Stress-Corrosion Cracking Behaviour of Lean-Duplex Stainless Steels in Chloride/Thiosulphate Environments

: The stress-corrosion cracking (SCC) behaviour of two lean-duplex stainless steels (DSS 2304 and LDSS 2404) was studied by slow strain-rate tests (SSRT) in 20% NaCl solution at 80 ◦ C (pH about 6) and in NACE TM-0177 solution at 25 ◦ C (pH 2.7), both in the absence and in the presence of thiosulphate ions (S 2 O 32 − ). The SCC susceptibility of the two alloys was compared to that of LDSS 2101 investigated in a previous study. LDSS 2404 was always immune to SCC, while DSS 2304 (and LDSS 2101) suffered this corrosion form at speciﬁc concentrations. The high SCC resistance of DSS 2404 in both environments was connected to its high Mo content, while the signiﬁcant SCC susceptibility of LDSS 2101 in NACE TM-0177 solution was likely due to the high Mn content of the alloy.


Introduction
Duplex stainless steels (DSS) are very interesting materials due to the favourable combination of their mechanical properties, weldability and corrosion resistance in various environments. These characteristics support their use in various fields, from the paper to the petrochemical industries and from the construction sector to nuclear energy production [1]. Compared to conventional DSS (i.e., DSS 2205), lean-duplex stainless steels (LDSS) are grades with lower nickel (i.e., LDSS 2101, LDSS 2404) and/or molybdenum (i.e., DSS 2304) content [2,3]. Their employment is considered when they can lead to economical benefits in the design and maintenance of industrial plants or civil construction. Oil and gas equipment often operates in contact with chloride-and sulphide-rich media. The use of LDSS is proposed mostly on the basis of their pitting and stress-corrosion cracking (SCC) resistance in these environments [4]. Many papers deal with the SCC resistance of DSS 2205 [5][6][7][8][9] and superduplex stainless steels [10][11][12][13] in hydrogen sulphide (H 2 S) media. By contrast, few data are reported about the SCC behaviour of LDSS in these media. Johansson et al. [4] showed that no SCC was observed at H 2 S levels up to 0.15 bar for LDSS 2101 and 1 bar for DSS 2304 in NACE TM-0177. Ruel et al. [14] found that DSS 2304 had a higher SCC resistance in comparison to LDSS 2101 in 50 g/L NaCl, 5 g/L CH 3 COONa, at pH 2.8, with 50% H 2 S, at the temperature of 20 • C. They also suggested that Mn has a negative effect on SCC resistance, while N has a positive one. The SCC susceptibility of LDSS 2404 in environments containing H 2 S has not yet been investigated. Therefore, the study of the corrosion performances of LDSS in simulated oil and gas conditions is reputedly quite important. DSS 2304 was the first commercialized lean duplex. It was developed as a σ-free and low-cost grade, due to the Mo savings (Mo mass % around 0.3%) [1]. Among LDSS currently on the market, the 2101 grade is a suitable candidate for application in the oil and gas industry, in particular for mild well conditions and topside components [4], where it can satisfactorily substitute standard austenitic grades [15,16]. On the contrary, LDSS 2404 is a recently introduced Mo-and N-containing LDSS grade, designed as an economical alternative to the 2205 grade [17]. In our laboratory, we studied the corrosion resistance of LDSS 2101 extensively [18][19][20][21]. In particular, we focused on its SCC susceptibility in chloride solutions containing thiosulphate ions (S 2 O 3 2− ) [18,20]. The approach to add S 2 O 3 2− instead of H 2 S gas in acid salt solutions in order to simulate the aggressiveness of sour well environments was suggested by Tsujikawa et al. [22], in order to study the pitting and SCC susceptibility of alloys in simulated well conditions, while minimizing the health hazards during laboratory tests and reducing the cost of equipment required to operate under safety conditions. In particular, they found that materials susceptible to SCC in H 2 S-containing solutions exhibited the same behaviour in the presence of S 2 O 3 2− [23]. Since S 2 O 3 2− is thermodynamically metastable, it may undergo disproportion and reduction reactions on the alloy surface to yield elemental S, which can be further reduced to H 2 S, depending on the solution pH and on the alloy corrosion potential [24]. As an example, H 2 S forms spontaneously on a carbon steel surface under free-corrosion conditions in acidic S 2 O 3 2− solutions [25].
On the contrary, in the same solutions a continuous surface scratching is needed to detect H 2 S gas formation on a 316-type stainless steel [23], because scratching causes a shift of the alloy-corrosion potential, into the H 2 S stability domain. Some research studies [26][27][28] helped to clarify the detrimental effect of S 2 O 3 2− dissolved in aqueous solutions on the corrosion resistance of stainless steels. They showed that adsorbed sulphur monolayers can form on Fe, Ni, and Cr, due to S 2 O 3 2− reduction on the metal surface, also under potential pH conditions in which usual Pourbaix diagrams predict no metal sulphide stability [26]. Moreover, they demonstrated that in neutral chloride solution, S 2 O 3 2− exerts a deleterious effect on the pitting resistance of a Fe-17Cr alloy on which the S 2 O 3 2− reduction occurs only in correspondence of bare alloy regions, leading to the formation of sulphide islands incorporated into the oxide layer and/or sulphur adsorbed at the metal/oxide interface [27]. Adsorbed sulphur is reported to inhibit repassivation in correspondence of passive film flaws formed by chloride ions, thereby affecting the corrosion resistance of stainless steels [26,29]. Depending on the conditions of the slow strain-rate tests (SSRT), in particular at sufficiently acidic pH, S 2 O 3 2− addition in chloride solutions may produce both H 2 S and S, which can inhibit repassivation and catalyse the ingress of hydrogen in the alloys thus inducing sulphide-induced SCC (S-SCC) in stainless steels [28]. In our preceding work [18] conditions. The adopted S 2 O 3 2− concentrations were selected on the basis of the results found for LDSS 2101 [18], so 10 −2 and 10 −1 M were used in 20% NaCl and 10 −3 and 10 −2 M were selected for NACE TM-0177 solution at pH 2.7. At the end of the tests, the samples were observed under an optical microscope (OM) to assess the morphology of the corrosive attack.

Materials and Methods
The tests were performed on LDSS 2404 and DSS 2304 stainless steels (supplied by Outokumpu S.p.A., Sheffield, United Kingdom, under annealed conditions), having the nominal chemical compositions (in mass %) reported in Table 1 [31], where also the composition of the reference LDSS 2101 is reported. The alloy Pitting Resistance Equivalent Number (PREN) values are also reported. This index, specifically developed to predict the resistance to pitting corrosion of Stainless Steels (SS) [32], is now considered a simplified way to compare SS corrosion performances against the dependence of their chemical composition and was recently applied also to DSS [33]. Usually, it is calculated on the basis of the Cr, Mo and N contents of the SS, from the formula reported in Table 1. The PREN values predict that LDSS 2404 is the most corrosion-resistant alloy because of its high Cr, Mo and N content, while DSS 2304 and LDSS 2101 are expected to present comparable resistance to localized corrosion attack. The microstructure and the corrosion attack morphologies were documented by observations of the long transversal sections of tensile samples under an optical microscope (OM), after surface polishing with diamond pastes (up to 1 µm) and etching with Beraha's reagent.
The susceptibility to SCC was evaluated by slow strain-rate tests (SSRT), with a strain rate of 1×10 −6 s −1 [34]. Tensile samples were machined from 1.5 mm-thick steel sheets, with their longitudinal direction parallel to the sheet lamination direction. They had an overall length of 230 cm and a gauge portion of 20 × 5 × 1.5 mm. Their surface was ground (always parallel to the stress direction) down to 800 grit emery papers, rinsed with deionized water, and degreased with acetone. Finally, they were screened by a two-component epoxy varnish, thereby leaving only the gauge portion exposed to the solution.
For SSRT, the tensile samples were inserted in an electrochemical cell filled by the following nitrogen-deaerated and thermostated solutions: 20% NaCl in the absence and in the presence of sodium thiosulphate (Na 2 S 2 O 3 ), at concentrations of 10 −2 and 10 −1 M, T = 80 • C (measured pH 6); or 2.
SSRT were also performed in demineralized water at 80 • C and in air at 25 • C, as reference conditions. During each test, the stress-strain curve was recorded. Stress (in MPa) is the ratio of the applied load to cross-sectional area of the gauge portion (5 mm × 1.5 mm) in tensile samples, while strain is the ratio of the sample elongation to its original length, as evaluated from the relative movement of the tensile-machine crosshead. The obtained strain values are unitless, but they were expressed in percentage, as they were multiplied by 100. In parallel to stress-strain curve recording, the open circuit potential (E OCP , versus saturated calomel electrode (SCE)) values were measured. Each test was performed in triplicate.
The SCC susceptibility was evaluated by the ratio (R) between the percentage strain to fracture (ε f %) in the test solution and that in air (for 25 • C tests) or water (for 80 • C tests). R values equal to or higher than 0.8 were considered an index of immunity to SCC [36].
At the end of the tests, the gauge length section of the samples were observed with an optical stereomicroscope and side surfaces were examined by an OM, after polishing and etching with Beraha's reagent, with the purpose of analysing crack initiation and morphology.
On unstressed samples with an exposed surface area of 72 mm 2 , E OCP measurements were performed during 42 h of immersion in the previously described aggressive solutions. After about 24 h of immersion, on still passive electrodes a scratch was performed by using a glass tip to verify their repassivation capability. At the end of the 42 h immersion, anodic polarization curves were recorded, starting from E OCP , with a scan rate of 0.2 mV/s. Figure 1 shows the microstructures of DSS 2304 and LDSS 2404 in their long transversal sections, characterized by elongated austenitic grains (lighter phase) embedded in a ferritic matrix (darker phase). These microstructures are quite similar to those of LSDD 2101 [18] and are typical of rolled-duplex stainless steels. On the three alloys, the volume fraction of ferrite was always quite close to that of austenite (51 ± 1% ferrite and balance austenite), which corresponds to the correct phase balance for these products [37].  The ε f % obtained in 20% NaCl solution for both alloys were very close to those measured in water and the samples presented a ductile type fracture like that shown in Figure 3a to about 37%. The sample presented a fracture of a more brittle type (Figure 3b), in comparison to that shown in Figure 3a, with black corrosion products deposited on the gauge length near to the fracture surface. As expected [9,25,38], iron sulphides were detected in that region by scanning electron microscopy/energy-dispersive X-ray spectroscopy (SEM-EDS) analyses. The micrograph of the long transversal section (parallel to the load direction in SSRT) presented in Figure 4a reveals the presence of selective corrosion of the ferrite phase, with a tendency to form crack-like attacks in some points along the sample gauge length.

Microstructures of the Studied Alloys
When S 2 O 3 2− concentration was increased to 10 −1 M, ε f % increased to a value close to that obtained with only chloride solution. The sample presented a moderately ductile type fracture, with black corrosion products close to the fracture region. In this case, only a selective attack of the ferrite phase can be observed by OM (Figure 4b). As already observed for LDSS 2101 [18], the increase in    Table 2 shows the average ε f % values from all SSRT, including the ε f % standard deviations.    Conversely, E OCP of LDSS 2404 was about −0.15 V SCE in this solution until a scratch was produced, which determined a significant drop of E OCP to values lower than −0.40 V SCE (in agreement with trends observed during SSRT). These values were then maintained until the end of the test. This suggests that on this alloy the passive film is more stable than that formed on DSS 2304, but its repassivation at large flaws (scratches) is hindered by the likely conversion of S 2 O 3 2− to sulphur and scarcely protective sulphides on the bare metal surface [27]. In the presence of 10 −1 M S 2 O 3 2− , both alloys exhibited E OCP values in the −0.45 ÷ −0.5 V SCE range, indicating active conditions. It is reasonable that the higher the S 2 O 3 2− concentration, the higher the adsorbed sulphur coverage, which tends to inhibit the repassivation of film flaws and stimulate the corrosion attack, particularly on the less noble ferrite phase. Galvanic coupling with austenite may favour this selective attack [30,[39][40][41][42].  These tests suggest that in 20% NaCl solution the alloys are scarcely affected by the presence of flaws in the passive film, such as those artificially produced or those caused by tensile straining during SSRT, because fast spontaneous healing of the passive film occurs. This may hinder both SCC and general corrosion attack of the alloys in this environment. S 2 O 3 2− addition affects the alloy oxide stability of DSS 2304 under dynamic ( Figure 6) and also under static (Figures 7 and 8 (Figures 6 and 7).
In 10 −2 M S 2 O 3 2− , the alloy is passive under static conditions, but undergoes a 250 mV shift of E OCP towards the negative direction if the film is damaged by tensile strain (Figure 6) or by scratching (Figures 6-8), due to the onset of ferrite corrosion. The histogram of Figure 9 compares the R index values calculated for DSS 2304 and LDSS 2404 to those obtained for LDSS 2101 [18]. An R value of 0.79 indicates a slight susceptibility to SCC of DSS 2304 in 20% NaCl solution at 80 • C in the presence of the lowest tested S 2 O 3 2− concentration (10 −2 M). The same behaviour was observed for LDSS 2101 [18].
No SCC susceptibility was detected for LDSS 2404. The different behaviour of the studied alloys can be predicted from their PREN values (Table 1) and is in strong correlation with their Mo content. In particular, LDSS 2404 with its Mo percentage about 5 times higher than that of DSS 2304 and LDSS 2101 ensures the best corrosion behaviour. Literature results [43,44] show that alloyed Mo reduces SCC and pitting-corrosion susceptibility of Ni-Cr-Mo-Fe alloy in a H 2 S-Clenvironment, due to the formation of stable molybdenum sulphide in the outer layer of the surface film which, being cation-selective, protects the inner chromium oxide layer from Cl − attack. Moreover, Mo containing SS and DSS grades also exhibit a higher pitting-corrosion resistance in Cl − /S 2 O 3 2− environments in comparison to low-Mo grades [45]. In the absence of S 2 O 3 2− , both alloys present stress-strain curves quite similar to those recorded in air at the same temperature, suggesting that they do not suffer SCC in this environment. At the lowest tested S 2 O 3 2− concentration (10 −3 M), DSS 2304 showed a stress-strain curve similar to that recorded in air ( Figure 10) and a ductile-type fracture, without secondary cracks. At the highest tested S 2 O 3 2− concentration (10 −2 M), this alloy underwent a significant ε f % reduction and several secondary cracks, likely developed from pits [30] (Figure 11), were detected in the gauge length portion of tensile specimens, characterized by a brittle fracture type. These secondary cracks propagated within the α phase and followed both γ/γ and α/γ grain boundaries, but an intense selective attack of the ferrite phase was not observed. Like DSS 2304, LDSS 2101 [18] also exhibited decreasing ε f % at increasing S 2 O 3 2− content in the NACE TM-0177 solution.  By contrast, LDSS 2404 maintained high ductility, comparable to that obtained in air at 25 • C, also in the presence of S 2 O 3 2− at both tested concentrations, as shown in Figure 10. The ductile-type fracture obtained on this alloy in air and in the presence of 10 −2 M S 2 O 3 2− is shown in Figure 12.     In spite of the absence of applied stress, the results were similar to those obtained during SSRT and shown in Figure 13. In fact, also under static conditions, the E OCP of DSS 2304 shifted rather quickly in the negative direction, from an initial value of −0.25 V SCE down to −0.5 V SCE . In this test, the electrode was not scratched. Conversely, LDSS 2404 exhibited stable noble E OCP values close to −0.1 V SCE , indicating the presence of passive conditions, capable of recovering quickly after scratching (at about 24 h of immersion, Figure 14). In Figure 15, the final anodic polarization curves are shown. They confirm that DSS 2304 is under active conditions, while LDSS 2404 exhibit a passive behaviour, with passive currents around 10 −5 A cm −2 or lower, in spite of the presence of the scratch. The histogram of Figure 16 compares the R index values calculated for DSS 2304 and LDSS 2404, to those obtained for LDSS 2101 [18]  Moreover, the latter alloy was significantly more susceptible to SCC in comparison to DSS 2304 at concentrations of 10 −3 and 10 −2 M. In NACE TM-0177 solution, LDSS 2101 results significantly more susceptible to SCC in comparison to DSS 2304 (Figure 16), and a S 2 O 3 2− concentration of 10 −4 M is high enough to induce a slight SCC susceptibility on this alloy [18]. This behaviour may be connected to the presence of a relevant content of Mn, which has a detrimental effect on pitting-corrosion resistance [14,46]. A PREN formula different from that reported in Table 1 has been proposed [47,48], in order to rate alloys regarding their pitting-corrosion resistance and including the influence of Mn content. According to this formula (PREN Mn = %Cr + 3.3%Mo + 30%N -1%Mn), LDSS 2101 is less resistant (PREN Mn = 24), than DSS 2304 ((PREN Mn = 27) and LDSS 2404 (PREN Mn = 34). Considering that in Cl − /H 2 S solutions SCC cracks are reputed to develop from pits [30], a high Mn content may contribute to affect pitting and thus SCC resistance in SSRT. In LDSS 2404, the detrimental effect of Mn is counterbalanced by the high content of Mo, so ensuring high SCC resistance [31,46]. In S 2 O 3 2 containing 20% NaCl solution at 80 • C, the detrimental influence of Mn is likely less relevant to explain differences in SCC behaviour of the studied alloys, because pitting is not detected. Instead, the selective corrosion of the ferrite phase appears competitive with SCC. Therefore, LDSS 2101 and DSS 2304 show almost the same behaviour, while LDSS 2404 maintains high performance because of its high Mo content.