 2102 or 2202, have similar corrosion resistance as 316L (UNS S31603⁽¹⁾), but at a lower price, however, some of these LDSS are limited in product form.^3-4 

The duplex microstructure has shown to have advantages over the polygonal grain γ-phase matrix of austenitic SS, avoiding severe crack branching as the γ phase will arrest the cracking, forcing the crack to go through the ferrite phase (α phase), or intergranularly through the α/γ interface.^5-6  This is due to the high-stress intensity factor for SCC (K_ISCC) that DSS is compared to austenitic, particularly in the K_ISCC of the γ phase.⁷  However, the interaction and properties mismatch between the two phases can also lead to the failure of the DSS, as microgalvanic coupling can initiate the pitting and/or cracking process due to the lower Cr content in the α phase granting a lower corrosion potential, thus acting as the anode.^8-9  In addition, secondary phases can also affect the corrosion and SCC behavior of DSS; σ, χ, and α’ phase produce Cr-depleted zones, considerably decreasing the pitting resistance, hence prematurely promoting pit nucleation facilitating the crack initiation.^10-11  Besides the effect of the phase properties, the environment plays a crucial role in the SCC mechanism of DSS, as the combination of ions, temperature, and pH will make one phase more prone to SCC than the other. In the case of strong oxidant or evaporated sea water (atmospheric corrosion), which leaves a saturated layer of MgCl₂, γ phase becomes more active than α phase, promoting transgranular SCC (TG-SCC).⁹  Prosek, et al., thoroughly studied the SCC behavior under atmospheric corrosion by means of chloride deposits of most DSS and austenitic SS available in the market, concluding that only at temperatures above 70°C SCC was found on DSS, where the mechanism found was stress-assisted selective dissolution (anodic dissolution).¹²  In the cases where only pitting was found, selective dissolution of the α phase was the main finding; however, the pit depth was one order lower than austenitic cases. In another work from the authors, the SCC at a low temperature with chloride deposits for austenitic and DSS was studied, where S32101 did not showed SCC development by U bend even after 22 weeks of exposure. It only showed preferential dissolution of the α phase and pitting nucleating from it.¹³  In simulated white liquor solutions, an industrial solution containing hydroxide and sodium sulfide, crack initiation, and propagation occurred in the γ phase, where the increase in temperature accelerates the SCC development, having a lower temperature threshold than the LDSS.¹⁴  The presence of the sulfides promotes the adsorption of sulfur on the pit bottom catalyzing Fe and Ni dissolution and enhancing the hydrogen effect.¹⁵  The mechanism behind the failure in this environment is a combination of continuous slip-step dissolution and hydrogen embrittlement.¹⁴  In caustic solutions, the SCC initiation and propagation depend on the alkalinity (concentration of OH⁻), where TG-SCC will be seen intermittently in between γ and α phase following the slip-step dissolution mechanism.¹⁶  In some cases, it has also been reported that hydrogen absorption was also present, promoting further SCC susceptibility to the alloy, even more in the case of S-containing solutions given the catalytic poisoning effect.¹⁷  In near neutral pH with chloride additions, the failure comes from the side of the α phase with a TG-SCC morphology, which is commonly arrested at the γ phase, and interganular SCC (IG-SCC) through the α/γ interface.⁹  DSS 2205 in neutral pH was able to withstand chloride concentrations up to 26 wt% NaCl without showing signs of SCC in the range of 25°C to 900°C, only reducing its elongation, nevertheless, after the polarization potentials reached the pitting potential, selective dissolution of the α phase was seen as a later trigger of the crack initiation through anodic dissolution promoting IG-SCC.¹⁸  At lower pH, LDSS 2101 suffered from SCC in the presence of thiosulfate, enhancing the SCC susceptibility due to hydrogen embrittlement, promoting cracks within the α phase but being arrested at γ/α grain boundaries.¹⁹ 

The superior corrosion susceptibility of DSS comes from the alloy composition enabling the formation of a stable passive film rich in Cr oxides, this protective film is thicker in DSS compared to austenitic SS and has fewer defects, including oxygen vacancies and metal interstitials.²⁰  This passive film stability is related to the ratio of γ/α phase, which by thermal treatments can be altered; in the case of welding, at the heat-affected zone, the γ/α-phase ratio changed due to the recrystallization and diffusion of Cr and Ni lowering the α-phase content, which enhanced the pitting corrosion resistance measured by the critical pitting temperature and the pitting resistance equivalence number calculations.^21-22  The thinning of the passive film on DSS due to the stress enhances the dissolution process by increasing the potential difference between the passive film and the electrolyte, doping the passive film with defects and generating soluble CrO₃, further decreasing the film thickness.²³  In a recent work from Macdonald, et al., the analysis of the passivity breakdown of hyper DSS 2707 in solutions containing bromide was studied using the point of defect model (PDM) theory, determining the critical density of defects for passivity breakdown.²⁴  By both PDM and Mott-Schottky analysis, the critical density of cation vacancies was calculated to be 10¹⁴ cm⁻², which tended to increase with temperature, as well as being correlated to the pitting potential, which decrease toward more negative values for increased temperature and bromide content.

This works aims to unravel the mechanism for the passivity breakdown and cracking mechanism of Type 2001 LDSS reinforcing bar in a high alkaline environment contaminated with chlorides at low temperature. For this task, thorough analysis of electrochemical tests as well as PDM theory would be done, relating the microstructure and passive film stability with the chloride content and applied stress. Critical electrochemical parameters such as repassivation rates, phase angle shifts, and density of defects are correlated with the passive film breakdown. The crack initiation and propagation are studied, analyzing the crack path and morphology to better understand how the chloride threshold and microstructure play a role in the SCC mechanisms.

The elemental composition of Type 2001 LDSS (UNS S32001) reinforcing steel may be seen in Table 1. The LDSS reinforcing bars were manufactured by the hot-rolling process at a temperature of 890°C with a postprocessing of pickling and passivation as per the manufacturer specifications (ACERINOX). The LDSS reinforcing bars were 3 mm in diameter. To accelerate the crack initiation process, the specimens were machined with a circular 60° V notch in the center of the sample.²⁵  Before the testing, samples were rinsed with deionized (DI) water, degreased with acetone, and blown-dried with air. A 2 cm² exposed area was selected by coating with epoxy lacquer (Midas 335-009 nonconductive paint).

The LDSS reinforcing bars specimens were tested under SCC conditions via uniaxial tensile test by the slow strain rate test (SSRT) while being immersed in a corrosive media following ASTM-G129.²⁶  SSRT experiments were conducted with a strain rate of 1 × 10⁻⁶ s⁻¹ to increase the number of environmental interactions.²⁷  The electrochemical tests conducted during the straining of the sample were performed using a three-electrode configuration cell setup with a Gamry potentiostat. The reference electrode (RE) used in this test was a saturated calomel electrode (SCE), the counter electrode (CE) used was a graphite rod, and the working electrode (WE) was the Type 2001 LDSS reinforcing bar. Three different chloride concentrations were tested 0 wt%, 4 wt%, and 8 wt% Cl⁻ by means of CaCl₂ additions to the simulated concrete pore solution (SCPS, pH 12.6) made out of a saturated Ca(OH)₂ aqueous solution. All tests were performed in triplicate to ensure reproducibility.

Prior to SCC testing, on separate samples without any applied stress, cyclic potentiodynamic polarization (CPP) tests were performed to better understand the electrochemical kinetics of the LDSS in the three different solutions (0 wt%, 4 wt%, and 8 wt% Cl⁻ SCPS) at RT in aerated conditions. Initially, open-circuit potential (OCP) was monitored, thereafter the polarization potential scan ranged from −0.3 V_OCP to +1.0 V_OCP at a potential scan rate of 0.1667 mV/s, in accordance with ASTM G61-86.²⁸  In addition, Mott-Schottky analysis was performed, also without any applied stress, to study the semiconducting properties of the passive film, in a potential range from –1.2 V_SCE to +0.6 V_SCE, with a potential step of 50 mV. The frequency for the capacitance measurements was 1,000 Hz, which corresponds to the relaxation process associated with the electrochemical response of the passive film based on the literature for SS.^29 -30 

The sequence for the different electrochemical test was the following: first OCP was monitored until a steady state value was reached, and then electrochemical impedance spectroscopy (EIS) measurements were performed. The EIS measurements were recorded at the OCP, in a frequency range of 10⁵ Hz to 10⁻² Hz with an applied 10 mV AC excitation signal and at a step rate of 5 points per decade. Both OCP and EIS experiments were repeated every 3 h, where in between repetitions, the current density was monitored during a potentiostatic hold with an applied potential (E_app) of +200 mV_OCP, which was kept constant throughout the entire SSRT. The E_app was selected based on the CPP to shift the LDSS reinforcing bars to the active dissolution region, allowing monitoring of the pit-to-crack transition. The EIS was not measured under the E_app, the SSRT was put on hold during the measurements to avoid unwanted creep.

For the microstructure analysis, the samples were epoxy mounted and polished to mirror finishing by SiC paper and diamond powder (1 μm). The etchant solution used to reveal the microstructure contained 40 wt% NaOH, the samples were exposed for 5 s at an applied potential of 3 V.³¹  Microstructural study was conducted on both, transversal and rolling directions. The metallographic study was performed using scanning electron microscopy (SEM) in a Tescan Lyra 3 XMU^†. Finally, x-ray diffraction (XRD) analysis was performed using a Rigaku SmartLab-3kW^† x-ray diffractometer, with a Cu target (K_α =1.5406 Å), and a scan step of 2°/min over the 2θ range of 35° to 100°. The surface morphology of specimens was studied via SEM. The SEM analysis was performed in secondary electron mode (SE) at an accelerating voltage of 15 kV and at a working distance of 10 mm. In addition, local compositional analysis was obtained by energy dispersive x-ray spectroscopy (EDS) technique.

The microstructure of Type 2001 LDSS reinforcing bars, cross section, and longitudinal rolling direction can be seen in Figure 1, where the γ-phase grains are embedded in the α-phase matrix.³²  In the transversal direction, it can be seen that the size of γ-phase grains is larger than that of α-phase (see Figure 1[a]).²³  Figure 2 depicts the XRD pattern of the as-received (AR) sample, where the diffraction peaks were composed of body-centered cubic (bcc) α phase (JCPDS No. 06-0694) and face-centered cubic (fcc) γ phase (JCPDS No. 33-0397).^33-34  The ratio of α/γ phase was quantified by the integration of the intensity peaks of the respective phases giving a 55% for α phase and 45% for γ phase. The obtained phase ratio was similar to the one from LDSS 2304.³²  The lattice parameter for α phase (a_α) and γ phase (a_γ) were calculated to be 0.287 nm and 0.360 nm, respectively.

The CPP of Type 2001 LDSS immersed in the three different environments (0 wt%, 4 wt%, and 8 wt% Cl⁻ SCPS) is presented in Figure 3, where the detrimental effect of the chloride content in the SCPS is clearly seen. The free-chloride solution at high alkaline pH developed a stable and protective passive film, seen by the negative hysteresis in the reverse scan of the CPP plot. The corrosion potential (E_corr) and the corrosion current density (i_corr) were –248 mV_SCE and 5.1 × 10⁻⁸ A/cm², respectively (see Table 2). In chloride-contaminated SCPS solution with 4 wt% Cl⁻, despite having a close E_corr value compared to the free-chloride (−254 mV_SCE) as well as a similar i_corr (6.3 × 10⁻⁸ A/cm²), and showing similar anodic kinetics; the Type 2001 LDSS immersed in this environment developed a pitting potential (E_pit) at 500 mV_SCE. The 4 wt% Cl⁻ concentration in the environment made for the passivity breakdown once the E_pit was reached, developing a positive hysteresis and returning to the initial E_corr and i_corr. Finally, the 8 wt% Cl⁻, while it also started at a similar E_corr to the other two solutions (−268 mV_SCE), the anodic kinetics shifted toward higher current densities (i_corr = 1.8 × 10⁻⁷ A/cm²), a sign of the higher susceptibility to pitting with the increase in chloride concentration. Like the change from free-chlorides to the 4 wt% Cl⁻ SCPS, the increase of chlorides further lowered the E_pit. Even more, the positive hysteresis returned to even lower potentials and higher current densities, both showing an increase in pitting susceptibility. The analysis of the repassivation potential (E_rep) showed that the ability to repassivate for the 4 wt% Cl⁻ was not notable, as it had a potential value close to the E_corr, thus showing substantial pitting susceptibility. Even more, at 8 wt% Cl⁻ the E_rep is found below the E_corr, indicating that no repassivation was achieved and that 8 wt% Cl⁻ was above the chloride threshold for 2001 LDSS reinforcements in an alkaline environment (pH 12.6).

The semiconducting properties of the passive film were analyzed by the Mott-Schottky model, which presents a relationship between the capacitance of the space charge (C) of the passive film and the potential of the applied electrode (E) given by Equation (1):

where E_FB is the flat-band potential, k is the Boltzman constant (1.38 × 10²³ J/K), T is the absolute temperature, q is the charge of the electron (1.602 × 10⁻¹⁹ C), ε_o is the vacuum permittivity (8.854 × 10⁻¹⁴ F/cm), ε is the dielectric constant of the passive film (15.6 F/cm for SS passive film),³⁵  and N is the carrier density.^36-37  As it can be seen from Figure 4 the increase in chloride content decreased by two orders of magnitude the C⁻², thus showing a thinner and more defective passive film. While the slopes n₂ and n₃ remained in the same range for all conditions (see point P), both chloride concentrations shifted the n₁ slope toward more cathodic potentials (see Table 3). This indicates that the chlorides hinder the formation of the passive film, preventing the formation of a stable Cr-rich passive film. For each of the slopes the corresponding N value was calculated, obtaining a similar order of magnitude between N values for each chloride concentration, which was going from low to high chloride content: 10¹⁷ cm < 10¹⁹ cm < 10²⁰ cm⁻³. The higher the N the more doped the passive film is with defects, thus providing less protection against chloride pitting corrosion. Accordingly, the E_FB became more negative with the chloride addition, hence reaching the anodic potential sooner for the donation of electrons.

The monitoring of the current density over the straining of the sample is shown in Figure 5. The solution without chlorides showed current densities in the order of µA/cm², initially decreasing for stresses lower than the yield strength (σ_y) due to the passive film formation, and stabilizing for the first 5,000 s, and then an increase was seen as the applied load approaches the ultimate tensile strength (σ_UTS). After reaching the σ_UTS, the current density stabilized. Finally at  the failure point, where purely mechanical fracture occurred, a sudden increase in the current density was recorded (see Figure 5[a]). The sample exposed to 4 wt% Cl⁻ revealed current densities of one order of magnitude higher (10 µA/cm²), which abruptly duplicated its value once the σ_y was reached. Afterward, the passivity breakdown was developed due to the combination of the chloride attack and the start of the dislocation movement. From that point forward, the current density increased for stress values close to the σ_UTS, where a faster increase in current density was seen until the failure point surpassed the 0.1 mA/cm² (see Figure 5[b]). Lastly, the sample exposed to 8 wt% Cl⁻ rapidly increased its current density over one order of magnitude before reaching the σ_y, showing the high chloride susceptibility of the Type 2001 LDSS reinforcement (see Figure 5[c]). Once the applied stress value got near the σ_UTS, the current density stabilized at 0.45 mA/cm² until failure, showing a pure dissolution process.

The Nyquist plots of 0 wt% Cl⁻ showed the highest impedance values from all three chloride conditions, with a slight improvement from the preload to the 75% σ_y, coinciding with the decrease in current density seen in the monitored electrochemical current transient. After the σ_y, due to the beginning of the dislocation movement, the mechanical stimulus decreased the corrosion protection of the LDSS sample, reducing the impedance values but staying in the same order of magnitude (10⁴ Ω·cm²) (see Figure 6[a]). The 4 wt% Cl⁻ reduced its impedance value with the straining, reducing the diameter of the second semicircle (low frequencies), as well as being more depressed than in 0 wt% Cl⁻ (see Figure 6[b]). Similarly, the 8 wt% Cl⁻ showed a similar trend as 4 wt% Cl⁻, reducing its lower global impedance with the straining, however, it started reducing at 10³ Ω·cm² (see Figure 6[c]). This indicated a lower protective character of the film, thus higher chloride susceptibility. All of the Nyquist plots presented two semicircles, which were related to the two time constants: passive film (high frequencies) and electrochemical double layer (low frequencies). The EIS data were fitted to a hierarchically distributed electrical equivalent circuit with two time constants, represented by a constant phase element (CPE) and resistor (R//CPE) (see Figure 6[d]).³⁸  The EIS fitting results can be seen in Table 4, where the total error was below 10% with a chi-squared (χ²) value below 10⁻⁴, proving the goodness of the fitting. In addition, Kramer-Kronig transforms were performed to show the robustness of the data, proving that there were no artifacts. The 0 wt% Cl⁻ exhibited the highest impedance value for both resistors, passive film (R_film) and charge transfer (R_ct), as well as having the lowest pseudo-capacitance. The absence of chloride ions made for the slight increase of the R_ct with immersion time, until high stresses were reached. The addition of the 4 wt% Cl⁻ resulted in a thinning of the passive film thickness, and consequently an increasing over an order of magnitude of the pseudo-capacitance of the passive film (Y_film) was found, thus resulting in the R_film values to rapidly decrease with the applied stress. For the Type 2001 LDSS reinforcing bar exposed to 8 wt% Cl⁻, the Y_film values remained in the mS/cm² proving the highly defective passive film that was formed onto the surface, therefore also showing low values of the R_film.

After straining to failure, the fracture morphology was analyzed by SEM. The sample tested in 0 wt% Cl⁻ presented a pure ductile fracture with a shape resembling a strained cone (see Figure 7[a]). Common ductile fracture by coalescence of dimples was seen through the fracture surface (see Figure 7[b]).³⁹  Some minor ductile overload areas were also present close to the center of the sample (see Figure 7[c]).

Once the 4 wt% Cl⁻ was added to the solution, as seen in the worsening of the corrosion performance by the electrochemical monitoring, Type 2001 LDSS altered fracture mode to brittle fracture. Clear signs of brittle fracture were seen at low magnifications, having a crack from one side of the cone lip to the center (see Figure 7[d]). The SEM analysis suggested a predominant cleavage fracture with some mixed cleavage/dimple rupture (see Figure 7[e]).⁴⁰  Figure 7(f) depicts numerous cleavage fracture facets on α phase, surrounded by some minor voids of different morphologies as well as cracks.^41-42  The cleavage facet showed higher chromium and lower nickel content as revealed by EDS, thus suggesting a higher density of α phase.⁴³ 

For the highest chloride concentration (8 wt% Cl⁻), the fracture morphology had more brittle features, including greater areas of ferrite cleavage fracture facets over the ductile overloads, and a well-defined crack passing through the center of the sample, going from side to side with a width of 83 µm (see Figure 7[g]). The surface close to the crack experienced a mixed cleavage/dimple rupture, while as the distance increased, ductile overload was formed instead. A large surface with brittle fracture, where crack coalescence was developed, could be seen along the broken sample; cracked cleavage facets resembled the aggressiveness of the environment for SCC development (see Figure 7[h]).⁴⁴  As previously seen in the 4 wt% Cl⁻ inside the crack, the preferential dissolution of the α phase was also developed in the form of a bundle of parallel γ phase left of both sides of the crack lips (see Figure 7[i]).¹⁸ 

As seen with the CPP analysis, the E_pit reduced with the increasing chloride content and increase in the maximum passive current density (i_ss), proving the higher pitting susceptibility and lower stability of the passive film. The higher activity of the chlorides did not enable the formation of a stable passive film, as the constant generation of vacancies from the Cl⁻ was at a higher rate that the annihilation of the same by the metal (passive film growth). Based on the point defect model (PDM) proposed by Macdonald, the critical vacancy concentration to promote passivity breakdown can be calculated and further corroborated with Mott-Schottky measurements.²⁴  The critical areal cation vacancy concentration (ξ) and the rate of annihilation of cation vacancies at the barrier layer/electrode interface (J_m) are inversely related and can be obtained empirically from the slope of the relationship E_pit/scan rate, which for these calculations, a value of 1 will be taken from literature²⁴  (see Equation [2]):

where χ is the barrier layer stoichiometry of Cr₂O₃ (3), α is variable from a potential dependence with a value of 0.22 based on literature, R is the ideal gas constant (8.314 J/mol K), T is the temperature, and F is the Faraday constant (96,487 C/mol). The J_m can be calculated from the i_ss following Equation (3):

where i_ss is the maximum passive current density, N_Av is Avogadro’s number (6.023 × 10²³ mol⁻¹), and χ and F have been defined above.

The J_m calculations for 4 wt% and 8 wt% Cl⁻ yield 7.24 × 10¹³ cm²/s and 3.21 × 10¹⁴ cm²/s, respectively, thus the ξ values are therefore calculated to be between 10¹³ cm^–2 and 10¹⁴ cm⁻². With the N₂ calculated from the Mott-Schottky, which corresponds to the donor point defect density, the ξ can also be estimated by dividing the N₂ by the thickness of the passive film (1 nm for stainless steel),²⁴  giving an approximated value between 10¹³ cm⁻² and 10¹⁴ cm⁻². The ξ comparison between the PDM and Mott-Schottky calculations shows that 4 wt% Cl⁻ does have some protection against the chloride attack, but its efficiency is severely restricted as the obtained value was near the ξ under no applied load. If stress is exerted, the passive film would thin due to the strain, decreasing the ξ and, making the annihilation flux of vacancies not enough to grow and repassivate the passive film, triggering the passivity breakdown. The 8 wt% Cl⁻ shows that the protective properties of the passive film are not enough for that high chloride content, resulting in a doped and unstable passive film even without the influence of the stress. Both results are in good agreement with the monitored electrochemical current transient, wherein for the 4 wt% Cl⁻ the needed stress for passivity breakdown was the σ_y, and for the 8 wt% Cl⁻ it was triggered after the preload.

To further corroborate the influence of the chloride content with the susceptibility of destabilizing the passive film and promoting pitting, the pitting susceptibility factor (PSF) was calculated (see Equation [4])⁴⁵ 

The PSF is a probabilistic variable that takes into account the E_corr, the E_pit, and the E_rep, which was the potential of the backward scan that had a corresponding current density of 1 µA/cm².⁴⁶  PSF values below 0.5 indicate low pitting susceptibility, while values equal or greater than 1 indicate high pitting susceptibility. Table 2 gathers the electrochemical results of the CPP analysis and shows that as the chlorides are added to the environment, the PSF immediately reached the unity and further increased to 1.1 with the 8 wt% Cl⁻, thus both chloride environments are enough to develop pitting. The development of the E_pit, the positive hysteresis, and the PSF reaching the unit, all together showed the pitting susceptibility of Type LDSS 2001 reinforcement to the environment, indicating that the 4 wt% Cl⁻ was above or at the chloride threshold. This is in good agreement with the PDM calculations, showing that the number of defects found on the passive film at 4 wt% Cl⁻ were near the ξ, while the 8 wt% Cl⁻ exceeded, a reason for the PSF higher than the unity.

While the γ phase would have been able to withstand that chloride amount and repassivate, the α phase, which is known to have lower chloride threshold values, was the detrimental factor influencing the premature nucleation of the pitting.⁴⁷  The higher corrosion potential of γ phase compared to α phase, due to the Ni and N constituents, promoted a lower oxidation rate of the γ phase, making the α-phase grains the anodic region.⁴⁸  The proof for the preferential dissolution of the α phase over the γ phase was a pit analysis after the CPP, which is presented in Figure 8. Looking at the same pit from both cross section (Figure 8[a]) and longitudinal rolling (Figure 8[b]) directions, it could be seen that the α phase was preferentially dissolved. Island-like shapes of γ phase can be seen in the cross-section direction, while in the rolling direction, the dissolution propagated through the α-phase matrix and the α/γ boundaries, stopping at the γ phase due to the higher corrosion protection properties, making the γ-phase work as the cathode. The findings coincide with the results seen in the literature for chloride-contaminated environments in high to neutral pH, where the lower E_corr of α phase promoted its preferential dissolution given the lower Ni content, thus becoming active as the mixed potential of UNS S32001 polarized above that threshold.^9,18 

The analysis of the passive film evolution was done by categorizing each of the individual transients (current peaks) to one of the following three types of transient (see Figure 5[d]):⁴⁵  The first type of transient (Type I) is categorized by a mild increase and a sudden drop in current, this is related to a slow rupture of the passive film compared to the fast recovery of the same (passive film growth). The second transient (Type II) starts with a vertical increase of current in a very short period of time, followed by a slower decrease in current, thus indicating for this particular case that the passive film breakdown is much faster than the repassivation. Finally, the third transient (Type III) is in a scenario where both rupture and growth of the passive film have a close value of slope and similar times, thus defining a transient state. Figure 5(d) shows the percentage of each type of transient for 0 wt%, 4 wt%, and 8 wt% Cl⁻. The transient deconvolution for 0 wt% Cl⁻ showed a low density of transients with low amplitudes, where the majority of those measured transient were Type I and Type III for stresses below the σ_UTS, while higher stresses destabilized the passive film, due to the thinning of the passive film, increasing the amplitude and duration times in addition to the density increase of Type II transient (see Table 5). The passivity breakdown under the 0 wt% Cl⁻ condition was reached purely by mechanical failure due to the mismatch of the mechanical properties of the passive film and reinforcing bars generating residual stresses at the interface, as well as by the stretching of the passive film until the maximum elongation it could stand was reached. Once the chlorides were added (4 wt% Cl⁻) the density of transients increased for all types, however, the majority of transients were of Type II, coinciding with the high pitting susceptibility that 2001 LDSS has. The repetition of Type II transient was accentuated as well as its increase in amplitude, a sign of metastable pitting. Nevertheless, once the σ_y was reached the dislocation movement began and the passivity breakdown was developed, as seen by the rate of transient and the larger duration times. The increase in stress was part of the responsible for the shift from metastable to stable pitting, as higher loads tend to increase the amplitude of Type II transients. In the most aggressive condition (8 wt% Cl⁻), the density of the transient severely increased for Type II along with its amplitude and duration times. Despite the increase in a number of recorded transients, Type I transient decreased. Passivity breakdown was developed before the σ_y, which prematurely increases the pit growth, thus indicating that those anodic sites would be the preferential flaws for exceeding the critical stress intensity factor to nucleate the first crack. After reaching the σ_UTS, the amplitude and density of Type II transient was dramatically boosted, giving place to the cracking process.

To further analyze the passive film evolution, the effective capacitance of the passive film (C_eff,film) and the electrochemical double layer (C_eff,dl) were obtained correcting the values of both CPE, as these elements take into account the nonhomogeneity of the area being electrochemically tested, including roughness, heterogeneities, and texture. To calculate the C_eff,film the equation from Mansfeld and Hsu was used, where the frequency is the value where the imaginary impedance has a maximum (see Equation [5]):⁴⁹ 

In the case of the C_eff,dl, the equation proposed by Brug, et al., was used (see Equation [6]):⁵⁰ 

With the corrected C_eff,film value, the thickness of the passive film (d_film) could be estimated using Equation (7):

where A is the exposed surface area (see Table 4).²⁰  As can be seen by the growth of both C_eff with the straining of the sample, the addition of the chloride ions further enhances the C_eff growth due to the depassivation effect of chlorides. As shown in the current transients, close to the passivity breakdown, a decreased d_film value around 1 nm is found, which indicates that no stable protective film is formed. The straining of the samples immersed in 0 wt% Cl⁻ made for the passivity breakdown occur by purely mechanical stimulus at the “Failure,” while 4 wt% Cl⁻ reduced its d_film to 1 nm after the σ_y, and finally 4 wt% Cl⁻ reached the 1 nm before the σ_y. The analysis of the d_film for the three chloride concentrations reflected the changes from the current transient, where the passivity breakdown was seen when the amplitude, duration, and density of Type II grew over the other type of transients.

The Bode plots for all chloride conditions as a function of the applied stress are represented in Figure 9. The decrease in phase angle (θ) with the increasing stress is associated with the damage taken by the passive film, which in the case of the 0 wt% Cl⁻ was purely done by thinning by straining (see Figure 9[a]).⁵¹  It was after the σ_UTS that the peak with the max θ (θ_max) shifted toward lower frequencies, a sign of the passivity breakdown, corresponding to the higher current densities registered. Surpassing the σ_y in the 4 wt% Cl⁻ triggered the pit-to-crack transition, developing another process, seen in the formation of a second peak at low frequencies, which got more defined for values close to the failure (see Figure 9[b]). It is at low frequencies (0.1 Hz to 0.01 Hz) that the cracking process can be seen and evaluated.⁵²  The 8 wt% Cl⁻ presented the low-frequency peak already at the preload, which then got better defined at the σ_y, the point from which the current density stabilized during the current transient (see Figure 9[c]). The shape transformation of the θ in between the preload and the σ_y can be understood as a transient state, responsible for the passivity breakdown given the high current densities in the order of 0.1 mA/cm².

To further analyze the θ, the evolution of the phase angle shift (Δϕ) as a function of the applied load is plotted in Figure 10, where the Δϕ is the subtraction of the preload to the other applied load points. Figure 10(a) shows the Δϕ for the 4 wt% Cl⁻, where after the σ_y is reached a relevant increase of the Δϕ can be seen, denoting the passive film breakdown and the start of the cracking. This is further shown by the θ_max shift at intermediate frequencies toward higher frequencies, while the low-frequency peak gets developed and shifts toward even lower frequencies. The severe decrease of the θ_max corresponds to the crack propagation process, which was notable after the σ_y, coinciding with the steep increase in current density seen in the current transients. The good agreement between the Δϕ and current transient shows the crack propagation becoming the dominant process. Figure 10(b), as seen in the Bode plot for 8 wt% Cl⁻, had main peaks after the current stabilization at the σ_y, however, the pit-to-crack transition triggered before the reaching the σ_y, seen by the abrupt increase in Δϕ and the formation of a well-defined peak at low frequencies. This peak at 0.1 Hz, representing the crack propagation, denotes a deep crack due to the low frequency and the good definition of the peak.⁵³  In addition, the low-frequency peak continued to shift toward lower frequencies, as well as the high peak shifting toward higher frequencies, both sing of the crack propagation process.

Figure 11 shows the crack path of the sample tested in 4 wt% Cl⁻. For DSS with lamellar dispersed structure, the cracking on the brittle fracture surface was developed in two steps: first the nucleation of the cracks in the α phase, which propagated up to the α/γ interface at the γ-phase islands; second the propagation through the γ phase, which was controlled by the stretching of the austenite ligaments (see Figure 11[a]).⁴¹  The majority of the cracking followed either the α/γ interface (IG-SCC), or the α phase (TG-SCC), coinciding with the anodic dissolution mechanism, where the most active path is the precursor of the cracking. Nevertheless, transgranular cracks in the γ phase were also seen, which correspond to the higher stresses that γ phase can withstand, eventually reaching the K_ISCC, and promoting TG-SCC by slip-step dissolution (see Figure 11[b]).¹⁶  The mechanism adjusts to the discontinuous cracks seen in Figure 11(b), where the crack path diverges from the α/γ interface and breaks the γ phase to further continue at the next α/γ interface.

The cracking morphology of 8 wt% Cl⁻ was also predominant on IG-SCC through the α/γ interface, but also on cracking through the α phase (see Figure 11[c]). Both crack path morphologies are in good agreement with the literature, as γ phase acted as a crack arrestor, redirecting the crack propagation to the α phase or grain boundary.¹⁹  The higher cracking susceptibility of the α phase is due to its bcc crystal structure and corresponding slip system,⁵⁴  where the crack at the α/γ interface propagated through the γ phase was due to its localized high-stress levels. These high-stress levels could be due to the deformation mechanism upon straining that γ-phase crystal structures undergo, like strain-induced martensitic transformation which has been previously reported on both DSS and LDSS.⁵⁵  Figure 11(d) shows the morphology of a crack propagating through the α/γ interface, where the γ-phase grains adjacent to the crack present a similar pattern to that of strain-induced martensite seen in the literature.⁵⁶  The microstructure of the γ-phase grains presented a lath-shaped martensite (darkest gray inside the γ phase) with different orientations (see Figure 11[d]).⁵⁶ 

In order to compare the SCC susceptibility between both chloride conditions, the maximum crack velocity (v_max) was estimated from the maximum crack length found on the samples (I_max), the time to failure (t_f), and the time to reach σ_UTS (t_UTS) with Equation (8):⁵⁷ 

The l_max of the 4 wt% Cl⁻ was found to be 126 µm, with a time lapse of 14,285 s, giving a v_max of 8.8 × 10⁻⁹ m/s, while the 8 wt% Cl⁻ had a 734 µm l_max with an associate time lapse of 10,073 s making for a v_max of 7.2 × 10⁻⁸ m/s. Doubling the chloride content, which as seen from the activity coefficient study did double the activity of chlorides and increased 50% of the oxygen activity, made for an increase of one order of magnitude of the v_max. This is well aligned with the high chloride susceptibility that Type 2001 LDSS showed from the electrochemical analysis.

This study presented an investigation of the SCC behavior of low nickel Type 2001 (UNS S32001) LDSS reinforcing bar in concrete pore solution-containing chlorides, by means of microstructural and fractographic characterization and electrochemical testing under a mechanical stimulus. From the results herein, the following conclusions may be made:

• The PSF values close to the unity corroborated the high pitting susceptibility of Type 2001 LDSS reinforcing bar, revealing a chloride threshold for a highly alkaline environment (12.6 pH) below 4 wt% Cl⁻. This was corroborated by the calculation of the critical areal cation vacancy concentration (ξ) for passivity breakdown via Mott-Schottky analysis, giving a threshold value close to the 10¹³ cm⁻². For higher chloride content (8 wt% Cl⁻), stable pitting was developed, seen in the PSF greater than the unity, and the lower rate of annihilation of cation vacancies at the barrier layer/electrode interface (J_m) than the ξ, not enabling the passive film growth.

• The doping of the passive film by an increase in defect density, in combination with the more cathodic E_FB, enhanced the electron donation from the passive film, forcing a premature rupture of the passive film which triggered the crack initiation once the σ_y was reached for the chloride threshold of 4 wt% Cl⁻, while 8 wt% Cl⁻ developed the passivity breakdown before the σ_y, promoting the crack propagation at low stress levels due to the lower protection capabilities of the passive film.

• The increase in the rate of Type II transients and their longer duration times once the σ_y was reached, correlated with passivity breakdown. While an increase in transient amplitude was related to crack propagation. Following passivity breakdown, the density of Type I transient was significantly reduced to almost negligible for 8 wt% Cl⁻, corroborating the low film growth capabilities under the chloride attack. Passive film thickness evolution corroborated the passivity breakdown development, where d_film values close to the 1 nm, triggered pitting. After the passivity breakdown, the d_film values kept decreasing, supporting the pit-to-crack transition without possible repassivation, as was seen with the PSF and PDM analysis.

• Passivity breakdown and crack initiation in 4 wt% Cl⁻ were triggered for stresses above the σ_y, developing a low-frequency peak (0.1 Hz to 0.01 Hz) in the θ angle from the Bode plot, responsible for the cracking process. For chloride content above the chloride threshold, the pit-to-crack transition triggered before the σ_y, severely decreasing the phase angle shift (Δϕ) as well as developing a well-defined peak at low frequencies.

• The preferential dissolution of α phase, which served as the so-called “anodic region” due to a lower N_i and N concentration, was arrested at the α/γ boundaries because of the lower dissolution rate of γ phase. The α phase worked as nucleation sites for the pits, which once reached a critical stress crack with predominant IG-SCC morphology through the α/γ interface. The doubling of the chloride threshold to 8 wt% Cl⁻ induced an increase of one order of magnitude in the crack propagation rate υ_max to 7.2 × 10⁻⁸ m/s, promoting a crack with a maximum length almost six times deeper than the one from 4 wt% Cl⁻.

• The mechanism responsible for the SCC failure of UNS S32001 reinforcement in an alkaline medium with chloride contamination at low temperature was anodic dissolution for the α phase (due to its higher corrosion susceptibility), making it the most active path for the cracks and propagating intergranularly by the α/γ interface. While for the γ phase, slip-step dissolution promoted TG-SCC when a γ-phase grain with high stresses was reached.

The authors acknowledge funding from Firestone Research Grant 639430, and The University of Akron Fellowship Programs FRC-207160 and FRC-207865. The authors acknowledge the technical support and facilities from the National Center for Education and Research on Corrosion and Materials Performance (NCERCAMP-UA), the College of Engineering and Polymer Science, and The University of Akron.