Effect of Aggressive Anions on the Stress Corrosion Cracking of Stainless Steel in Hot-Pressurized Alkaline Water

Deriving energy products from abundant and renewable biomass continues to garner interest to help meet our increasing sustainable energy demand. Conversion of biomass into bioenergy products can be achieved through either a biochemical or thermochemical pathway.^1-4  Among them, hydrothermal liquefaction (HTL) is one thermochemical method that can efficiently and sustainably convert wet biomass and industrial biowaste into biofuels.^5-7  Hot-pressurized water (between 250°C and 374°C and 4 MPa and 22 MPa) is used as the reaction medium to dissolve and break down (depolymerize) raw solid biomass, while suppressing the repolymerization of broken-down components, to produce liquid biofuel. Both yield and quality of the biofuel produced are enhanced by the addition of a catalyst, with the addition of 0.5 M to 1.0 M K₂CO₃ (alkaline catalyst) showing promise.^8-10  However, selecting corrosion-resistant structural alloys to fabricate core components (conversion reactor in particular) for commercial scale-up remains a significant technical challenge.^11-13  Corrosiveness of the HTL conversion medium is not only affected by the hot, pressurized, alkaline nature of the aqueous reaction medium, but also by aggressive (HS⁻ and Cl⁻) anions, as intrinsic components, and organic (carboxylic) acids, as byproducts, which are released during HTL processes.^5-7,11-14 

Recent corrosion screening testing conducted in simulated HTL alkaline water shows that both ferritic and austenitic stainless steels suffer less corrosion than ferritic-martensitic steels.^12-13  Oxide films formed on these steels are similar in composition and structure to those grown in hot-pressurized water relevant to nuclear power plants,^15-17  namely a double-layered oxide consisting of a less protective Fe-rich outer oxide layer, formed by a dissolution precipitation mechanism, and a more-protective Cr-rich inner oxide layer, formed by a solid-state mechanism. The formation of a more protective inner M₃O₄ layer due to increased Cr incorporation from the dissolving metal is believed to be responsible for improved corrosion protection.^13,17  The lowest corrosion susceptibility in simulated HTL alkaline water occurs when a Cr₂O₃ layer can form at the inner layer/metal interface, as is the case for Type 310 (UNS S31000⁽¹⁾) stainless steel (Fe-25Cr-20Ni)¹³  and Alloy 33 (Fe-33Cr-32Ni):¹⁸  stainless steels with relatively high bulk Cr contents. Formation of an inner Cr₂O₃ layer can also occur on lower Cr-containing (more cost-effective) stainless steels when chromized, as demonstrated by corrosion screening testing of chromized Type 409 stainless steel (Fe-11Cr) in simulated HTL alkaline water.¹⁹  Chromizing is a surface treatment that diffuses Cr into the surface of metal substrates, to enhance, in this case, corrosion protection.

Stress-corrosion cracking (SCC) also needs to be considered during materials selection and design considering that HTL is a pressurized process and that fabricated components will likely contain welds and associated residual stresses. An analysis of field-exposed metallic components from bench-top, pilot-scale, and commercial-scale HTL systems revealed a cracking susceptibility of stainless steel.¹¹  Electron probe microanalysis (EPMA) mapping of cracks formed in Type 316L stainless steel (Fe-17Cr-12Ni-2.5Mo) revealed both O and S as major scale-forming elements with significant incorporation of Cl. Such an SCC susceptibility in HTL alkaline water should be unexpected based on reported SCC studies of stainless steel in hot, pressurized (deaearted) hydrogenated water relevant to nuclear power plants.^20-23  Hot-pressurized alkaline sulfide solutions relevant to chemical digestion of wood chips in pulp production^24-27  and hot, pressurized acidic sulfate solutions simulating extreme crevice solutions relevant to occluded secondary-side steam generator surfaces in nuclear power plants.^28-32  SCC of stainless steel in hot-pressurized (deaerated) water involves an interplay between Cr content, cold work, heat treatment, and water chemistry. In hot-pressurized alkaline sulfide solutions, SCC involves an interplay between S and Cl, albeit at lower temperatures with appreciable OH⁻ and HS⁻ concentrations, whereas SCC is S assisted, occurring without Cl⁻ ions, in hot-pressurized acidic sulfate solutions. Ferritic stainless steels are typically less prone to SCC than austenitic stainless steels when comparatively tested in hot-pressurized alkaline sulfide solutions^22-23  and in hot-pressurized water.^33-35  Thus, it is of technical interest (materials selection and design) to determine if the same holds true in hot-pressurized HTL alkaline water reaction medium given the unique process condition and solution composition relative to these reference cases.

The objective of this study was to determine the effect of aggressive anions (Cl⁻ and HS⁻) on the SCC susceptibility of stainless steel U bends immersed in simulated HTL alkaline water (10 MPa at 310°C) for 120 h. Stainless steels considered include Type 409 (ferritic), chromized Type 409 (ferritic), and Type 304L (austenitic). U bends were examined postexposure using electron microscopy. Focussed ion beam (FIB) milling was used to extract electron transparent cross-sectional foils of the oxide films and crack tips (Type 304L stainless steel) for the postexposure analysis using complementary analytical transmission electron microscopy (TEM). Special attention was given to the structure and composition of the oxide scales formed to analyze and discuss the combined role of S and Cl in activating corrosion and SCC.

U bends were made following the practice described in ASTM G30. Rectangular strips (100 mm × 20 mm × 1 mm) were prepared from the starting stainless steel plates using CNC machining. Holes 8 mm in diameter were created (by drilling) on both ends of the strips after machining to facilitate the fastening of the loading bolt (Alloy 625) and associated nuts (Alloy 625) across the ends after bending. Strips were mechanically abraded to a 600 grit surface finish using SiC abrasive paper and water as a lubricant, rinsed with ethanol, immersed in ethanol within an ultrasonic cleaner for 10 min, and then dried in a flowing air stream.

Chromized SS409 stainless steel U bends were also investigated (SS409-Cr). A subset of the rectangular strips prepared from the starting SS409 stainless steel plate was first electroplated with Cr using a commercial process (Ontario Chrome, Oakville, ON). Electroplating was conducted in a mixed chromic acid (250 g/L to 300 g/L)—sulfuric acid (2.5 g/L to 3 g/L) solution at an elevated temperature (52°C to 60°C) at an applied cathodic current density (−390 mA/cm² to 470 mA/cm²). The goal was to plate the strips with a Cr coating of about 50 µm in thickness. Chromizing was achieved by annealing the Cr electroplated strips at 975°C for 1 h in N₂ gas at ambient pressure. A horizontal tube furnace with N₂ gas flow rate 90 cm/s of was used for this purpose. Electroplated strips were rinsed with deionized water, rinsed with ethanol, and then dried in a flowing air stream before annealing. After annealing, SS409-Cr strips were lightly ground on all sides using 600 grit SiC paper to remove the oxide scale that developed. Afterward, SS409-Cr strips were rinsed with deionized water, rinsed with ethanol, immersed in ethanol within an ultrasonic cleaner for 10 min, and then dried in a flowing air stream before bending.

Once prepared, strips were bent into U bends with a bending radius of 16.67 mm using a custom-built loading jig (CanmetMATERIALS). Loading bolts and nuts were then fastened after bending to the target radius, without removing the bend strips from the loading jig. Ceramic (ZrO₂) inserts were used to isolate the Alloy 625 loading bolts and nuts from the stainless steel U bends. Clamped U bends were rinsed with ethanol and dried in a flowing air stream before immersion. Duplicate U bends were tested to ensure reproducibility in terms of a “go/no go” SCC susceptibility. Figures 1(c) and (d) show a light optical microscopy image of the microstructure in a cross section after U bending. Both microstructures were largely unaffected by the plastic deformation accrued during U bending. The nanohardness of the tensile surface, as measured in cross section at the outer tensile surface of the U bend, was 316 Hv for SS409 and 370 Hv for SS304L. An Anton Paar Nanoindentation Tester NHT³  with a Berkovich tip was used for this purpose. The different hardness suggests a different dislocation density in the plastically deformed outer fiber of the U bends when immersed in the alkaline HTL water solutions. The U bends were mounted in cross section and polished to a mirror finish, using conventional metallographic sample preparation techniques. Measurements were made using a 40 mN load force, 10 Hz acquisition rate, and 10 s dwell time.

Clamped U bends were immersed in simulated HTL alkaline water (10 MPa at 310°C) for 120 h. The test temperature of 310°C is within the temperature range considered necessary for effective HTL conversion of biomass.^5-7  The pressure is the saturated vapor pressure at this temperature. Table 2 lists the composition of the three HTL alkaline water (aqueous solutions) chemistries used to determine the effect of aggressive anions on SCC susceptibility. The concentrations selected are within the range reported for HTL conversion of woody biomass.^5-7,11-14  Solutions were prepared by mixing together reagent-grade chemicals and deionized water. All three materials were tested in solution A (Cl and S), whereas only SS304L was tested in solutions B (Cl only) and C (S only) as it was the only one of the three to suffer SCC after immersion in solution A (Cl and S). A Hastelloy C-276^† autoclave, equipped with an Alloy 625 liner, was used for immersion testing. A single set of duplicate U bends was placed inside the autoclave, in a manner that exposed the bending radius surfaces (top [tensile], bottom [compressive], and sides) of each, before adding the desired solution. The autoclave was then sealed and purged with N₂ before heating. It was expected that the starting dissolved (air-saturated) oxygen content was rapidly scavenged by corrosion processes involving the Alloy 625 liner and duplicate stainless steel U bends. It typically took about 4.5 h for the autoclave to attain 310°C. Both temperature and pressure were manually monitored to ensure both were being maintained during the immersion. After 120 h of immersion at 310°C, the autoclave was shut off and allowed to cool for 24 h before U-bend retrieval.

Table 2.

Simulated HTL Alkaline Water Solutions (10 MPa at 310°C)

View large

View Large

U bends were examined in a cross section after immersion using electron microscopy. U bends were cut in half along the midpoint (apex) of the bend (line of maximum tensile stress) perpendicular to the bending direction. Leg sections were also removed by cutting. One half was mounted to reveal the cross-sectional plane cut perpendicular to the bending direction, whereas the other half was mounted to reveal one of the edge surfaces. Only the cross-sectional plane cut perpendicular to the bending direction of SS409-Cr U-bend samples was examined as the edges contained pre-existing cracking (from the coating process). Mounted surfaces were polished to a mirror finish using conventional metallographic techniques and procedures. Final polishing was done using a fumed SiO₂ suspension (OPS by Struers) before etching with acetic glyceregia for 45 s to reveal the grain structure. FIB milling was used to prepare electron transparent TEM cross-section foils containing cracks found on mounted SS304L surfaces, using planar lift out and thinning to preserve the cracking for TEM examination. A Helios G4^† plasma FIB with Xe⁺ ion milling was used for this purpose.

A JEOL 6610LV^† SEM instrument equipped with an Oxford^† Instruments EDS system and associated Aztec^† software was used to examine the mounted surfaces. Secondary electron images, backscatter electron images, and EDS maps/line scans were all acquired at an accelerating voltage of 15 kV at a working distance of 10 mm. FIB-prepared cross-section foils were examined using complementary analytical TEM techniques. High-angle annular dark field (HAADF) imaging and elemental map/line scans, using EDS and electron energy loss spectroscopy (EELS), in scanning TEM (STEM) modes were acquired using a Thermo Scientific Talos 200X^† instrument equipped with four symmetrically positioned SDD Super-X detectors for STEM-EDS acquisition at improved solid angles and a CMOS detector for EELS acquisition. The low background-to-noise ratio of the EDS system used was ideal for characterizing small concentrations of S and Cl, if present. Analysis was conducted using an accelerating voltage of 200 kV.

Secondary electron images of the U-bend tensile surfaces (top-down view) after immersion are shown in Figure 4. Although the tensile surface of the SS409 U bend immersed in solution A (Cl and S) exhibited shallow pit-like indications (Figure 4[a]), it did not exhibit any cracking when viewed at this magnification; either within, or adjacent to, the pit-like indications. In contrast, the SS409-Cr U bend immersed in solution A (Cl and S) exhibited cracking (Figure 4[c]). However, the cracking was very similar to that observed on the unexposed surface (Figure 4[b]). Thus, it is unclear from this image if any new cracks were initiated from SCC during immersion. None of the three SS304L U bends surfaces exhibited obvious cracking (or even pit-like) indications. The only major difference in surface appearance between the three was the decoration of the solution B (Cl only) surface by small nodule (bright) particles (as observed at this magnification).

A summary of the U-bend SCC testing results in simulated HTW alkaline water is provided in Table 3. Of the test conditions considered, SCC was only observed for SS304L in solution A (Cl and S). Even with pre-existing cracks present in the chromized layer, SS409-Cr was not prone to SCC in this test. However, SS409-Cr suffered corrosion of SS409 at the coating/metal interface. Such corrosion would undermine the improved protection imparted by the chromized layer if it were to cause the coating to spall off. Longer testing is required to better understand this risk from a materials selection and corrosion control perspective. Regardless, a major finding here is that ferritic stainless steel is much less susceptible to SCC in simulated HTL alkaline water than austenitic stainless steel. Having said that, we recognize that dislocation density is likely not the same in both materials.

Table 3.

U-Bend Sample SCC Susceptibility Overview

View large

View Large

The fact that SS304L exhibited SCC in solution A (Cl and S), but not in solution B (Cl only), or C (S only), combined with SS409 not exhibiting SCC in solution A (Cl and S) (under the conditions tested), provided an opportunity to better understand the critical factors that lead to SCC in simulated HTL alkaline water. Critical factors, when combined, that are deemed to be sufficient to increase SCC susceptibility include: (i) formation and preferential corrosion of deformation bands, (ii) S segregation and/or Ni-S compound formation at the oxide/metal interface, and (iii) S and Cl incorporation into Cr-rich oxides. A discussion of why SS304L suffered SCC in solution A (Cl and S), but not in solution B (Cl only), or C (S only), is provided below, which is then followed by a discussion on why SS409 did not suffer SCC in solution A (Cl and S) (under the conditions tested).

Plastic deformation is responsible for the formation of the deformation (slip) bands observed in the SS304L U-bend microstructure. Whether deformation bands form in an alloy depends on the stacking fault energy, which in turn depends on composition.^40-41  Low stacking fault energies, such as for SS304L, favor deformation band formation, as opposed to cross-slip (such as for SS409, as discussed later), during plastic deformation. Deformation bands are more defective than the matrix within which they form as the sub-grains within the band tend to be significantly misoriented, elongated, and contain many tangled dislocations.^23,42  As these deformation bands intersect the exposed surface, creating localized regions of high stress, they have been identified as precursors for SCC of stainless steel in hot-pressurized water relevant to nuclear power plant plants.^43-46  In addition to creating a localized region of high stress at the intersecting interface, defective deformation bands serve as fast diffusion and corrosion pathways (corrosion of Fe coupled to the ingress of aggressive anions) that enhance corrosion at intersecting interfaces. This feature has been identified as a key role in propagating intergranular cracks in cold-worked Type 304 stainless steel in hot-pressurized water relevant to nuclear power plants.^23,47-48  Preferential corrosion of deformation bands was observed in the SS304L U bends immersed in solution A (Cl and S) and C (S only). Ingress involved S and Cl along with O when present in the solution A (Cl and S) and S and O in solution C (S only). Because SCC was only observed in solution A (Cl and S), deformation bands and their preferential corrosion may contribute to the SCC mechanism by inducing oxide film rupture but do not cause SCC in isolation.

Ni enrichment at the oxide/metal interface within cracks has been reported for cold-worked stainless steel exhibiting SCC when immersed in hot-pressurized water^23,47-49  and alkaline sulfide solutions.^24-27  It has been argued that SCC in hot-pressurized alkaline sulfide solution is driven by a mechanism involving dealloying at the crack tip,²⁵  which leaves behind a brittle nanoporous Ni layer, similar to what occurs in hot caustic solutions.^50-51  However, the results reported herein do not agree with this mechanism. The Cr content in the oxide that forms within deformation bands (solutions A [Cl and S] and B [Cl only]) and cracks (solution A [Cl and S]), was neither enriched nor depleted relative to the unaffected matrix. This suggests that the oxide shares the same formation mechanism as the oxide film formed on the external surface; one that only involves selective dissolution of Fe, not Cr. Oxide film formation in hot-pressurized water involves a thick double layer consisting of an outer Fe-rich M₃O₄ (with M = Fe, Cr, and Ni) type oxide and inner Cr-rich M₃O₄ type oxide. The outer layer likely forms with a dissolution precipitation mechanism involving the outward diffusion of Fe through the outer layer, whereas the inner layer forms with a solid state mechanism involving the inward diffusion of O through the inner layer.^15-17  Thus, Ni enrichment is more likely a consequence of solute rejection by the growing inner Cr-rich (M₃O₄) layer. Along with Ni, S was found to be co-enriched at the oxide/metal interfaces within the SS304L U bends. S is known to enhance degradation of Ni-Fe-Cr alloys, typically through one of three scenarios: (i) impaired stability of the Cr-rich passive film by incorporation of S, (ii) formation of a less protective mixed sulfide-oxide film, and (iii) S adsorption on bare metal surfaces resulting in less stable metal surface atoms.^52-54  The latter two effects are particularly pronounced for Ni.⁵²  The EELS analysis (Figure 15) indicates oxidized Ni at locations of co-enrichment with S, suggesting that sulfide compound formation could be the scenario in this case. The formation of less protective mixed oxide/sulfide on the bare metal surface at a crack tip has been argued to play a key role in the SCC susceptibility exhibited by Fe-Ni-Cr alloys in both hot-pressurized alkaline sulfide^24-27  and acidic sulfate³²  solutions. In this case (simulated HTL alkaline water), the sulfide film forms separately from the Cr-rich M₃O₄ oxide, separating the latter from the unaffected metal. As Ni-S co-enrichment was observed in the oxidation of deformations bands of the SS304L U bend immersed in solution B (Cl only), this factor is considered necessary to disturb passivation, but not sufficient to induce the SCC observed in solution A (Cl and S).

The role of Cl in inducing SCC of stainless steels in hot-pressurized water^32,55  and alkaline solutions^24-27  has also been argued to involve the formation of a less protective oxide on the bare metal surface exposed at the crack tip. Chloride ions are well known to adversely affect passive oxide film stability of metals and alloys by inducing localized breakdown in one of three ways: (i) localized thinning induced by preferential Cl adsorption and subsequent formation of less strongly bound metal chloride surface complexes,⁵⁶  (ii) localized voiding by interdependent cation vacancy buildup due to single charge Cl penetration into the oxide via doubly charged O vacancies,⁵⁷  and (iii) stress-induced fracture⁵⁸  by Cl penetration into oxide/metal interface via short-circuit diffusion paths and subsequent formation of metal compounds, which generate osmotic stress due to a Pilling-Bedworth ratio greater than 1.⁵⁹  The EELS line scans (Figure 15[b]) clearly showed Cl incorporation into the inner Cr-rich M₃O₄ spinel oxide layer that formed on the crack wall surface, as well as Ni-Cl compound formation in between the Cr-rich M₃O₄ spinel oxide and the Ni-S compound on the crack wall surface. Very little Cl was incorporated into the Cr-rich M₃O₄ spinel oxide that grew on the boldly exposed U-bend surface in solution B (Cl only). It follows that S is required to create a more defective Cr-rich M₃O₄ spinel oxide; one in which Cl is more easily incorporated into further impairing protectiveness.

The three critical factors discussed above can be considered within a slip dissolution mechanism^59-60  to account for the observed SCC susceptibility in simulated HTL alkaline water. The model is cyclic and involves protective film rupture by slip bands intersecting with the crack tip (providing dynamic plastic strain), subsequent dissolution of the metal along a favorable crack path (intergranular and/or low-index crystallographic plane), and eventually reformation of a protective film. As argued by Lozano-Perez, et al.,²³  deformation bands (critical factor) can contribute to crack propagation in two ways: (i) create localized high-stress regions as they intersect with the crack path and (ii) enhance dissolution at the crack tip by serving as fast diffusion/corrosion pathways. Ni-S compound formation (critical factor), along with S and Cl incorporation into the Cr-rich M₃O₄ oxide (critical factor), that forms at the crack tip makes film rupture easier and/or repassivation more difficult (S adsorption occupies sites normally occupied by OH⁻ as a precursor to passive film formation). Cl incorporation is key as the corrosion/oxidation along fast diffusion pathways (grain boundaries and deformation bands), as observed in solution C (S only) (Figure 17), did not transition into stress corrosion cracks (i.e., there is an additional effect of Cl on impairing film passivity). This implies the Cr-rich oxide that formed in solution C (S only) was sufficiently protective and/or could repassivate to inhibit a slip dissolution mechanism, despite Ni-S compound formation at the oxide/metal interface. It is worth noting that the SCC observed, likely occurred under deaerated conditions given that the 120 h immersion involved a static autoclave. From thermodynamic considerations (published Pourbaix diagram for the Ni-S-H₂O system at 300°C⁵⁴ ) Ni-S compound formation is the stable anodic dissolution product under reducing (deaerated) conditions for pH 11 (measured starting pH of solution A [Cl and S] at room temperature). While the role of hydrogen in assisting SCC in alkaline HTL water, particularly in the presence of sulfur anions,⁶¹  cannot be ruled out,⁶²  it seems unlikely that hydrogen had one. If hydrogen was playing a role, with assistance from sulfur, then one would expect to see SCC in the Type 304L stainless steel U bend immersed in solution B (S only), which was not observed. Despite this, the role of hydrogen in SCC of stainless steels in alkaline HTL water is worthy of further investigation.

This qualitative framework can also account for why SS409 did not exhibit SCC in solution A (Cl and S). Unlike austenitic stainless steels, ferritic stainless steels undergo cross-slip during cold working.^40-41  Plastic deformation is accrued with the formation of wavy, rather than parallel, slip bands with significantly less dislocation density.^63-64  Thus, short-circuit diffusion and corrosion pathways were significantly reduced. Preferential corrosion/oxidation of internal deformation bands (critical factor), or any other fast diffusion path (grain boundaries) was not observed in the SS409 U bend. Thus, both S accumulation at the oxide/scale interface, concomitant Ni-S compound formation (critical factor), and Cl incorporation into the Cr-rich M₃O₄ oxide (critical factor) were significantly impaired. In other words, film rupture at local deformation bands can quickly repassivate, limiting the time available for S or Cl to segregate or adsorb at the oxide/metal interface. However, a competition mechanism between corrosion and cracking cannot be ruled out.⁶⁵  Despite not being prone to SCC in solution A (Cl and S), SS409 was significantly more prone to corrosion than SS304L, as qualitatively judged by the formation of a much thicker oxide scale. The formation of a more protective inner M₃O₄ layer due to increased Cr incorporation is believed to be responsible for improved protection against corrosion.^13,17 

Chromizing SS409 was shown to be very effective in reducing solution access and corrosion in simulated HTL alkaline water.¹⁹  Despite containing pre-existing cracks in the coating, the SS409-Cr U bend also did not exhibit SCC in solution A (Cl and S). However, the U bend did exhibit corrosion at the coating/metal interface, presumably because of the pre-existing cracks linking together to form a continuous path to the interface, and/or the applied tensile stress opening and/or extending the pre-existing cracks. Fe-rich oxides were found to fill the pre-existing Kirkendall voids present at the interface. Such corrosion is undesirable from a coating performance perspective as these indications could link up after prolonged exposure time and cause the coating to spall off.

The SCC susceptibility of low Cr ferritic and austenitic stainless steel U bends immersed in simulated aqueous HTL conversion environments was evaluated. The following conclusions from the experimental results and analysis were drawn:

• Austenitic stainless steel SS304L U bends were prone to SCC in the simulated HTL alkaline water, but only when both aggressive impurity anions (HS⁻ and Cl⁻) were present. SCC was not observed in alkaline water with only one of these anions present. Cracking was in mixed-mode.

• Three critical factors were identified as being sufficient for SCC to occur, namely (i) preferential corrosion of deformation bands (arising from cold working), (ii) Ni-S compound formation at the oxide/metal interface, and (iii) S and Cl incorporation into the inward growing Cr-rich oxide. These critical factors were considered within an overall slip dissolution mechanism to account for the SCC observed.

• Ferritic stainless steel SS409 U bends were not prone to SCC in the solution that caused SCC of SS304L. Of the three critical factors, the lack of deformation bands, and associated preferential corrosion, was likely the missing SCC precursor.

• Chromized SS409 (SS409-Cr) U bends were also not prone to SCC in the solution that caused SCC of SS304L, despite having pre-existing cracks in the coating. However, localized corrosion was observed at the coating/metal interface, which could adversely affect the materials performance during service lifetime, and may eventually lead to SCC in the long term as the chemistry evolves.

Funding was provided by the Natural Resources Canada’s Office of Energy Research and Development (OERD) Clean Energy and Forest Innovation programs and the Natural Science and Engineering Research Council of Canada (NSERC) Discovery Grant program. Technical support for autoclave testing and electron microscopy was provided by the staff at CanmetMATERIALS. Technical support for electron microscopy was provided by staff at the Canadian Centre for Electron Microscopy (CCEM) at McMaster University.