A. Identification of the corrosion initiation site and effect of the σ-phase

From the surface morphology of the steel weld after the immersion test in reference to ASTM G48, ²⁹ which can be illustrated in Fig S1, it is found that the corrosion damage is too severe to analyze the corrosion mechanism. For this reason, an electrochemical polarization test involving the pre-exposure process was used using a methodology reported elsewhere. ^{Reference Kim and Hong28} After the electrochemical evaluation, the surface was analyzed by FE-SEM in back scattered electron (BSE) mode, as shown in Fig. 1(a). It is clear from the analysis that the localized corrosion was initiated preferentially around the 2nd phase particle, which is expressed as a bright color in the interdendritic area. For a clearer understanding, the electron probe microanalysis (EPMA) was additionally performed, and the results are demonstrated in Fig. 1(b). It reveals that the 2nd phase particles characterized as a typical σ-phase with a tetragonal structure in a previous investigation ^{Reference Kim and Hong28} were enriched by Cr and Mo and depleted by Ni. Considering these results, the localized corrosion attack observed in the weld is caused mainly by the difference in composition of Cr and Mo between the σ-phase and matrix. This can be supported by the 3DAP analysis, which was found in Fig. 2. Around the σ-phase, the Cr and Mo depleted zones were clearly formed. The depletion phenomenon was somewhat different from that determined by TEM in a previous investigation ^{Reference Kim and Hong28,Reference Kim and Hong30} in that the Mo depletion phenomenon was identified accurately in the present study. As the diffusion length of the solute element is proportional to the square root of the diffusivity of the element, ^{Reference Kim and Hong28,Reference Porter and Easterling31} the slow diffusion kinetics of Cr and Mo in the γ-matrix could lead to restricted diffusion length, resulting in a narrow depleted region.

FIG. 1.

(a) FE-SEM image exhibiting the surface morphology of the specimen obtained by the electrochemical technique and (b) EPMA around the σ-phase.

FIG. 2.

(a) Sample preparation for the analysis of 3DAP and (b) atomic distribution of Fe, Cr, and Mo around the σ-phase.

To confirm the detrimental influence of the σ-phase on the resistance to localized corrosion of the steel weld, three types of samples with different volume fractions of σ-phase in the microstructure were evaluated using an electrochemical polarization test, and this can be found in Fig. 3. The fractions of σ-phase in the microstructure after the heat treatment conducted at 1180 °C for 30 s, 60 s, 90 s are 5.16, 3.25, 2.27, respectively, meaning that the σ-phase is gradually dissolved by the additional heat treatment. From the polarization measurement, it is evident that the decrease in the fraction of σ-phase in the microstructure by the additional heat treatment resulted in higher corrosion resistance.

FIG. 3.

Potentiodynamic polarization evaluation of the three types of steel weld with different volume fraction of σ-phase in the microstructure.

B. Effect of oxide inclusion

The inclusion contained in the steel weld has a spherical shape with a diameter of up to 15 µm, which can be identified in Fig S2 showing the optical microscopy and SEM images of the oxide inclusion distributed in weld metal of the steel with low cleanliness. This can also be found in our previous report. ^{Reference Porter and Easterling31} The EDS analysis suggested that the inclusions are mostly (Mg, Al)-oxide, which is suspected to be a spinel-type inclusion of MgAl₂O₄. On the other hand, the inclusion was barely observed in the steel with high cleanliness. The mean area fraction of the inclusion in the two tested steels (high cleanliness steel: 0.02%, low cleanliness steel: 0.4%) was measured quantitatively using an image analyzer. From the EDS analysis, it appears that the composition of the inclusions was similar in the two steels tested. Considering the fact that the oxide inclusions were distributed in base metal as well as weld metal, it is obvious that the formation of inclusion was not associated with the welding conditions, and the manufacturing process of the steel was not properly controlled and optimized.

It is considered that the metallurgical inhomogeneity in the stainless steel, caused by the presence of coarse-sized inclusion, leads to increased susceptibility to pitting corrosion of the steel. Figure 4 presents the potentiodynamic polarization curves of the two steels tested in an acidic chloride environment. As the steels are classified as high-alloyed austenitic steel, the overall trend of the polarization curves showed a similar pattern. However, the passivation current density at which the potential exceeds 0.2 V_SCE between the two tested steels was different. This suggests that the passive film formed on the steel containing a higher density of coarse-sized inclusion is less stable, and the steel is more susceptible to localized corrosion than the steel with high cleanliness. It is also suggested that the pitting corrosion associated with the inclusion can be initiated at a potential of approximately 0.2 V_SCE during anodic polarization.

FIG. 4.

Potentiodynamic polarization curves of the two tested steels with high and low cleanliness, respectively. ^{Reference Porter and Easterling31}

As mentioned in the Introduction, the detrimental effect of oxide inclusion on the localized corrosion of various types of stainless steel has been explored, and several theories have been provided to explain the precise role of the inclusion on the corrosion initiation behavior of the steel. Nevertheless, it is unclear even today due mainly to the difficulty in analyzing a passive film with only a few nanometers thick. Therefore, the present authors have focused our attention on clear observations of the cross-sectional morphology of the passive film around the inclusion, by HR-TEM, and a reasonable mechanism is proposed based on the experimental evidence.

Figure 5 presents the surface and cross-sectional morphology of the inclusion contained in the steel with low cleanliness employing FIB and SEM. Figures 5(a) and 5(b) show the morphology of the steel before the electrochemical corrosion test; Figs. 5(c) and 5(d) present the morphology of the steel after the corrosion test. Firstly, it is interesting to observe a pre-existing cavity in the interfacial area between the inclusion and steel matrix even before the corrosion test. This could be caused by the difference in the thermal expansion coefficient between the spinel type of MgAl₂O₄ (8–9 × 10⁻⁶ °C⁻¹) ^{Reference Kaprálik32} and the austenitic stainless steel (19–20 × 10⁻⁶ °C⁻¹). ^{Reference Khatak and Raj33} The pitting was concluded to have initiated around the inclusion from the observation of the inclusion after the corrosion test. In addition, it can be inferred that the pitting could propagate more easily via the pre-existing cavity in the interfacial area. Although the cross-sectional morphology exhibiting pitting corrosion propagated around the inclusion cannot be observed clearly due to the presence of Pt coated for the subsequent HR-TEM observations, it is evident that the pitting hole with a large diameter of approximately 500 nm is open to aggressive ionic species.

FIG. 5.

Surface morphology of the inclusion contained in the sample with low cleanliness (a) before and (c) after the corrosion test, and cross-sectional view around the inclusion (b) before and (d) after the corrosion test. ^{Reference Porter and Easterling31}

The cross-sectional morphology was observed utilizing the HR-TEM to gain further insight into the property of passive film around the inclusion. Figure 6(a) shows that a thin passive film with an approximate thickness of 3 nm is normally formed on the steel surface. On the other hand, Figs. 6(c) and 6(d) show that the film thickness decreases as it approaches the inclusion. In addition, the EDS analysis showed that the compositional ratio of Cr and O with respect to Fe(Cr(0.32)*O(0.27)/Fe(0.40) = 0.22) in the passive film around the inclusion was much lower than that (Cr(0.41)*O(0.39)/Fe(0.19) = 0.84) in the passive film formed normally on the steel surface. These suggest that the 3 nm thick passive film cannot be continued homogeneously on the steel surface due to the presence of inclusions. Therefore, the local region covered by a very thin film with a smaller Cr content can be attacked easily by aggressive ionic species such as Cl⁻ ions. The resultant pitting propagation of steel in the chloride environment can be observed clearly, as shown in Fig. 6(e). The restricted growth of the passive film around the inclusion in the steel weld could be attributed mainly to the hindrance of Cr-diffusion around the inclusion to the surface of steel weld. A simple schematic diagram of the proposed mechanism is depicted in Fig. 6(f). As this mechanism is based partly on our speculation, further study with supporting evidence will be needed.

FIG. 6.

(a) Cross-sectional view of the passive film formed on the steel; (b) cross-sectional morphology of the inclusion observed by SEM; (c) magnified view of the cross-sectional morphology, marked in (b); (d) magnified view of the passive film, marked in (c); (e) pitting morphology around inclusion after corrosion test; (f) schematic diagram of local corrosion around the inclusion. ^{Reference Porter and Easterling31}

C. Effect of alloying elements of Cu and Al

To examine the effects of Cu on the resistance to localized corrosion of the steel weld, three different samples containing 0.1, 0.183, and 0.37 wt% Cu, respectively, were tested by electrochemical polarization measurement. Smuk et al. ^{Reference Smuk, Nenonen, Hänninen and Liimatainen34} have indicated that Cu particles in the stainless steel could prevent the motion of the σ/γ₂ interface boundaries, and the δ → σ phase transformation could be suppressed. However, the addition of Cu up to 0.37 wt% had no effect on the reduction in the volume fraction of the σ-phase, which can be found in Fig. S3. Figure 7 presents the potentiodynamic polarization measurements of the three different samples with different amounts of Cu. It is evident that the addition of Cu to the steel weld decreases the localized corrosion resistance in a chloride environment. The current spike for the formation and repassivation of meta-stable pits increased with increasing Cu content in the experimental alloys. Similar behavior was also reported by Jeon et al. ^{Reference Jeon, Kim, Lee, Park, Kim, Kim and Park24} From their study, with the addition of Cu exceeding 1 wt%, the number and area of numerous oxides and oxy-sulfides in the microstructure increase, which results in a significant decrease in the resistance to pitting corrosion. From the present study, however, a smaller amount of Cu below 0.4 wt% has no significant effect on the fraction of oxide inclusion. As the electrochemical potential of Cu is much higher than that of Fe, the enrichment of metallic Cu on the surface could lead to the local dissolution of Fe, suggesting that the difference in electrochemical potential could facilitate the local galvanic interaction.

FIG. 7.

Potentiodynamic polarization measurement of the three different samples with different content of Cu.

The addition of Al to the stainless steel is expected to lead to the formation of an Al-based oxide layer with a protective nature on the steel surface, which could contribute to the improvement of corrosion resistance. However, the addition of a small amount of Al below 0.25 wt% does not have any discernible effect on the corrosion resistance, which can be found in Fig. S4. Although the 3DAP analysis identified in Fig. 8 has demonstrated that Al is enriched in a form of Al₂O₃ at the outermost layer on the passive film, it could not contribute effectively to enhancing the corrosion resistance in the chloride environment. Our previous study based on atomic distribution in 3DAP also revealed that the outermost layer is Al-based oxide and Cr is mainly located right below the layer. Further work considering a larger amount of Al contained in the steel weld will be reported in the near future.

FIG. 8.

(a) Sample preparation for the analysis of 3DAP and (b) atomic distribution of Al and O around the steel surface.