Selective Oxidation of Fe-Cr-Mn Stainless Steel in Supercritical Water Free

Material selection of the fuel cladding for the Generation IV International Forum supercritical water-cooled reactor (SCWR) concept is a key technical design consideration.^1-3  Current designs are considering a maximum supercritical water coolant exit temperature ranging from 450°C to 625°C. Among others, key material performance considerations include corrosion/oxidation and stress corrosion cracking (SCC) in supercritical water, along with irradiation damage. Ferritic-martensitic steels and austenitic stainless steels have an attractive combination of the key material performance properties for this application. Hence, both corrosion/oxidation^4-11  and SCC^12-18  of such alloys in supercritical water have received considerable attention. When compared, test results show that relative performance is strongly influenced by crystal structure: austenitic stainless steels showing lower corrosion/oxidation, but higher SCC relative to ferritic-martensitic steels.^19-23 

Corrosion/oxidation and SCC of austenitic Fe-Cr-Mn stainless steels in supercritical water have received less attention.^24-25  Such alloys have been considered for nuclear (fusion) reactor components because the replacement of Ni by Mn maintains the favorable face-centered cubic austenitic structure and significantly lowers long-term radioactivity.^26-29  Oxide scales formed on ferritic-martensitic steel and austenitic stainless steel are typically comprised of a thinner inner layer consisting of FeCr₂O₄ and/or Cr₂O₃, depending on the Cr content, and a thicker outer layer comprised of Fe₃O₄ and Fe₂O₃, with the thinner inner layer providing the bulk of corrosion/ oxidation control.^4-11  Nitronic®^† 50 (Fe-22Cr-12.5Ni-5Mn-2.2Mo) was found to form a different inner layer in supercritical water: a mixture comprised of MnCr₂O₄, Mn₂O₃, and Cr₂O₃, as identified by ex situ Raman spectroscopy and x-ray photoelectron spectroscopy (XPS).²⁴  The inner layer was argued to be an effective barrier against outward cation diffusion, and thus provided a measure of corrosion/oxidation control.

Of particular interest to fuel cladding performance in the SCWR design concept is the phase instability of austenitic Fe-Cr-Mn stainless steel that results from the selective high-temperature oxidation of Mn from the subsurface region. This instability promotes the transformation of austenite to ferrite in the Mn-depleted zone.^30-37  This instability may prove to be an effective means to improve the material’s performance of austenitic stainless steel in supercritical water by forming a more SCC-resistant ferrite surface layer on a more corrosion-resistant austenite core. Thus, the objective of the study was to determine the composition andstructure of the oxide/metal interface of Nitronic® 32 (Fe-17.2Cr-12.5Mn-1.5Ni) formed in supercritical water to validate the in situ formation of a ferrite surface layer on an austenite core by selective oxidation of Mn. In doing so, this study also provided the opportunity to further assess the corrosion/oxidation of Fe-Cr-Mn stainless steel in supercritical water.

The starting material used was a section of commercial Nitronic® 32 (N32) stainless steel rebar (∼15 mm diameter) provided in the mill-annealed condition. The chemical composition was analyzed using inductively coupled plasma-optical emission spectroscopy (ICP-OES) and combustion analysis to be (wt% basis): 16.7% Cr, 12.4% Mn, 0.7% Ni, 0.5 % Si, 0.3% N, and 0.06% C, which meets the chemical composition requirements ASTM A276 (Grade XM-28). This alloy was selected simply because it was readily available for this proof-of-concept evaluation. Figure 1(a) shows a light optical microscopy image, and Figure 1(b) shows a backscattered electron image of the starting microstructure (transverse plane). The images show a single-phase, equiaxed, and twinned grain structure. The average grain size (diameter) of 4.7±0.8 µm was determined using ASTM E112-13 (point intercept method). Small dark particles were observed to be randomly distributed within the microstructure. EDS elemental composition line scans were acquired across regions containing these dark particles. A typical example of the results is shown in Figure 1(c). In each case, the spectra revealed a composition enriched in Si, Al, Ca, and O, relative to the matrix, likely present as inclusions. Ca-bearing inclusions are not unexpected, and Fe-Cr-Mn steelmaking routinely involves the addition of Ca to modify the shape, size, and composition of inclusions for improved castability and cleanliness.³⁸  The relatively high density of inclusions (black spots) suggests poor material quality. However, as discussed later, neither the SEM (Figure 5) nor the TEM (Figure 6) analysis of the oxide/metal interface in cross section revealed any evidence to support a detrimental effect of the inclusions on the corrosion resistance exhibited by N32 in the supercritical water environment under study.

Disk coupons (∼2 mm thick) were cut from the rebar and subsequently drilled to produce a 2 mm diameter hole required to mount (suspend) the coupons on an exposure tree. A replicate set of four coupons was tested. Surface preparation involved wet mechanical abrading (800 grit surface finish) using SiC abrasive paper and water, rinsing using ethanol, and drying using warm air. Coupons were immersed in supercritical water (25 MPa at 550°C) for 500 h using a static autoclave system (CanmetMATERIALS^†). No attempt was made to control the dissolved oxygen content in the deionized water used for the exposure. Thus, a deaerated condition is assumed to have prevailed once the initial dissolved oxygen content (from air saturation) had reacted. When it comes to corrosion testing in supercritical water, static autoclaves are attractive as they are relatively simple; no pumps and auxiliary equipment to complicate operation. The caveat is that water chemistry (dissolved oxygen concentration, for example) is difficult to control since the autoclave is sealed. Despite this caveat, static autoclaves have been widely used for corrosion testing in supercritical water without any control of the dissolved oxygen content to produce meaningful results.²¹  Mass change measurements were made using an analytical balance with a precision of 0.001 mg.

Scanning electron microscopy (SEM), x-ray diffraction (XRD), and transmission electron microscopy (TEM) were used to determine the composition and structure of the oxide/metal interface. SEM (JEOL JSM-7000F^† microscope) was used to image the oxide scale both in planar and cross-sectional views (prepared using standard metallographic procedures). Energy dispersive spectroscopy (EDS) (Oxford^† Synergy system) was used to determine the elemental composition. XRD (Bruker 8D^† advanced x-ray diffractometer) data, acquired using a Co-kα radiation source, were acquired from an oxidized coupon using both low-angle (2°) diffraction and coupled bulk diffraction modes. Focused ion beam (FIB) milling (Zeiss NVision 40 CrossBeam®^† Workstation) was used to prepare TEM cross-section foils, which were lifted out of the surface using the standard H-bar procedure. The TEM (JEOL 2010F TEM/STEM microscope) examination was performed at 200 kV. Acquired selected area diffraction (SAD) patterns indexed using patterns of oxide phase created using the JEMS^† software. EDS analyses (Oxford Inca) and corresponding images were obtained using the bright-field (BF) scanning TEM (STEM) mode. Electron energy loss spectroscopy (EELS) and corresponding images were obtained using the high-angle annular dark-field (HAADF)-STEM mode to help distinguish oxide structure, in particular those with the same crystallographic structure, but different compositions. This was done using the energy shift of 510 eV and dispersion of 0.2 eV/channel to include O-K, Cr-L_2,3, Mn-L_2,3, and Fe-L_2,3 peaks in one spectrum. The relative intensity ratio of the peaks was then compared with the values reported in the literature^39-42  using the Gatan Digital Micrograph^† software.

Of the set of four coupons exposed, two were removed with the oxide scale completely intact, yielding a matte dark gray surface appearance. One of the remaining two coupons showed minor oxide scale spallation around the small mounting hole, whereas the other showed major oxide scale spallation over most of the surface. The average weight gain measured after 500 h exposure is 25±5.9 mg/dm², with the error representing one standard deviation confidence, when excluding the coupon that exhibited major oxide scale spallation. Figure 2 compares the average weight gain of N32 stainless steel with that of Type 316L (UNS S31603⁽¹⁾) (Fe-16.3Cr-10.2Ni-2.1Mo-1.6Mn-0.02C) and Type 310S (Fe-24.3Cr-19.6Ni-1.0Mn-0.06C) stainless steel in supercritical water. Type 310S coupons were exposed along with N32 coupons, whereas Type 316L coupons were exposed separately.⁸  The weight gain of N32 is lower than that exhibited by Type 316L, but higher than Type 310S. This comparison is considered meaningful as all three materials were corrosion tested in supercritical water using identical sample preparation procedures and supercritical water chemistry (using a static autoclave starting with air-saturated water).

Figure 3 shows a set of SEM images of the oxidized N32 stainless steel surface in planar view, using a coupon that exhibited significant oxide scale spallation and another that showed none. Figure 3(a) shows a low-magnification view of adjacent spalled and nonspalled regions, and Figure 3(b) shows the same interface between the two regions at higher magnification. Grinding lines are visible on the alloy surface (spalled region), whereas significant cracking is observed on the surface of the intact oxide scale. The higher magnification image of the intact oxide scale reveals a plate-like granular structure. Figure 3(c) shows a low magnification image of the intact oxide scale on the coupon that exhibited no spallation. The intact oxide scale is uniform and free from any major defects such as cracks and pores. The plate-like granular structure of the intact oxide scale on this coupon is again revealed at high magnification (Figure 3[d]).

Figure 4(a) shows the results of the XRD analyses conducted using the coupled bulk mode. Two sets of peaks were readily identified for oxides: one set for a cubic oxide structure (M₂O₃) and the other for the spinel oxide structure (M₃O₄), where M represents an oxidized metallic element (Fe-Cr-Mn). Two more peaks were readily identified for the underlying metal: austenite and ferrite. The spectrum acquired using low-angle (2°) mode is shown in Figure 4(b). As expected, the intensity peaks for the matrix phases are significantly reduced, indicating the analysis was, in large part, confined to the oxide scale. The intensity peak for ferrite is higher than that of austenite, suggesting that there is more ferrite in a mixed ferrite-austenite layer subsurface or that a ferrite layer exists on top of the austenite matrix. Regardless, the ferrite peak indicates that sufficient selective oxidation of Mn has occurred from the austenite matrix to induce transformation to ferrite. These XRD results agree well with those reported for a Fe-17Cr-10.5Mn alloy after high-temperature oxidation in air.³⁶  In that study, the pristine alloy exhibited only austenite peaks, whereas the oxidized surface exhibited peaks corresponding to ferrite, cubic M₂O₃, and spinel M₃O₄ in addition to austenite.

Figure 5(a) shows a backscattered electron image of the intact alloy/oxide interface in cross section along with one (typical) of the elemental intensity (concentration) depth profile determined from EDS line scans (Figure 5[b]). The intensity for each element plotted in the depth profile was normalized against itself (maximum intensity) to improve plot clarity. The image in Figure 5(a) shows that the oxide scale is reasonably uniform and compact in appearance. However, it is detached from the surface (internal gap), which most likely occurred during cold mounting of the sample in cross section, because the scale was intact upon removal from the autoclave. Upon closer inspection (higher magnification, Figure 5[c]), a subsurface layer of brightness contrast is observed in the alloy in addition to the formation of an external oxide scale. The element intensity (composition) depth profile (Figure 5[b]) indicates the oxide scale consists of two main layers; both comprised of Mn, Cr, and O as major scale-forming elements and Fe as a minor scale-forming element. The STEM-EDS analysis presented and discussed next (Figure 6) shows there are three layers; the innermost layer being too thin to be resolved by the SEM-EDS analysis presented in Figure 5. The concentration of Mn is higher in the outer layer, whereas the concentration of Cr is higher in the inner layer. The concentration of Fe is similar in both layers. Upon closer inspection (higher magnification, Figure 5[c]), the oxide fractured along the outer/inner oxide interface. The most logical explanation for the gap not showing up in the EDS line scan is that the gap is shallow, that is, it only exists at the surface. Therefore, EDS signals from within the sample (beyond the depth of the gap) are being collected and analyzed. The profile also indicates the formation of a subsurface zone depleted in both Cr and Mn (and concomitantly enriched in Fe), relative to the unaffected metal at the far left of the profile. The depth of the depleted zone is about 2.5 μm in depth for Cr and Mn. The zone is heterogeneous as there are localized regions of significant Cr-enrichment (relative to both the depleted zone itself and the unaffected bulk metal) within the depleted zone, closer to the alloy/oxide interface.

A bright-field TEM image of the oxide scale (Figure 6[a]) along with an associated set of STEM-EDS composition (element) depth profile (Figure 6[b]) and element maps (Figures 6[c] through [f]) for O, Cr, Mn, and Fe are all shown in Figure 6. Collectively, the STEM-EDS analysis (depth profile and maps) shows that the oxide scale comprises both Cr and Mn as major metallic components. Further inspection of the Mn and Cr maps and the depth profile indicates that the oxide scale consists of three layers: (i) thicker Mn-Cr-O outer oxide (layer 1), (ii) thinner Mn-Cr-O middle oxide (layer 2) that is enriched in Cr and depleted in O relative to the outer layer, and (iii) very thin Cr-Mn-O oxide (layer 3) at the alloy/oxide interface. The latter is more readily apparent in the element composition depth profile. Note that the composition depth profile includes only a small thickness of the outer oxide (layer 1). The region of alloy shown underneath the oxide scale is entirely within the depleted zone, as identified in Figure 5(b). Localized regions enriched in Cr (and Mn to a smaller degree) and depleted in Fe, relative to the adjacent depleted zone matrix, are apparent. Internal oxidation is not observed. The TEM acquired (bright field) image shows the oxide scale to be somewhat thinner than the SEM acquired (backscattered electron) image (0.5 µm to 1 μm in Figure 6[a] vs. about 1.5 μm in Figure 5[a]). This difference is likely related to the local variations in the oxide scale thickness, as revealed in the SEM acquired (Figure 5[a]), coupled with the much smaller length scale of examination in the TEM image (Figure 6[a]).

Figure 7(a) shows a HAADF-STEM image of the oxide/alloy interface in cross section. The STEM-EELS spectrum obtained from site 1 (outer oxide) and site 2 (middle oxide) on the HAADF-STEM image is shown in Figure 7(b). Both layers show peaks for O, Cr, and Mn. Oxide structure information can be inferred by the energy loss values associated with the O-K edges (peaks), as shown in Figure 7(c). O-K edges associated with site 1 (outer oxide) occur at different energy loss values when compared with the edges associated with site 2 (middle layer). The arrows pointing to features in the site 2 (middle layer) O-K edge spectrum indicate characteristic features in the O-K edge spectrum of M₃O₄.³⁹  A comparison between the energy loss values associated with the O-K edges and the relative intensities of the Cr-L_2,3 and Mn-L_2,3 edges associated with both sites with published values indicate that site 1 (outer oxide) is most likely a M₂O₃ type oxide and site 2 (middle oxide) is most likely a M₃O₄ type oxide.^39-42 

A bright field TEM image (Figure 8[a]) of the entire subsurface depleted zone within the alloy and associated set of STEM-EDS element maps (Figures 8[b] through [e]) are shown in Figure 8. The Mn and Cr maps clearly show depletion, relative to the presumably unaffected metal, in a region below the alloy/scale interface. The average thickness of the depleted region (based on the Mn map) is 1.8±0.1 µm. The zone is heterogeneous as it contains regions that are significantly enriched in Cr relative to the unaffected alloy, located closer to the alloy/scale interface. Mn is also present in this phase, but it is not enriched to the extent Cr is. Fe is enriched, relative to the presumably unaffected alloy, in this depleted region, except for the localized regions of Cr enrichment.

Figure 9 shows a set of bright field TEM images and associated SAD patterns acquired from the depleted zone and the unaffected metal region of the alloy. Two patterns were acquired from the depleted zone, one from a Fe-rich (Figures 9[a] and [d]) and another from a Cr-rich region (Figures 9[b] and [e]). The pattern acquired from both regions in the depleted zone was indexed to be the body-centered cubic structure (ferrite phase). The pattern acquired from the unaffected metal region (Figures 9[c] and [f]) was indexed to the face-centered cubic structure (austenite phase). Thus, the selective oxidation of Mn from the prior austenite matrix at the alloy surface was successful in producing an appreciably thick ferrite layer in the alloy. Figure 9(a) also reveals significant recrystallization occurred within a shallow subsurface region. Mechanical grinding applied to prepare a sample surface is well known to induce subsurface recrystallization during high-temperature oxidation/corrosion testing.⁴³  As a ferrite sublayer formation has been reported for electropolished, in addition to mechanically-ground, Fe-Cr-Mn samples during high-temperature oxidation,³²  it is unlikely that ferrite sublayer formation is an artifact of mechanical grinding.

The primary objective was to characterize the composition and structure of the alloy/oxide interface of N32 formed in supercritical water to validate the in situ formation of a ferrite subsurface layer on an austenite core by selective oxidation of Mn. Selective oxidation of Mn from the alloy indeed creates a subsurface depletion zone that has transformed from austenite to ferrite during immersion in supercritical water. Selective oxidation of both Cr and Mn from the alloy results in the formation of an external layered oxide scale. Based on the composition (STEM-EDS) and structure (XRD and STEM-EELS) information acquired, the oxide scale comprises a thicker Mn-rich (Mn,Cr)₂O₃ outer layer residing on a thinner (Mn,Cr)₃O₄ inner layer, which, in turn, resides on a very thin continuous (Cr,Mn)₂O₃ layer at the alloy/oxide interface.

Table 1 summarizes the structure and composition of oxide scales formed on Fe-Cr-Mn alloys in oxidizing high-temperature environments.^{24,32-33,35-37,44}  At lower temperatures studied, an intact external layered scale forms with an outer cubic oxide (M₂O₃) forming on inner spinel oxide (M₃O₄). The composition of each layer is influenced by the alloying content and temperature. More protective scales form on alloys with higher Cr content and exposures at lower temperatures. Under these conditions, the outer M₂O₃ layer is enriched in Mn at the scale/gas interface, whereas the inner M₃O₄ is enriched in Cr at the alloy/oxide interface. It seems that none of the alloys have sufficient Cr content to maintain a protective inner Cr₂O₃ layer during exposure. However, not all oxide scales were characterized using TEM and associated techniques, which may be necessary given the thinness of the layer. Oxide scales formed on alloys with lower Cr content and/or exposure at higher temperatures have Fe incorporated into both the inner M₃O₄ and outer M₂O₃ layers and include the formation of Fe-rich oxides largely associated with nodules. These oxide scales are more prone to spall, especially during cooling. Oxidation kinetics have been reported to follow a cubic rate law³⁶  and a parabolic rate law,^{32-33,35,37,44}  the latter with two stages; the initial stage associated with the formation of a thinner M₃O₄ | M₂O₃ base scale and the final stage associated with the formation of Fe-rich oxide nodules.³⁵  The large thermodynamic driving force for Mn oxidation coupled with its relatively high diffusion in austenite (face-centered cubic metal),⁴⁵  spinel oxides (M₃O₄),⁴⁶  and cubic oxides (M₂O₃)⁴⁷  is the typical explanation invoked for Mn enrichment in the scale, particularly at the scale/gas interface. Formation of mixed (Cr,Mn) oxides is also consistent with the appreciable mutual solid solubility that exists between Cr₂O₃ and Mn₂O₃, and the appreciable solubility of Mn and Cr in the Mn_yCr_3–yO₄ spinel phase.^48-49  Thus, it is clear that the (Mn,Cr)₃O₄|(Mn,Cr)₂O₃ layered scale structure formed on N32 in supercritical water reported herein agrees well with those scales formed on Fe-Cr-Mn alloys when exposed in high-temperature oxidizing environments.

Although not included in Table 1, formation of a ferrite subsurface layer also accompanies oxide scale formation during high-temperature oxidation of austenitic Fe-Cr-Mn alloys. Formation involves the transformation of austenite to ferrite within a subsurface zone depleted in both Mn and Cr.^30-37  Such a selective oxidation-induced transformation requires Mn to fall below the critical value required to stabilize austenite. Using the binary Fe-Mn phase diagram as a guide, such a critical value can be estimated from the equilibrium solubility of Mn in ferrite, which is approximately 4.7 wt% at 500°C.⁵⁰  Conversion of the Mn intensity data acquired during the STEM-EDS analysis (Figure 7) to concentration reveals a Mn content of 4.2±1.8 at% in the Fe-rich α phase, which agrees well with the equilibrium solubility of about 4.7 at% at 500°C.⁵⁰ 

As revealed by the STEM-EDS (Figures 6 and 8) and TEM-SAD results (Figure 9), a secondary phase formation process also occurs within the depleted zone: the formation of Cr-rich regions adjacent to the scale/alloy interface. The regions are presumably a Cr-rich phase that forms from thermal aging. Thermal aging of Fe-Cr-Mn stainless steel tends to promote the formation of intermetallic sigma (σ) and chi (χ) phases, in addition to Cr-rich carbide (M₂₃C₆) and nitride (CrN) phases.^26-29,51-54  A ThermoCalc^† prediction of precipitate phases formed as a function of temperature within a Fe-18Cr-10Mn-0.5N alloy (similar in composition to N32 [Fe-17Cr-12Mn-0.3N]) includes σ, CrN, and M₂₃C₆.⁵²  The body-centered cubic Cr-rich phase that formed in this work is likely not either σ, CrN, or M₂₃C₆ based on crystal structure differences: tetragonal structure for σ,⁵⁵  trigonal for CrN,⁵⁶  and face-centered cubic for M₂₃C₆.⁵⁷  However, the Cr-rich phase could be the χ phase as this phase has the body-centered cubic crystal structure.⁵⁸  The χ phase has a lattice parameter of about 8.8 Å.⁵⁰  The lattice parameter calculated from the SAD pattern of the Cr-rich phase (Figure 9[e]) is about 2.9 Å. The difference is significant and, thus, is not particularly convincing for definitive identification of χ phase formation. However, of the phases known (reported) to have formed during thermal aging, the χ phase seems most probable.

Another possibility considered, but discounted by the lack of imaging evidence, is the formation of the Cr-rich α' (body-centered cubic) phase by phase separation of the ferrite (α) phase. Such a transformation is predicted in the equilibrium Fe-Cr phase diagram⁵⁹  and is well known to occur in ferrite-containing stainless steels when exposed at temperatures ranging from 200°C to 500°C.^60-62  Formation has been observed in Fe-Cr alloys with a Cr content as low as 12 wt%.⁶³  Evidence in support of such a transition is a mottled contrast arising from Cr and Fe fluctuation as revealed by bright field TEM imaging.^64-66  No such evidence was found when imaging using TEM. Moreover, annealing at 550°C (the temperature of our SCW exposure) has been shown to completely dissolve the α-α' phase separation.^64-65 

A secondary objective was to further assess the oxidation performance of austenitic Fe-Cr-Mn stainless steel in supercritical water. Figure 2 shows that the weight gain of N32 is lower than that exhibited by Type 316L, but higher than Type 310S. The relative differences can be explained by the composition and structure, thus the protectiveness of the oxide scale forms on each. Oxide scales formed on stainless steels immersed in supercritical water typically consist of two main layers: a thicker, less compact outer layer formed by outward cationic diffusion and a thinner, more compact inner layer formed by inward anionic diffusion, with the interface coinciding with the original alloy surface.^22,67-68  Sublayers can form depending on the alloy composition and the supercritical water environment (temperature, pressure, and dissolved oxygen content).²¹  There is a consensus that the inner layer forms by a solid-state growth mechanism,^67,69  but there is an ongoing debate about whether the outer layer grows by a dissolution-precipitation mechanism⁷⁰  or a solid-state mechanism.^69,71  Regardless, the thinner, more compact inner layer is key to imparting protection, which is strongly affected by the type of oxide that is formed. An inner layer consisting of Cr₂O₃ is more protective than one consisting of Fe-Cr oxide spinel, in large part because of the significantly lower ionic diffusion rates in Cr₂O₃.^72-73 

The inner layer formed on Type 316 stainless steel typically consists of Fe-Cr M₃O₄-type spinel oxide, which is usually reported as FeCr₂O₄.^22,67-68  Treatments/processes designed to increase short-circuit diffusion paths (dislocations and/or grain boundary density) have successfully promoted formation of a thin continuous Cr₂O₃-based sublayer at the scale/alloy interface with a concomitant improvement in protection.^74-76  The weight gain for Type 316L exhibited in Figure 2 coincides with an inner layer consisting of Fe-Cr M₃O₄-type spinel oxide, presumably FeCr₂O₄.⁸  The inner layer formed on the more oxidation-resistant Type 310 stainless steels typically consists of Cr-Fe M₂O₃-type oxide, presumably (Cr,Fe)₂O₃.^77-80  Interestingly, the layered oxide scale formed on N32 consists of a Cr-Mn M₃O₄ middle layer that resides on a thin continuous Cr-rich M₂O₃-type oxide (presumably Cr₂O₃) inner layer at the alloy/oxide interface. Therefore, the formation of a more protective Cr₂O₃-based layer is likely responsible for the lower weight gain exhibited by N32 and Type 310S relative to Type 316L. The difference in the inner oxide layer composition is likely responsible for the difference in weight gain between N32 and Type 310S. The inner layer formed on N32 consists of the less protective (Mn,Cr)₃O₄, with Cr₂O₃ being restricted to a very thin layer at the alloy/oxide interface. In contrast, the inner layer formed on Type 310S likely consists entirely of the more protective (Cr,Fe)₂O₃. The higher Mn and lower Cr contents in N32, coupled with the high Mn (relative to Fe, Cr, and N) cation diffusion in Cr₂O₃-based scales, are likely responsible for the difference in the inner oxide composition.

The design thickness of the fuel cladding in the Canadian SCWR concept is approximately 600 μm, which includes about 200 μm corrosion allowance over the service lifetime (30,660 h).²  Oxide scale accumulation is also problematic as it could impede heat transfer and cause accelerated corrosion due to overheating. Consequently, an oxide accumulation no more than 25 µm (at 830°C) is permissible during a fuel cycle (10,200 h).²  One of the N32 stainless steel coupons was descaled to permit a weight loss measurement, which yielded a value of 33 mg/dm² for 500 h exposure. Linear extrapolation over the service lifetime, and subsequent conversion to thickness loss, gives a thickness loss of 26 µm, which is well within the corrosion allowance. A similar extrapolation of the oxide scale thickness formed after 500 h (1.8 µm) over one fuel cycle gives an accumulation of 37 µm, which is greater than permissible. The oxide scale thickness seems a bit low given the weight gain exhibited in Figure 2. The weight gain for Type 316L of approximately 80 mg/dm³ coincides with an oxide scale thickness (by SEM) of about 8 μm,⁸  whereas the weight gain of N32 of about 25 mg/dm³ coincides with an oxide scale thickness of 1.8 μm. Based on these values, N32 exhibits a weight gain that is 3.2 times lower and an oxide scale that is 4.4 times lower. Thus, it seems as though the oxide on N32 should be a little thicker than what is reported for the weight gain exhibited. The reason for this apparent discrepancy is unclear. Regardless, it is plausible that the Mn content in N32 is too high from a corrosion perspective and that a lower Mn content may be more suitable to achieve the desired ferrite surface layer formation, while forming a thinner and more adherent oxide scale. The alter is important to reduce the spallation tendency. As already mentioned, N32 was selected because it was readily available for study. Investigating other commercially available Fe-Cr-Mn alloys would be a logical next step to determine if alternative Cr and Mn contents can produce a thinner oxide scale, which would reduce the growth stresses responsible for spallation.⁸¹ 

The extent of the selective oxidation of Mn and associated ferrite formation within the subsurface depleted zone exhibited by N32 stainless steel is promising from a SCC control perspective. Formation of the Cr-rich phase in the subsurface deleted zone, and the associated embrittlement expected, on the other hand, may be problematic. Further microstructure instability (secondary-phase precipitation) is also expected with prolonged exposure at the high temperatures of interest. Secondary phases reported to form include Cr-rich M₂₃C₆ carbides, σ and χ phases, and CrN.^26-29,51-54  Such precipitation is of concern as it may affect the high-temperature mechanical properties, creep resistance, and SCC resistance of the fuel cladding.²  The effect of the phases on the SCC of Fe-Cr-Mn alloys in supercritical water also deserves consideration in future work.

The corrosion/oxidation performance of Nitronic® 32 (N32) stainless steel was examined in 25 MPa supercritical water at 550°C to help guide the selection of a suitable material for the fuel cladding in the Canadian SCWR design concept. Major conclusions drawn from this study can be summarized as follows:

• Selective oxidation of both Cr and Mn from the alloy results in the formation of a bi-layer oxide scale externally on the surface comprising a Mn-rich (Mn,Cr)₂O₃ outer layer residing on a Cr-rich (Cr,Mn)₃O₄ layer, which includes a Cr-rich (Cr,Mn)₂O₃ sublayer at the alloy scale interface. Selective oxidation of Mn from a subsurface depletion zone induces a primary phase transformation from austenite (starting material) to ferrite (oxidized material) and the formation of a secondary Cr-rich phase (presumably the chi [χ] phase).

• The weight gain of N32 is lower than that exhibited by Type 316L, but higher than Type 310S. Given that protection against oxidation is primarily provided by the inner oxide layer, it follows that a thin (Cr,Mn)₂O₃ inner layer at the alloy/oxide interface is more protective than FeCr₂O₄, but less protective than (Cr,Fe)₂O₃.

• The thickness of the oxide scale is problematic, considering current design constraints for the fuel cladding. However, it is expected that this concern can be controlled to some extent by optimizing the Mn and Cr content in the Fe-Cr-Mn alloy, while maintaining the desired selective oxidation-induced phase instability, required an improvement in SCC control.

Funding was provided by the Natural Sciences and Engineering Research Council of Canada (NSERC) under the Discovery Grants program (RGPIN-05182-2015). The Canadian Centre for Electron Microscopy (CCEM) is a national facility supported by NSERC, Canada Foundation for Innovation (CFI), under the MSI program, and McMaster University.