Stress Corrosion Cracking of 316L Stainless Steel in Concentrated Ammonium Chloride Solution with Very Low Dissolved Oxygen Levels

The corrosion resistance of austenitic stainless steel comes from the presence of a thin, passive oxide layer that forms on the metal surface. While superior in protection to carbon steels, austenitic stainless steels are still subject to corrosion failure, with stress corrosion cracking (SCC) being a leading cause.¹  SCC occurs when there is a combination of tensile stress, susceptible material composition, and a particular environmental chemistry.^2-4  The SCC of austenitic stainless steels in chloride-containing solutions is considered one of the classic cracking mechanisms because it can occur at very high rates (tens of mm/y) and very low stress levels (∼10% to 20% of the yield stress).^5-6  Using the empirical definitions suggested by Newman⁶  two different electrochemical conditions can cause chloride-SCC: Type B, which occurs where the SCC is initiated from sites of slow localized corrosion in an otherwise passive surface; and Type C, where there is a state of surface dealloying with no continuous oxide. However, the film-induced cleavage model^7-10  effectively unifies these types, because it involves dealloying within the localized corrosion sites of Type B.

The most widely known example is perhaps the external SCC of stainless steels due to chloride salts deposited from the atmosphere.^11-13  The selection of stainless steels for applications in these conditions is largely based on the concept of empirically established maximum acceptable temperatures for the different steel grades.¹⁴  The idea of a critical temperature for SCC first arose from a survey of industry failures of common 300 series stainless steels, and a value of 100°C was initially suggested¹⁵  then revised¹⁶  down to 60°C. Tsujikawa and co-workers^17-19  later explained that the SCC velocity in these conditions increases more rapidly with increasing temperature than does the localized corrosion velocity, giving rise to the critical temperature thresholds. Higher temperature thresholds are often reported for more highly alloyed stainless grades.²⁰  In recent years, however, there has been some downward pressure on the accepted limits for various alloys, predominantly due to incidents and laboratory experiments with very high chloride concentrations,^21-23  especially where MgCl₂ was present instead of (or as well as) NaCl, which had previously been more widely used in such work.²⁴  Shoji and Ohnaka²⁵  showed SCC of 304 (UNS S30400⁽¹⁾) and 316L stainless steels at ambient temperature in controlled humidity environments simulating atmospheric corrosion, with surface deposits of MgCl₂, CaCl₂, or ZnCl₂. Samples with surface deposits of NaCl did not crack at ambient temperature or 50°C but did crack at 70°C. The increased SCC susceptibility at lower temperatures in the Mg, Ca, and Zn salt solutions is due to the significantly higher chloride concentration in the saturated solutions of these salts that form in these conditions. On the other hand, SCC of stainless steels in acidic chloride solutions, where the surface state is no longer passive, is well known to occur at ambient temperature, even at relatively low chloride concentrations.^26-29  For all these cases with cracking at relatively low temperatures, the solutions are typically open to the air and dissolved oxygen (DO) will be present in the solution at levels up to a few ppm.

For internal process environments, where the oxygen concentration is typically relatively low (i.e., ppb levels) or zero, it is very well known that stainless steels can be used safely at much higher temperatures.³⁰  For example, a collation of experimental data^31-32  is widely used to suggest that SCC will not occur at any chloride concentration up to about 10,000 ppm, at temperatures of up to 260°C, provided that the DO concentration is <10 ppb.³¹  Nonetheless, there are some industrially important environments where it is at least suspected that chloride SCC can occur in the absence of DO—for example, in oil and gas production involving formation waters with very high MgCl₂ concentrations^33-34  or in refinery applications where NH₄Cl salts can be deposited.^1,35  For both these cases, the bulk solution is a very concentrated solution of an acid chloride salt.

The atmospheric corrosion of various metals beneath surface deposits of airborne salts has been studied extensively.^12-13  Some of these salts, including MgCl₂ and NH₄Cl, are described as hygroscopic, which means they absorb water from the air when the relative humidity (RH) is above some critical value. Some of these salts also exhibit deliquescence, which means they can absorb sufficient water to dissolve and form an aqueous solution. The deliquescence RH (DRH) is the RH above which this can occur and is the point at which the saturated salt solution is in equilibrium with the vapor pressure of water in the atmosphere above it.¹³  The time of wetness (TOW) is recognized as a critical parameter for atmospheric corrosion^13,36  and considerable effort has hence been made to establish accurate DRH values for a wide variety of salts on different surfaces and at different temperatures.³⁷  However, it is also known that the presence of hygroscopic salt deposits can lead to significant increases in corrosion rate or corrosion susceptibility, even at RH values well below the DRH^38-39  Nishikata, et al.,⁴⁰  studied deposits of MgCl₂ on 304 stainless steel in controlled RH environments at ambient temperature, finding that pitting initiated when the RH was approximately 65%. In the work of Shoji and Ohnaka²⁵  the maximum SCC susceptibility of stainless steel at temperatures from ambient to 70°C was at RH levels of 30% for MgCl₂, 20% for CaCl₂, and 10% for ZnCl₂. Vainio, et al.,⁴¹  showed that at 71°C the DRH was 78±2% for carbon steel with surface deposits of NH₄Cl and 25 vol% H₂O and 5% O₂ in the vapor phase, but that rapid corrosion occurred at 80°C with an RH of only 54%, well below the DRH value. Similarly, Toba, et al.,⁴²  tested a range of materials in conditions simulating reactor effluent streams in refinery hydroprocessing units, showing for carbon steel with NH₄Cl deposits at 80°C that corrosion began above about 30% RH, while the maximum corrosion rate was observed at about 60% RH. For Type 304 (UNS S30400) stainless steel and 25Cr duplex (UNS S39274) stainless steel, pitting was observed at RH values of 50% to 60%.

API 932-B Standard⁴³  indicates that solid NH₄Cl deposits can form directly by desublimation from a vapor phase that contains NH₃ and HCl and that they can then absorb water from the process stream and cause severe corrosion when the temperature is above the nominal water dewpoint. It suggests that some refiners define dry conditions as being those with <10% RH, noting this is probably representative of the threshold for NH₄Cl corrosion defined by Lin and Lagad⁴⁴  at 204°C and is likely conservative at lower temperatures. In practice, it is easier for the refining industry to consider this problem in terms of temperature rather than RH, taking advantage of the fact that deliquescence can be modeled as an effective increase in the local water dewpoint above a saturated solution of the salt. Sun and Fan¹  provided a detailed analysis, comparing the thermodynamic behavior of a tertiary NH₃-HCl-H₂O system with that of a binary HCl-H₂O system, showing that while 100 ppm of HCl in the vapor does not significantly raise the dew point of water, the combination of 100 ppm HCl with 10 ppm NH₃ can raise the dewpoint by nearly 20°C. Similarly, Shargay, et al.,⁴⁵  note that the presence of NH₄Cl salts can raise the effective water dewpoint by as much as 28°C to 40°C. Lordo⁴⁶  then suggests there is a “typical recommendation” to maintain the operating temperature at least 25°F (14°C) above the salt formation temperature, while Ruel, et al.,⁴⁷  note the application in their facilities of a 10°C safety margin. Once formed, these wet salt deposits can cause rapid under-deposit corrosion and SCC. If the concentrated salt solution is diluted (e.g., by more water condensing from the gas phase) then the corrosiveness decreases accordingly. So-called water-wash is a proven refinery technology in which significant amounts of clean water (usually boiler feedwater) are injected upstream of at-risk equipment to ensure that a liquid water phase is present before any de-sublimation might occur, thereby avoiding the presence of corrosive wet salt deposits. However, the potential introduction of oxygen with water wash has been identified as a major problem. API 932-B Standard⁴³  recommends a desirable wash water DO content of <15 ppb and a maximum of 50 ppb, while also suggesting that alloys such as 625 and C-276 may be necessary where wet chloride salts can be present.

In reactor effluent air cooler (REAC) systems, the process stream is likely to include H₂S as well as NH₃ and HCl. API 932-A Standard⁴⁸  explains that corrosion can hence be caused in these systems by the deposition of ammonium chlorides or bisulfides, but that the deposition temperatures are quite different. Although numbers will vary with specific refinery conditions, it is suggested the chloride salt usually deposits first, at 177°C to 232°C, while the bisulfides deposit further downstream at 27°C to 65°C. Mahajanam, et al.,⁴⁹  reports the SCC failure of Type 2205 duplex stainless steel (DSS) air cooler tubes exposed to overhead vapor containing H₂S, HCl, and NH₃, with modeling suggesting a risk of both ammonium chloride and MEA (monoethanolamine) hydrochloride salt deposition, at a temperature of 121°C to 127°C. Similarly, Nicacio, et al.,⁵⁰  state that a failure of Type 304L stainless steel tubes occurred due to the presence of ammonium chloride deposits in a reactor effluent system at approximately 134°C, again with H₂S also present. However, Ruel, et al.,⁴⁷  describe a detailed failure investigation for Type 321 stainless steel piping in a biorefinery hydrotreatment effluent stream without any H₂S. In this case, dry ammonium chloride deposition was expected to occur during normal operation at temperatures around 300°C, and it was determined that the SCC occurred due to humidification during transient start-up procedures, causing deliquescence of the salt deposits and rapid SCC at about 180°C. DO levels were not discussed in any of these refinery case studies,^47,49-50  but in the absence of water wash the potential for oxygen ingress into hydroprocessing systems is very limited.

Laboratory data concerning the localized corrosion of stainless steels in concentrated ammonium chloride solutions are quite rare. Forsen, et al.,⁵¹  showed polarization curves for various stainless-steel grades in ammonium chloride solutions with chloride contents up to 35,000 ppm and at temperatures up to 90°C, reporting pitting and crevice corrosion potentials essentially as would be expected from the general literature on pitting corrosion in sodium chloride solutions. Nicacio, et al.,⁵⁰  similarly showed polarization curves for Types 304, 316L, and 317L stainless steels in 3.5 wt% solutions of both NaCl and NH₄Cl at ambient temperature, with no significant differences between results with the different salts. Furthermore, in this work, we are concerned in particular with the impact of DO content. Challenges in completely removing DO from laboratory experiments have long been recognized,^52-54  while it has also been noted^5,55  that the presence of oxidizing impurities such as ferric ions may play a significant role when the DO has been removed. In this context, for laboratory experiments using conventionally sourced salts, one might expect a difference in purity between NH₄Cl (made by reacting HCl and NH₃, and thus nominally free of heavy metals) and MgCl₂ (commonly prepared from seawater or other salt sources). Toba, et al.,⁵⁶  studied boiling solutions of 20% NH₄Cl (b.p = 105°C) and 40 wt% NH₄Cl (b.p. = 115°C), reporting pitting corrosion for the common stainless grades, but no pitting for the more highly alloyed materials (with pitting resistance equivalent number [PRE_N] values over ∼40). The DO content was not measured, but boiling solutions are expected to have low DO.^5,52,57  Sundararajan, et al.,⁵⁸  investigated the pitting of 316L SS in NH₄Cl solutions purged with nitrogen to achieve low DO (<25 ppb), using a combination of open-circuit potential (OCP) and repassivation potential measurements, finding that the critical concentration of NH₄Cl for stable pitting at the OCP was approximately 40 wt%, 20 wt%, and 10 wt% at 65°C, 80°C, and 95°C, respectively.

Published data for SCC testing in NH₄Cl solutions are also sparse. Backensto and Yurick⁵⁹  performed tests with stressed wire specimens of 304, 316, 321, 347, 309, and 310 stainless steels at room temperature in solutions of up to 25 wt% NH₄Cl, finding no evidence of SCC in the as-received condition, while sensitized wires cracked in all tested NH₄Cl concentrations, even with nitrogen bubbling through the test solution. He, et al.,⁶⁰  investigated the SCC of 304 stainless steel using stressed U bends under atmospheric exposure conditions with various salts deposited on the sample surface and controlled RH levels in the test chamber. For salt deposits containing a mixture of ammonium nitrate and sodium chloride, with nitrate to chloride ratios of 3:1 and 6:1, they showed that the deposits had a mutual DRH (MDRH, or eutonic point) below that of either pure ammonium nitrate or pure sodium chloride and that extensive SCC occurred at 45°C in atmospheres with 44% RH, despite the expected inhibitory effect of nitrate on localized corrosion. Sun, et al.,⁶¹  performed U-bend tests of 316L stainless steel in boiling solutions at up to 43% NH₄Cl, reporting pitting and microcracks in the boiling 10% solution (assumed to be about at 100°C), and then significant pitting and SCC in the 43 wt% solution (approximately 115°C). Regniere, et al.,⁶²  performed U-bend tests in autoclaves, using saturated NH₄Cl solutions that were deaerated by bubbling with nitrogen for 3 d before the tests commenced. A range of materials were tested, including two 6Mo stainless steels (UNS S31254 and S32654), duplex and super duplex stainless steels (UNS S32205 and S32750) and several nickel-base alloys (UNS N08825, N08028, N06625, N08935, and N10276), all of which had PRE_N >33. The concentration of NH₄Cl increased with temperature, from 49 wt% at 130°C, to 59% at 180°C, and 64% at 220°C. At 130°C, only the S31254 was tested, and no SCC was observed. At 180°C, SCC was found on all of the tested alloys, which included UNS S31254, S32205, S32750, and even N08825. At 220°C, there appeared to be a significant change in behavior, with all tested samples being covered by a black surface scale. At this temperature, Alloys N08935 and S32654 nonetheless suffered SCC, while S31254, N06625, and N10276 did not.

In this paper, we aim to establish the critical temperature threshold for SCC of 316L stainless steel in concentrated NH₄Cl solutions, with very low (<100 ppb) levels of DO.

The material investigated in this study was commercial Type AISI 316L austenitic stainless steel, with the chemical composition listed in Table 1, which results in a PRE_N of 23.9. The steel was supplied in the cold drawn condition, with an average hardness of 239 HV10. The microstructure was fully austenitic (Figure 1) and showed no trace of magnetism when tested with a strong magnet.

Tests were performed in solutions of ammonium chloride at concentrations of 15 wt%, 30 wt%, and 40 wt%. The ammonium chloride was 99.5% grade, supplied by Sigma-Aldrich company. According to the supplier, the expected contaminant concentration is Ca < 10 mg/kg, Fe < 2 mg/kg, Mg < 5 mg/kg, total sulfur as < 20 mg/kg, < 2 mg/kg, heavy metals < 5 mg/kg, and sulfurated ash < 0.01%. The composition of the product was also analyzed in our laboratory using inductively coupled plasma mass spectrometry (ICP-MS), and the contaminants were identified as only 0.125 mg/kg Mn and 0.114 mg/kg Zn. The test solutions were then made up using demineralized water, made using Smart2Pure 3 equipment (manufactured by Barnstead™^†), with a conductivity <1.0 µS/cm.

The experimental setup consisted of one deaeration cell and one test cell, with a liquid volume of 0.6 L (Figure 2). The test setup was configured with the deaeration cell and the test cell arranged in series, allowing both cells to be purged simultaneously, from a single high-purity N₂ source. The solution in the deaeration cell was sparged with high purity nitrogen (99.999%) gas overnight (>16 h), at a flow rate of 0.2 L/min, through a porous glass frit which was immersed in the solution. The solution was also heated to the desired temperature before transferring to the test cell through perfluoroalkoxy (PFA) flexible tubing by a difference in pressure. After transferring the solution into the test cell, the slow strain rate testing (SSRT) was initiated. Nitrogen sparging through an open end of the tube immersed in the test solution was continued throughout the whole test to minimize the ingress of oxygen. Tests were performed at temperatures of 60°C, 80°C, 95°C, and 105°C (see Table 2). Temperature was controlled using a circulating bath connected to the test cell. The solution temperatures were recorded throughout the tests and maintained at the target test temperatures, with a measured variation of ±0.4°C.

We have used OLI Studio^†: Corrosion Analyzer software (version 11.5) for calculating solution pH, DO content, and the OCP of stainless steels in our test solutions. The basic thermodynamic model for solution chemistry and the electrochemical model for corrosion kinetics are summarized by Anderko, et al.,⁶³  while further development for application to stainless steels in chloride solution is described by Sridhar, et al.⁶⁴  Sundararajan, et al.,⁶⁵  then explain improvements to this model for stainless steels in chloride solutions with low DO.

Both direct and indirect measurements of the DO in the test solution were made. For the direct measurement, the DO content in the brine was measured using a polarographic oxygen sensor (model InPro 6850i manufactured by Mettler Toledo^†) immersed in the solution in the SSRT test cell. Measurements were made every 2 min for the total test duration (roughly 10 d). This method was used during two of the SSRT at 80°C before the sensor became damaged. Although the sensor body was made from titanium, the sensor utilizes an oxygen-permeable membrane that is reinforced with stainless steel mesh that likely suffered from corrosion in the test solution. Subsequently, the indirect measurement method was used, in which the oxygen content in the vapor phase exiting the test cell was measured using an oxygen sensor (model S80 manufactured by electro-chemical devices [ECD]). The measured oxygen content in the gas phase was then used to calculate the amount of DO in a solution using Henry’s Law (Appendix [A 1]). The indirect DO measurements were made every 30 min. This method was used during two of the SSRT at 95°C.

SSRT were performed using an LFV-100-HH^† servo-hydraulic machine from Walter+Bai, with the experimental setup shown in Figure 2. The ports for sensors and tubes in the cells were sealed using polytetrafluoroethylene (PTFE) tape. Tests were conducted twice or thrice, and the test matrix was summarized in Table 2. The specimens for SCC experiments had a cylindrical geometry with a gauge length of 25.4 mm and gauge diameter of 6.35 mm and were wet ground circumferentially up to 600 grit with silicon carbide paper shortly before mounting into the test setup. ASTM G-129⁶⁶  suggests that for most materials and conditions a strain rate between 10⁻⁵  s⁻¹ and 10⁻⁷ s⁻¹ is adequate. However, there was a concern that higher rates might be too fast for localized corrosion to initiate, which was expected to be necessary for SCC initiation.⁶⁷  Therefore, it was decided to use a value near the lower end of that range, which is consistent with strain rates employed by other researchers in similar studies.^68-69  A secondary concern was the balance between this and the need to complete a reasonable number of tests in the available time.⁶⁷  Therefore, the strain rate for all tests was set as 4 × 10⁻⁷ s⁻¹. The tests were stopped as soon as the specimens fractured. After testing, specimens were sequentially rinsed with ethanol followed by deionized (DI) water. One-half of the specimen was used for examination of the gauge length and fracture surface in a scanning electron microscope (SEM). The other half was mounted in resin, ground, and polished to reveal transverse cross sections through the sample, which were then also examined in the SEM.

Figure 3 shows the measured DO concentration as a function of time during the initial nitrogen sparging of DI water at 50°C, in a blank test. It was observed that after roughly 3 h of continuous nitrogen sparging, the DO concentration dropped to values below 20 ppb. Figure 4 then shows the measured DO in the test cell for two experiments using the direct method for roughly 10 d, during SRRT in a solution of 40% NH₄Cl at 80°C. Despite small fluctuations, the DO levels remained ≤12 ppb for most of the time in both tests. The DO levels were <20 ppb for at least 86% of the time, and <30 ppb for at least 95% of the time in test one (F1), while the values were 97% and 100% for test two (F2). Figure 5 then shows the measured DO in the test cell for two experiments using the indirect method for roughly 8 d, during SSRT in solutions of 30% and 40% NH₄Cl at 95°C. Results indicate a higher variation compared to the tests at 80°C, but the average DO levels remained ≤17 ppb in both test conditions.

Representative stress-strain curves are shown in Figure 6. All curves show three distinct regions. First, a linear elastic region up to the yield stress, followed by a strain hardening region (curved plateau) up to the ultimate tensile stress. Finally, in the last region, the stress starts to decrease, representing the necking of the sample that fails. Graphically, it is evident that there is a reduction in ductility for higher temperatures, causing the specimens to break at lower strains. Figure 7 shows the plastic elongation ratio relative to the control sample, which was the sample tested in demineralized water air at 80°C, where a ratio closer to 1 indicates minimal mechanical property alteration compared to the control sample. SSRT results reveal a significant decrease in plastic elongation ratio with increasing NH₄Cl concentrations and temperatures. For instance, the ratio dropped to 0.59 when tested in 40 wt% NH₄Cl solutions at 105°C.

Figure 8 shows the exterior surface of the specimens after testing. The specimen tested in the control environment exhibited a smooth surface, with no visible cracks on the surface (Figure 8[a]). However, this is not the case for specimens tested in NH₄Cl solution, as all of them exhibited linear features oriented perpendicular to the tensile direction (Figure 8[b]). These features are mainly concentrated in the necking region of the specimen. As the NH₄Cl concentration and temperature increased, these features became more numerous and propagated deeper into the specimen. Additionally, at temperatures of 80°C and above, corrosion pits could be found alongside the linear features (Figure 8[c]). The pits observed on the specimens tested at 80°C and 95°C exhibited similar sizes and were mainly found near the fracture surface. However, at 105°C, the depth of pits increased significantly across the whole specimen surface. The deepest pit observed was approximately 1.5 mm deep and was located at the base of the specimen, outside the gauge area (Figure 10[b]).

Figure 9 shows the fracture surfaces of the samples after testing. The control samples (tested in demineralized water) show typical cup-and-cone morphology with the fracture surface covered by dimples associated with microvoid coalescence, a common indication of a ductile fracture mechanism. The specimens tested at 60°C in the ammonium chloride solutions also presented ductile features, as shown in Figure 9(b). However, the fracture surface of the specimens tested at 80°C and above showed two distinct features; the center was covered by dimples, while a quasicleavage region was observed at the edge of the fracture surface, as shown in Figures 9(c) and (d).⁵⁹ 

Figure 10 shows the cross sections of the specimens observed by SEM and optical microscopy (OM) after testing. This reveals that the linear surface features observed in Figure 8(b) can be described as trenches, with a triangular cross section, oriented perpendicular to the stress (Figure 10[a]). They are deeper when found closer to the final fracture surface and can be explained as short, blunted cracks. The distinctive morphology identified near the edges of the fracture surfaces in Figures 9(c) and (d) can be seen as the trench walls. Furthermore, the specimens tested at higher temperatures (95°C and 105°C), exhibited deep pits (Figure 10[b]) and finely branching secondary cracks in addition to the final fracture (Figures 10[c] and [d]).

Butler, et al.,⁵²  review four common methods for deaeration of laboratory test solutions, concluding that none of these methods resulted in complete DO removal. The best technique they considered was nitrogen purging of the solution in a PTFE stoppered Pyrex beaker through a sintered glass bubbler for 20 min to 40 min, which reduced the DO level in solution to 200 ppb to 400 ppb. More recently, Hesketh, et al.,⁵³  considered the possible oxygen purge techniques for use in SCC testing. It was first noted that a widely accepted criterion for oxygen-free test solutions is that provided by EFC 17,⁵⁴  which specifies DO <10 ppb. Similarly, they note that a commonly used method for deaeration is recommended in NACE TM0177,⁷⁰  which is to sparge with nitrogen at a rate of 0.1 L/min for 1 h/L of test solution. However, they then demonstrated using tests in autoclaves that these recommendations were insufficient in many cases, depending on factors such as the size of the vessel, whether the nitrogen sparger was submerged, the salinity of the test solution, and whether the vessel was partially or completely filled. For testing in their laboratory, they practice overnight deaeration of up to 20 L test solutions, which was shown to produce DO levels <10 ppb.

In our experiments we have applied 16 h of deaeration with high-purity nitrogen at a flow rate of 0.2 L/min, for test solutions of 0.6 L, which exceeds the requirements recommended by NACE TM0177⁷⁰  and by Hesketh, et al.⁵³  As shown in Figure 3, this reduced the DO level to approximately 10 ppb after about 3 h of purging. However, as shown in Figures 4 and 5, this DO level was not maintained throughout the full duration of the tests; the mean and maximum DO values were 17.4 ppb and 34.4 ppb in 40 wt% NH₄Cl, and 16.5 and 41.8 in 30 wt% NH₄Cl solutions at 95°C.

Figure 11(b) compares the expected equilibrium DO values calculated using the OLI Studio for saturated NH₄Cl solution under aerated atmospheric environments (pO₂ = 0.21 bara) as a function of temperature, against the measured DO values in our experiments at 80°C and 95°C. Butler, et al.,⁵²  in their review considered boiling at atmospheric pressure or under vacuum as common laboratory techniques for deaeration, and it is routine in industrial plants to deaerate boiler feedwater by boiling the water while sparging with steam. As the temperature of the water approaches boiling point, the partial pressure of the water vapor approaches the total pressure, and so the partial pressure of oxygen in the vapor phase tends to zero, and hence (according to Henry’s Law) the DO concentration in the liquid phase also tends to zero. Cottis and Newman⁵  suggest that boiling solutions contain “no oxygen,” and hence that the electrochemical potential is determined by cathodic reactions other than oxygen reduction, which would include hydrogen evolution but also the reduction of oxidizing impurities such as ferric ions. In practice, however, kinetic limitations mean that boiling solutions often still contain residual levels of DO. Butler, et al.,⁵²  reported DO levels no lower than about 200 ppb after boiling DI water under vacuum conditions for 30 min, while industrial deaerators are typically designed to reduce DO below about 5 ppb or 7 ppb,⁵⁷  with additional use of chemical oxygen scavengers often required. Cottis and Newman⁵  and Turnbull⁷¹  noted that the relatively rapid ingress of oxygen from the atmosphere into thin liquid films means that, despite being at boiling point, the solution at the sample surface during drop-evaporation SCC tests is not oxygen-free, while concentrations of dissolved metal ions may also build up in these cases, be further oxidized by reaction with oxygen in the solution, and then be available as alternative cathodic (oxidizing) reactants. In our experiments, the highest temperatures are close to, but not at, the boiling point, and the average DO values we have measured are more than two orders of magnitude below the calculated equilibrium values (Figure 11[b]). Furthermore, ICP-MS analysis of the NH₄Cl used to make up our solutions showed the presence of 0.125 mg/kg Mn and 0.114 mg/kg Zn, but the dissolved ions of these metals are not expected to act as oxidants to the stainless steel samples in this work.

In these SSRT experiments, the potential of the stainless steel samples was not measured. However, the potential of 316L SS was previously measured by Sundararajan, et al.,⁵⁸  overexposures of up to 72 h in concentrated NH₄Cl solutions in similar apparatus, continuously purged with high purity nitrogen and with DO values measured continuously through several of the experiments.⁵⁸  For experiments in 5 wt% to 15 wt% NH₄Cl with DO levels < 25 ppb and at temperatures from 65°C to 95°C, the OCPs remained below the repassivation potentials measured in cyclic polarization tests, and no significant pitting was observed, so the final OCP values are considered to represent the OCP of the passive surface, which is around −300 mV_Ag/AgCl.

In comparison, the potential of the reversible hydrogen electrode (RHE) in these solutions can be estimated (see Figure 12) by assuming that the partial pressure of hydrogen at the surface is 1 bar, giving values of −461 mV_Ag/AgCl at pH 4 and 60°C and −416 mV_Ag/AgCl at pH 3 and 95°C, which covers the range of test conditions in this work (as shown in Table 2). Hence, the OCP values measured by Sundararajan, et al.,⁵⁸  are up to approximately160 mV above the maximum value that, based on this calculation, could be supported by hydrogen evolution alone. Researchers working on SCC of stainless steels in high-temperature waters have performed extensive studies concerning the variation of OCP with the DO content in solution. Both Turnbull and Psaila-Dombrowski⁵⁵  and Kim and Andresen⁷²  provide summary plots showing results from various sources for the OCP of 304 or 316 stainless steels as a function of DO in water at 288°C. The OCP for DO values <5 ppb is around −600 mV_Ag/AgCl to −900 mV_Ag/AgCl and then increases steeply with increasing DO before leveling out around −100  mV_Ag/AgCl to −200  mV_Ag/AgCl for DO values above about 10 ppb to 20 ppb. Factors such as the solution flow rate and the presence of H₂ have also been shown to affect the results,⁷³  but the steep increase at very low DO levels is always evident. We have used the OLI Studio to estimate the OCP value for 316L SS in NH₄Cl solutions with concentrations from 0.7 M to 7.2 M (3.7 wt% to 34.8 wt%), and the results are shown in Figure 11. The results are qualitatively similar to those shown by Turnbull and Psaila-Dombrowski⁵⁵  and Kim and Andresen,⁷²  although the steep increase in OCP for the 7.2 M solution is displaced to a relatively high DO level of ∼100 ppb. The pH in this work is lower than typical values found in nuclear reactor water environments considered by Kim and Andresen⁷²  which is one of the reasons that contributes to the slight displacement of the OCP to higher values measured by Sundararajan, et al.⁵⁸ 

The DO values measured in our SCC experiments are shown along with the corresponding range of passive OCP values according to the OLI Studio calculations. It reveals that these values are located at the lower end of the sigmoidal curve, and further reduction of DO would have little effect on the OCP. Sundararajan, et al.,⁶⁵  previously compared OLI Studio model predictions against measured OCP values for 316L stainless steel in seawater. In that work, the steep increase in OCP was shown to occur at DO levels in the region of about 5 ppb to 50 ppb, increasing with the flow rate in the solution. It is not clear why the OLI Studio calculations for the most concentrated NH₄Cl solution are displaced to such high DO levels, though of course the pH of this solution is much lower (Figure 14) and this may significantly affect both the anodic and cathodic kinetics. Also shown in Figure 11 are the range of DO values measured in the SCC experiments reported in this paper, and the range of passive OCP values measured in our previous work.⁵⁸  Comparing the results in this figure, we estimate that the OCP of the passive samples in these SCC tests was at intermediate values, around −350 mV_Ag/AgCl to −300 mV_Ag/AgCl

As shown in Table 3, the type of damage seen on the samples in these tests changed as the temperature and salt concentration were increased. At 60°C in up to 30 wt% NH₄Cl, no pitting or SCC was observed, though some trenching was seen (Figure 8[b]). At 80°C, no pits were seen in the 15% NH₄Cl, but some small pits were seen in the 30 wt% and 40 wt% solution (Figure 8[c]). At 95°C and 105°C in the 30 wt% and 40 wt% solutions, larger pits are seen (Figure 8[d]), together with heavily branched SCC (see Figure 10). These observations of pitting corrosion are broadly consistent with the results of Sundararajan, et al.,⁵⁸  work and summarized in Figure 13, which shows that the repassivation potential may fall below the OCP for the higher NH₄Cl concentrations and temperatures. The concentration required for the OCP to exceed the repassivation potential decreases as the temperature increases so that it is approximately 30 wt% at 65°C and 20 wt% at 90°C.

The solutions used in this work contained 15 wt%, 30 wt%, and 40 wt% of NH₄Cl, corresponding to 3.3  molal, 8.0  molal, and 12.4 molal (m) solutions. In studies of atmospheric corrosion, Nishikata, et al.,⁴⁰  showed that pitting of 304 stainless steel initiated beneath MgCl₂ deposits when the RH was approximately 65%, corresponding to a dissolved chloride concentration of approximately 7 m. Similarly, Maier and Frankel⁷⁴  studied drying droplets of MgCl₂ solution and found that pitting of 304 stainless steel initiated at chloride concentrations between 3.2 m and 12 m. It is well established^75-76  now that the pitting of stainless steels depends on the maintenance of a critical local chemistry (C_crit) within the pit, which is typically expressed as some high fraction (e.g., 70%) of the saturation concentration of dissolved metal ions (C_sat). This understanding then underlies the use of parameters such as the critical pit stability product (i.x _crit)^75,77-78  that is required to maintain C_crit, and the transition potential (E_t), which is the potential required to maintain a salt film inside the pit.⁷⁹  For relatively dilute bulk solutions, it is also well known that pitting potentials decrease linearly with increasing log [Cl⁻].^79-80  However, for 316L stainless steel Ernst and Newman⁸¹  reported a decrease in the value of C_crit/C_sat from approximately 65% at 1 M bulk chloride to as low as approximately 10% in solutions with more than 6 M chloride. Laycock, et al.,⁸²  then showed how this would result in an even steeper decrease in the critical potentials for pitting as the bulk chloride concentration increases above about 1 M. Consequently, for all of the ammonium chloride concentrations considered in the present work, and for all cases of corrosion in wet deposited salts of NH₄Cl, it seems likely that C_crit values will be <65%, such that values of i.x_crit and E_t will also be much lower than those measured in the majority of the pitting literature, where experiments have typically been performed in relatively dilute bulk chloride concentrations, ≤1M. This may be an important factor in the occurrence of pitting corrosion in the present experiments with low DO levels and correspondingly low passive OCP values.

As shown in Figures 8 and 9, the samples in this work remained in the passive region, with no indications of general corrosion, only localized pitting and SCC. This is as expected from the measured pH values of the solutions (Table 2), which range from about 4 in the 15% solution to 3 in the 40% solution and are consistent with values calculated using the OLI Studio, as shown in Figure 14. All of these values remain above typical depassivation pH values (<2.5) for stainless steel at the temperatures used in these experiments.^83-84  Figure 10 also shows secondary cracking initiating from pits. Hence, this can be classified as Type B SCC, which initiates from sites of slow localized corrosion.

As shown in Table 3, the type of damage seen in these tests changed as the temperature and salt concentration were increased. At 60°C in up to 30 wt% NH₄Cl, no pitting or SCC was observed, though some shallow trenching was seen on the gauge length close to the final fracture (Figure 8[b]). At 80°C, no pits were seen in the 15% NH₄Cl, but the edges of the fracture surface showed some small areas of quasicleavage (Figure 9[c]), and small pits were seen in the 30% solution. At 95°C and 105°C, unambiguous evidence of heavily branched SCC was seen in both the 30 wt% and 40 wt% solutions (see Figure 10), but not in the 15% solution. These results suggest an SCC temperature threshold in these tests between 80°C and 95°C. In comparison, for other cases with low or zero DO, the Shearwater failure^33-34  of 22Cr DSS (Type 2205) occurred at about 140°C in a solution containing ∼350,000 ppm MgCl₂ and ∼ 420,000 ppm CaCl₂, together with relatively minor amounts of NaCl, KCl, and NaHCO₃. Similarly, for refinery hydro-processing conditions with ammonium salt deposits, Mahajanam, et al.,⁴⁹  reported SCC failure of Type 2205 duplex stainless steel (DSS) at approxiamtely 121°C, and Ruel, et al.,⁴⁷  described SCC of Type 321 SS at about 180°C. In the laboratory, Sun, et al.,⁶¹  showed SCC of 316L stainless steel in boiling solutions of 43% NH₄Cl (likely at ∼115°C), while Regniere, et al.,⁶²  found SCC of alloys including 6Mo, DSS, and 825 in deaerated, saturated (59%) NH₄Cl solutions at 180°C. It should also be noted that the steel in our work was heavily cold-worked, with a yield strength estimated from the control test of approximately 600 MPa, compared to typical values for solution-annealed material of approximately 200 MPa. Cold work is well known to decrease time to failure for stainless steels in hot, concentrated MgCl₂ solutions^32,85  and has also been implicated in bolt failures by external chloride-SCC at ambient temperature in offshore applications.⁸⁶  Consequently, one might expect that the threshold temperature identified in the present work will represent a lower bound for the commercially available variations of 316L stainless steel. However, Tsujikawa and co-workers^17-19  showed that these temperature thresholds arise from the competition between corrosion velocity and crack velocity and hence that crevice corrosion, being slower and more stable than pitting corrosion, will result in lower threshold temperatures.

In Figure 15, we have compiled literature data from various sources^87-96  for the SCC of 304/316 stainless steel. This pooling of data excludes results of tests using sensitized steels or that had H₂S, it also ignores all variables except oxygen and chloride content, and, if no DO content information is provided, it assumes the system was aerated—unless it was boiling. Extensive data (G1) is available for temperatures higher than 200°C, usually associated with the usage of water in nuclear reactors. A second group (G2) of data is available for lower temperatures, but aerated conditions, typically associated with atmospheric SCC. The last group (G3) is for data from tests in boiling saturated magnesium chloride solutions and are assumed to have low DO due to the solubility decrease as a function of the temperature and brine concentration (the DO of boiling salt brines was considered as 100 ppb DO unless other info was provided). Finally, it also shows the results of SCC tests in concentrated NH₄Cl solutions from this work. Overall, for solutions of any chloride salt with concentrations exceeding about 10,000 ppm, it would seem unwise to assume that maintaining DO levels <10 ppb will prevent SCC at any temperature. However, even at these higher chloride concentrations, if the DO is <50 ppb and the temperature is <80°C, then the probability of SCC seems very low, provided that crevice corrosion sites are not present. We expect that the reduction of the DO level to <1 ppb, definitively below the steep rise in the OCP in Figure 11, would extend the temperature threshold to higher values, although the failure cases discussed earlier^47,49-50  suggest that even then it will likely be no higher than about 115°C.

• In this study, the susceptibility of cold-drawn 316L austenitic stainless steel to SCC in concentrated ammonium chloride solutions with very low DO was examined using SSRT. Maintaining very low DO levels for long-term experiments proved challenging. 3 h of sparging with high-purity nitrogen at a flow rate of 0.2 L/min was sufficient to reduce the DO in a 0.6 L test solution to ∼10 ppb. However, despite continuous nitrogen sparging, DO levels rose slightly over the full test duration (approximately 10 d), but remained low overall (<42 ppb). For 316L stainless steel in these conditions, pitting corrosion was not seen at 60°C, became evident at 80°C, and was severe at 95°C and above. Similarly, SCC was not seen at 60°C, was possibly initiating at 80°C without propagating significantly, and was severe at 95°C, underscoring the importance of temperature for SCC in very low DO environments. In comparison, a literature review reveals industrial case histories and laboratory measurements that show SCC of 300 series SS and 22Cr DSS in concentrated ammonium chloride solutions with assumed zero or very low DO at 115°C to 180°C.

This publication was made possible by NPRP-Standard (NPRP-S) Thirteenth (13th) Cycle grant NPRP13S-0205-200268 from the Qatar National Research Fund (a member of Qatar Foundation). The findings herein reflect the work and are solely the responsibility, of the authors.