Introduction

The corrosion resistance of iron aluminum alloys makes them ideal candidates for use as weld overlays in coal combustion environments. Corrosion resistance increases with aluminum concentration [1, 2], but susceptibility to hydrogen cracking [3] limits the practical aluminum concentration to around 10–15 wt%. Researchers are currently considering ternary additions to such iron–aluminum alloys to improve their corrosion resistance. Chromium additions have been reported to have both beneficial [4] and detrimental [5] effects on the corrosion resistance in this alloy system; further analysis is required to conclusively determine the relative advantages and disadvantages of chromium additions. Additionally, much of the prior work on the corrosion resistance of these alloys has been performed at higher temperatures, where formation kinetics are faster and the slower growing α-alumina phase is stable. Considerably less research has been performed on the corrosion resistance of these alloys at lower temperatures (500–700 °C).

In Part I of this paper [6], the corrosion resistance of three model iron–aluminum alloys was discussed in the context of analysis by scanning electron microscopy and thermogravimetric measurements. In the present analysis, scanning transmission electron microscopy and electron probe microanalysis is used to further analyze the corrosion products that form on these alloys in order to gain mechanistic insight into the factors controlling the reported corrosion behavior [6]. The differences in corrosion resistance are compared and contrasted in terms of alloy composition and the corrosion products that develop over long exposure times (up to 5000 h).

Experimental Procedure

The three alloys used in this study [6] were prepared at Oak Ridge National Laboratory by drop casting high purity elements into copper cooled molds. The compositions of the alloys are summarized in Part I [6]; throughout this work, the alloys are referenced by their intended nominal composition (i.e. Fe–10Al–5Cr, Fe–12.5Al and Fe–12.5Al–5Cr). Details of the corrosion experiment conditions and scanning electron microscopy analysis of the as-corroded coupons are presented in Part I of this paper [6]. Quantitative chemical analysis of metal compositions after corrosion testing was performed for select coupons using a JEOL 733 electron probe micro analyzer (EPMA) equipped with wavelength dispersive spectrometers (WDS). Kα X-ray transitions were analyzed for all elements at an accelerating voltage of 20 kV and a probe current of 30 nA. A φ (ρz) correction factor was used to quantify the compositional data to weight percentage values [7]. Samples for scanning transmission electron microscopy (STEM) were prepared from selected long term corrosion coupons using a FEI Strata 235 DB Focused Ion Beam (FIB). All STEM samples were prepared from the wedge region of the coupon, in areas that were representative of the corrosion products observed on the majority of the coupon. STEM analysis was performed on a VG HB603 analytical electron microscope (AEM) equipped with a Nion Inc. aberration corrector and operated at 300 kV. Spectrum images were obtained using an Oxford/Link windowless Si(Li) X-ray energy dispersive spectrometer (XEDS) detector. Kα lines were analyzed for each element. Measurements of the inner corrosion product thickness were determined by analysis of the STEM bright field (BF) and annular dark field (ADF) images using ImageJ quantitative image analysis software.

Results

Analysis of both short and long term corrosion kinetics for these alloys was discussed in detail elsewhere in Part I [6]; further microstructural analysis of the corrosion products that formed on long term test coupons is the focus of the present paper. The long term corrosion test results can be summarized as follows: weight change data for all three alloys are presented in Fig. 1. Fe–10Al–5Cr was protective at short times, but exhibited considerable weight gain and formation of thick iron sulfide scales after 5000 h. Iron sulfide was found on the surface of the binary Fe–12.5Al alloy at all exposure times, with near complete coverage of the entire coupon surface after 5000 h. Fe–12.5Al–5Cr did not experience significant weight gain for any exposure time, and no iron sulfide was found by SEM analysis of this alloy.

Fig. 1
figure 1

Weight change data for long term corrosion test coupons (from [6])

Full size image

In order to assess the susceptibility of these alloys to failure by depletion, wedge shaped coupons were used in the long term tests [6]. Electron probe microanalysis was performed on the Fe–12.5Al alloy after 5000 h due to the fact that this coupon experienced the highest weight gain for all tested alloys; it was expected that this coupon would exhibit the largest amount of alloy depletion. Metallographic preparation of microprobe samples was performed without the use of Al2O3 polishing media to prevent contamination errors. The electron probe diameter employed was approximately [7] 200 nm and the beam interaction volume was estimated to be 1 μm3 at 20 kV. Elemental WDS line scans were performed on polished cross sections both along the length and through the thickness of the coupon. No significant depletion of aluminum or iron within 1 μm of the inner corrosion product in the region of the sample that was less than 100 μm thick.

STEM–ADF images and XEDS maps of cross section samples are shown in Figs. 2, 3, 4, 5, 6 and 7 for all three alloys after both 500 and 5000 h of exposure time. Regions of platinum (a remnant of the Pt coating strip deposited as part of FIB sample preparation process) surrounding external corrosion products are visible in the annular dark field images for some maps. Also helpful in the interpretation are the schematic visualizations of the development of the corrosion products, derived from the elemental mapping data, shown in Figs. 8, 9 and 10.

Fig. 2
figure 2

ADF image and elemental maps (derived from the STEM–XEDS spectrum image) of Fe–10Al–5Cr alloy exposed for 500 h to mixed oxidizing/sulfidizing environment at 500 °C

Full size image
Fig. 3
figure 3

ADF image and elemental maps (derived from the STEM–XEDS spectrum image) of Fe–10Al–5Cr alloy exposed for 5000 h to mixed oxidizing/sulfidizing environment at 500 °C

Full size image
Fig. 4
figure 4

ADF image and elemental maps (derived from the STEM–XEDS spectrum image) of Fe–12.5Al alloy exposed for 500 h to mixed oxidizing/sulfidizing environment at 500 °C

Full size image
Fig. 5
figure 5

ADF image and elemental maps (derived from the STEM–XEDS spectrum image) of Fe–12.5Al alloy exposed for 5000 h to mixed oxidizing/sulfidizing environment at 500 °C

Full size image
Fig. 6
figure 6

ADF image and elemental maps (derived from the STEM–XEDS spectrum image) of Fe–12.5Al–5Cr alloy exposed for 500 h to mixed oxidizing/sulfidizing environment at 500 °C

Full size image
Fig. 7
figure 7

ADF image and elemental maps (derived from the STEM–XEDS spectrum image) of Fe–12.5Al–5Cr alloy exposed for 5000 h to mixed oxidizing/sulfidizing environment at 500 °C

Full size image
Fig. 8
figure 8

Schematic representation of scale development on Fe–10Al–5Cr alloy. Initial exposure results in formation of aluminum and chromium rich oxide layer (a). Iron sulfide develops due to diffusion of sulfur and iron through existing oxide layer (b). Continued exposure leads to nodule coalescence and formation of aluminum-sulfide layer at scale-metal interface (c)

Full size image
Fig. 9
figure 9

Schematic representation of scale development on Fe–12.5Al alloy. Initial exposure results in formation of aluminum rich oxide layer (a). Iron sulfide develops due to diffusion of sulfur and iron through the existing oxide layer (b). Coalescence of nodules results in sulfide film formation (c)

Full size image
Fig. 10
figure 10

Schematic representation of scale development on Fe–12.5Al–5Cr alloy. Initial exposure results in formation of aluminum rich oxide layer (a). Chromium–sulfur rich layer forms after penetration of sulfur through existing oxide (b). With increased exposure time, an isolated chromium–sulfur rich protrusion can form (c)

Full size image

Depths of the inner corrosion layer on all three alloys measured from the STEM–ADF maps are summarized in Table 1. The border between the inner corrosion layer and the external scale was assumed to be the line between the sulfur-rich nodules or film, and the oxygen-rich layer, this being the most distinct boundary in the corrosion products. For the Fe–12.5Al–5Cr alloy, the thickness of the entire scale was measured. This total thickness (oxide plus sulfide) correlates well with the weight gain data for the three alloys.

Table 1 Measured depth of internal corrosion layer for 500 and 5000 h exposures
Full size table

Figures 2 and 3 show the ADF image and elemental XEDS maps of the Fe–10Al–5Cr alloy after 500 and 5000 h of exposure, respectively. While most of the 500 h test coupon exhibited mixed flake corrosion products, the area analyzed in Fig. 2 is from a region of coupon that contained some iron sulfide nodules, such as shown in Fig. 6a of Part I of this study [6]. Interestingly, a faint outlining of these nodules with aluminum and oxygen, and to a lesser extent, chromium, was observed on multiple samples. The inner corrosion layer immediately beneath the original metal surface was found to be approximately 0.2 μm thick (Table 1) and is rich in aluminum and oxygen, with some chromium and sulfur. Additionally, there are protrusions into the metal that contain both aluminum and sulfur. At 5000 h, the external iron sulfide nodules have coalesced to form a continuous film (see Fig. 7b in Part I [6]), although encapsulation of small regions with aluminum and oxygen is still evident. The inner scale layer is considerably thicker (0.8 μm) and the aluminum–sulfur protrusions into the metal are further elongated.

After 500 h of exposure, the binary Fe–12.5Al alloy, Fig. 4, developed external iron and sulfur rich blocks or nodules that exhibit the same aluminum–oxygen layer encapsulation as described above. The area mapped is representative of the corrosion morphology presented in Fig. 9c in Part I [6], with separate sulfide blocks on top of a mixed oxide/sulfide layer, rich in aluminum and oxygen, with some traces of iron and sulfur. At 5000 h, the most significant change in the scale is the coalescence of the individual iron sulfide nodules into a continuous film, Fig. 5. The external iron and sulfur rich layer in this Figure is the sulfide film that was observed using scanning electron microscopy in Figs. 13b and 14 of Part I [6]. Additionally, an aluminum–oxygen rich layer was observed at the metal—inner scale interface; this layer was characteristically observed on multiple repeats samples from this alloy. Sulfur was present at the interface of this layer with the un-corroded metal. Inner layer thickness increased from 0.5 μm at 500 h to almost 2 μm at 5000 h (Table 1).

ADF images and XEDS elemental maps of the Fe–12.5Al–5Cr alloy after 500 and 5000 h of exposure are shown in Figs. 6 and 7, respectively (note the higher respective magnifications of these maps in comparison to those presented thus far). At 500 h, a characteristic layered structure, approximately 25–50 nm thick, is observed for the alloy, consisting of a chromium–sulfur layer sandwiched between two aluminum–oxygen rich layers. Somewhat isolated large, blocky nodules rich in chromium and sulfur were also observed on this alloy at 5000 h, Fig. 7. The scale observed on this alloy at both exposure times using scanning electron microscopy (see Fig. 5 in Part I [6]) exhibited flake-like morphology; it is likely that the large, sulfur rich protrusion shown in Fig. 7 is one of these flakes. As with the other alloys, a thin (i.e. less than 100 nm) layer enriched in aluminum and oxygen completely surrounds the chromium–sulfur phase.

By combining the schematic diagrams of the scale development from Figs. 10, 11 and 12, the difference in the corrosion resistance between the different alloys becomes more apparent. Figure 11 is an alternate representation in which the diagrams for each alloy are plotted according to the degree of attack or scale morphology, regardless of exposure time. Stage I represents the initial reaction between the gas and the alloy. At Stage II, the Fe–12.5Al–5Cr alloy has developed the layered scale observed after 500 h (Fig. 6); there is no comparable scale morphology for the two other alloys investigated. Development of blocky sulfides (Stage III) is representative of the Fe–12.5Al–5Cr alloy after 5000 h (Fig. 7), and the Fe–10Al–5Cr and Fe–12.5Al alloys after 500 h (Figs. 2 and 4 respectively). The thickness of the aluminum and oxygen rich layers in this particular stage in Fig. 11 is drawn approximately to scale, based on the data presented in Table 1, to indicate the degree of attack on the different alloys. By Stage IV, the Fe–10Al–5Cr and Fe–12.5Al alloys have developed a continuous sulfide film; no such morphology exists for the Fe–12.5Al–5Cr alloy after 5000 h exposure. From this figure it is evident that the overall scale thickness on the Fe–12.5Al–5Cr alloy at 5000 h is still dramatically less than that of the other two alloys at 500 h.

Fig. 11
figure 11

Composite rearrangement of schematic corrosion diagrams from Figs. 10, 11 and 12, with diagrams aligned by similar corrosion morphology or degree of attack for the alloys (irrespective of exposure time). Stage III is drawn approximately to scale; other stages magnified to show detail

Full size image
Fig. 12
figure 12

ADF image and elemental maps (derived from the STEM–XEDS spectrum image) of Fe–10Al–5Cr alloy after 5000 h of exposure. Note aluminum and sulfur rich plate-like features at corrosion-product metal interface

Full size image

To determine the stoichiometry of the sulfur and oxygen rich scale regions on these alloys, semi-quantification of the STEM–XEDS maps was attempted using Oxford Instrument’s INCA Microanalysis Suite 14 software utilizing standard Cliff-Lorimer k-factors. The bulky iron and sulfur rich areas (nodules and film) on all samples were found to have an iron to sulfur concentration ratio (in at.%) of between 0.97 and 1.0, suggesting that the sulfide phase was FeS, not FeS2. Conversely, the block-like chromium and sulfur rich nodule on the Fe–12.5Al–5Cr after 5000 h of exposure (Fig. 7) had a chromium to sulfur concentration ratio of 0.76, which is close to the Cr–S ratio for Cr2S3 (0.67). Analysis of the other sulfur and oxygen rich regions on the samples could not conclusively identify specific compounds.

Discussion

The EPMA analysis showed no change in concentration of aluminum or iron after 5000 h of exposure for the Fe–12.5Al alloy. Even in a region of the wedge where the sample thickness was less than 100 μm, both aluminum and iron levels were within the experimental error of the nominal concentration. Selection of the binary alloy for the detailed EPMA analysis was based on the considerably greater weight gain for this composition as compared to the other two alloys (Fig. 1), since it is reasonable to assume that depletion effects would be greatest for this particular alloy. However, as shown in Fig. 5, the STEM analysis revealed that an aluminum- and oxygen-rich layer developed at the alloy-scale interface sometime between 1000 and 5000 h. It is highly likely that this layer, which the present work cannot conclusively identify as Al2O3, acted as a barrier to slow the diffusion of iron or sulfur and limit the extent of attack, as evidenced by the lack of any significant weight increase between the 1000 and 5000 h tests (Fig. 1). If any depletion of aluminum or chromium had occurred prior to the 1000 h test, the subsequent exposure for 4,000 additional hours without any significant scale formation may have allowed enough time for diffusion in the alloy to eliminate any depletion profile that may have developed at earlier times.

The STEM–XEDS maps show significant differences in scale composition, morphology and thickness for the alloys in this study. Comparing the two alloys with 12.5 wt% aluminum (Figs. 9, 10), the major difference in corrosion products is the formation of chromium sulfide on the Fe–12.5Al–5Cr alloy (at long times) and the formation of iron sulfide on the Fe–12.5Al alloy. Of the three elements in these alloys, aluminum forms the most stable oxide, according to the free energies of formation [8, 9]; however, if equilibrium has not been reached, oxides and sulfides of aluminum, chromium and iron may all form initially. The development of the different sulfides on the two 12.5 wt% aluminum has an important consequence on the corrosion resistance, since iron sulfide typically has a more defective structure than chromium sulfide. In a review of sulfidation of metals and alloys, Mrowec [10] reported that the maximum deviation from stoichiometry for FeS at 800 °C is 0.2, while for Cr2S3 at 700 °C, it is 0.08. While there is a difference in temperature in the present work, and the stoichiometry of the sulfides formed here could not be determined, Mrowec’s data indicates that iron sulfide will have a higher lattice defect concentration than chromium sulfide. Diffusion of oxygen or sulfur towards the metal, or diffusion of iron out, therefore, will be easier through iron sulfide than chromium sulfide, which will inevitably result in an increased growth rate of the former compound. This is evidenced by the measurements of inner scale thickness made from the STEM–ADF images.

Not only is the overall thickness much lower on the Fe–12.5Al–5Cr alloy, but there is also little change in scale thickness for this particular alloy between 500 and 5000 h of exposure. While the convoluted interface between the metal and the inner scale resulted in a large standard deviation for the average thickness measurement on the binary alloy, it is obvious from the STEM–ADF images and the maximum thickness data that the inner scale on Fe–12.5Al increases considerably with exposure time. Additionally, this measurement on the binary alloy is for the inner corrosion product only; previous analysis [6] of this sample indicated that the outer iron sulfide layer thickness was on the order of 20 μm, indicating that the scale after 500 h is not an effective diffusion barrier. In contrast, the scale that forms on the Fe–12.5Al–5Cr alloy remains thin (on the order of 100 nm) up to exposure times of 5000 h, indicating that the combination of the initial oxide and chromium sulfide acted as a diffusion barrier to significantly slow the diffusion of metal ions or sulfur through the scale.

The formation of iron sulfide on the Fe–12.5Al alloy is not necessarily indicative of classic breakaway corrosion. The similar weight gain observed for this alloy between the 1000 and 5000 h tests (Fig. 1) suggests that the scale formed after 1000 h minimizes further attack by slowing the diffusion of corroding species. Breakaway behavior in sulfidizing environments [11] is usually associated with significant ingress of sulfur (or other corroding species) through a protective oxide layer and the formation internal sulfides; corrosion kinetics typically increase rapidly with the onset of breakaway corrosion. With the Fe–12.5Al alloy, there is a period of rapid growth (up to some time between 500 and 1000 h) followed by negligible weight gain (1000–5000 h). Given the defective structure and rapid growth kinetics of iron sulfide, it is unlikely that FeS is providing the observed protection against further attack after 1000 h. Rather, the aluminum and oxygen rich layer observed at the alloy/scale interface at 5000 h (Fig. 5) is the most probable origin of the observed protection after 1000 h. While exact composition of this layer was not determined, it is assumed to be aluminum oxide. Other researchers have observed similar “healing layers” in different alloy systems [12–14]. At the interface between the corrosion product and the metal, the partial pressures of oxygen and sulfur are controlled by the dissociation pressure of corrosion product. The formation of this oxygen rich layer is most likely due to a decrease in the sulfur pressure at this interface, which would make the oxide phase more stable. Development of this layer is a significant result, as it shows that the formation of substantial iron sulfide scale on a binary Fe–Al alloy does not necessarily imply a complete loss of protection or breakaway corrosion, since the internal aluminum and oxygen rich layer may act as an effective diffusion barrier after a certain incubation period.

Comparison of the Fe–10Al–5Cr and Fe–12.5Al–5Cr alloys illustrates the effect of changing the aluminum concentration. From the STEM–XEDS elemental maps (Figs. 2 and 3), it is evident that the lower aluminum alloy did not develop chromium sulfide, but iron sulfide, similar to that observed on the binary alloy. It is hypothesized that upon initial exposure to the corrosive gas, the aluminum concentration in the Fe–10Al–5Cr alloy is not high enough to selectively oxidize only aluminum. In contrast to the Fe–12.5Al–5Cr alloy, oxides of both aluminum and chromium form on the low aluminum alloy (Fig. 8a); this is evident in the 500 h EDS map (Fig. 2), where chromium alongside aluminum and oxygen was found in both the inner corrosion layer and around the iron sulfide nodules. The formation of these scales decreases the concentration of aluminum and chromium close to the surface and also consumes some amount of oxygen from the local atmosphere. Iron sulfide begins to form and continued diffusion of sulfur and oxygen into the metal causes the inner corrosion scale to thicken, as iron diffuses out to form more iron sulfide. Coalescence of the sulfide nodules eventually creates regions of sulfide film, and continued ingress of sulfur results in the formation of the aluminum–sulfur rich protrusions observed at the metal-scale interface (Fig. 8c).

As with the binary alloy, the growth of iron sulfide may not necessarily result in a significant loss of corrosion resistance for the Fe–10Al–5Cr alloy. An aluminum oxide healing layer at the alloy/scale interface may also eventually develop here, although at much longer exposure times. The thickness of the inner scale (Table 1) on the Fe–10Al–5Cr alloy is only approximately half as that of the inner scale on the Fe–12.5Al alloy. The lowering of the partial pressure of sulfur to the critical level at which oxides become thermodynamically favored may not occur until this inner scale develops further into the alloy. Some evidence of this layer is visible in the 5000 h sample (Fig. 3). However, this layer was observed on only some of the STEM samples obtained from this alloy at this particular exposure time, whereas it was present on nearly all of the STEM samples of the Fe–12.5Al alloy at 5000 h.

These results suggest that depletion of chromium beneath the initial oxide layer leads to the formation of iron sulfide corrosion products on the Fe–10Al–5Cr alloy. A similar mechanism was considered responsible for the formation of iron sulfides on iron aluminum alloys in a sulfidizing environment at 700–900 °C by Kai and Huang [15]. Lee et al. [16] have shown that depletion regions developed beneath an external Al2O3 scale during oxidation of Fe3Al. Shebany and Douglass [17] pre-oxidized iron, nickel and cobalt alloys prior to sulfidation and observed that the protective scale on the cobalt alloys failed after 4 h, leading to rapid sulfidation. It was hypothesized that depletion of chromium in the cobalt alloys during pre-oxidation allowed cobalt sulfides to form.

Sub-surface depletion of the less noble metal during oxidation of a binary alloy was first discussed by Wagner [18] and others have since modified or updated that original model [19–21]. Using the data of Akuezue and Whittle [22] and Akuezue and Stringer [23], interdiffusion coefficients of iron, aluminum and chromium at the temperature of the corrosion tests performed here (500 °C), are estimated to be on the order of 10−11 to 10−12 cm2 s−1. additionally, it is likely that at this temperature, transitional aluminas (θ, γ, δ) will form in addition to or rather than the α-phase, since these metastable phases grow more rapidly than the α-phase [24, 25]. Thus, in the present work, it is reasonable to assume that at 500 °C, diffusion of aluminum and chromium may be too slow to prevent a sub-surface depletion layer from forming during the initial stages of exposure. This is in agreement with the work of Lesage et al. [26], who showed that the oxidation of MA956 resulted in aluminum depletion profiles below 1080 °C. When oxidation occurred above 1080 °C, however, a decrease in the overall aluminum concentration from the nominal alloy level was noted, but a depletion profile did not develop, since diffusion rates were fast enough at this temperature to effectively eliminate any concentration gradient that may have been created by oxidation.

The corrosion behavior of the two ternary alloys in this work indicates that the critical concentration of aluminum to form a protective aluminum oxide is somewhere between 9.2 and 11.6 wt% (the actual composition of the Fe–10Al–5Cr and Fe–12.5Al–5Cr alloys) under the current set of test conditions. Other researchers have noted critical aluminum concentrations for protective behavior in iron aluminum alloys close to these values, although the level varies with both corrosion environment and temperature. Natesan and Johnson [27] reported Fe–Cr–Al and Fe–Al alloys with greater than 10 wt% aluminum exhibited protective behavior in sulfur-bearing environments at 650–1000 °C. Stott et al. [28] showed that alloys containing 8 wt% aluminum and 5 wt% chromium did not develop Al2O3 scales at 500 °C in a high sulfur, low oxygen environment, while an alloy with 12 wt% aluminum was protective and did not develop significant amounts of sulfide at 700 °C. In a cyclic sulfidation test at 500 °C, Klower [29] showed that metal loss for Fe-10 wt% Al-2 wt% Cr after 2,000 h was 120 μm, but only 1 μm for an Fe-17 wt% Al-2 wt% Cr alloy. In an atmosphere similar to the one employed in the current study, Kai and Huang [15] showed that an iron–aluminum alloy with less than 15 wt% aluminum formed both sulfides and Al2O3, whereas an alloy with 25 wt% aluminum formed only Al2O3 at 700 °C, though some minor iron sulfide component was found on this alloy at higher temperatures.

In addition to the different sulfides that formed on Fe–12.5Al–5Cr compared to the other alloys, another distinguishing feature is the absence of the aluminum and sulfur rich protrusions observed at the inner scale-metal interface on the Fe–10Al–5Cr alloy (Fig. 12). These platelets were observed in parallel groups and resemble the Al2S3 platelets observed by others [30, 31] at the scale-metal interface during the sulfidation of iron aluminum alloys. Patnaik and Smeltzer [30] showed that the inward diffusion of sulfur resulting from the dissociation of external FeS caused Al2S3 plates to grow internally on iron aluminum alloys. Przybylski et al. [32] observed the plates on an Fe-25 wt% Cr-10 wt% Al alloy and showed through serial sectioning that they form at select orientations. The chromium sulfide and aluminum oxide layers on the Fe–12.5Al–5Cr alloy in the present work may slowed inward diffusion of sulfur sufficiently to prevent Al–S platelets from forming. Conversely, the iron sulfide adjacent to the un-corroded metal on the Fe–10Al–5Cr alloy may have acted as a source of sulfur for reaction with the alloy to form the Al–S platelets observed on this alloy.

In contrast to the ternary alloys, the Al–S protrusions were not observed by STEM–XEDS analysis on the binary Fe–12.5Al alloy. Lower magnification SEM images of the metal interface on this alloy, however, show that this particular alloy does form high aspect ratio platelets or needles at this interface (Fig. 13a, b). Banovic et al. [33] observed similar needle-like corrosion products at the metal interface of an iron alloy with 5 wt% aluminum exposed to a sulfidizing environment, as shown in Fig. 13c. Using electron microprobe analysis, those authors determined that the internal needles were the spinel FeAl2S4 τ-phase. Considering the similarity of both the alloy composition, it is likely that the plates observed in the present work are also the FeAl2S4 phase. The needles observed on the alloy here appear to be aligned along characteristic directions, indicating a possible crystallographic orientation relationship between this phase and the ferrite matrix. Banovic et al. [34] reported that the FeAl2S4 phase grew with either 4 or 6 growth directions in ferrite grains, indicating perhaps alignment of the spinel with an easy growth direction in the ferrite.

Fig. 13
figure 13

Secondary electron image of the inner corrosion and metal interface on the Fe–12.5Al alloy after 5000 h of exposure (a, b), compared to (c) the similar features observed by Banovic et al. [33] on a Fe-5 wt% Al alloy

Full size image

In that same study, the authors observed a 5 wt% increase in the oxygen concentration (as determined by EPMA) of this spinel corrosion product, with a concurrent decrease in the sulfur concentration of approximately 5 wt%, over a 6 day storage period within a dessicator. A similar degradation of the Al2S3 phase upon exposure to water vapor has also been reported by Mrowec and Wedrychowska [35] and it is possible that a similar decomposition has occurred in the present study as well. Decomposition of the sulfur rich corrosion products could in principle have been accelerated in the STEM work presented here, as the ultra-thin electron-transparent samples would have more surface area exposed to oxygen than the bulk samples in the Banovic et al. [34] or Mrowec and Wedrychowska [35] studies. This could explain the discrepancy of oxygen enrichment in this phase between the work presented here and the prior findings of Banovic; further investigation is required to fully understand the formation mechanism of these corrosion products.

The aluminum and oxygen rich layer observed outside the external sulfides in the STEM–XEDS elemental maps (see in particular Figs. 3, 4 and 7) is an interesting aspect of the corrosion of these alloys. It is possible that the iron sulfide a significant amount of aluminum that reacted with the oxygen in the environment to form this encapsulation layer. Such doping of iron sulfide with aluminum has been observed by others [30, 36, 37]. Strafford and Manifold [36] reported the presence of approximately 1 wt% aluminum dissolved in the iron sulfide that formed on an Fe-5 wt% Al alloy at 600 °C. Similarly, Kai and Douglass [37] found dissolved aluminum in external iron sulfide phases on Fe–Mo–Al ternary alloys in H2–H2O–H2S environments. In both of these studies, oxygen was not measured in the sulfide phase and aluminum was described (through analysis with EPMA traces through the FeS nodules) as being dissolved throughout the sulfide phase, rather than being concentrated in the form of an oxide layer at the sulfide surface as observed in the present study. No aluminum in solution with the iron sulfide was detected in the STEM–XEDS analyses of the 500 and 5000 h samples presented here, possibly indicating it has all moved to the surface and oxidized in less than 500 h.

Regardless of the mechanism, the presence of the oxide outlining is significant in that it provides a type of secondary corrosion protection. In the event of a breach by a sulfide through the protective layer, the formation of an oxide over the protruding sulfide will effectively seal off the fast diffusion paths of the sulfide. This is evidenced in Fig. 7, where a large chromium and sulfur rich nodule is seen breaking through the protective aluminum and oxygen rich outer layer. The defect structure of the sulfide, along with its interface with the oxide layer, would provide an opening for accelerated diffusion into the alloy. Development of the aluminum and oxygen rich coating over this nodule, however, should limit any further attack at this location.

Conclusions

Three model iron aluminum alloys were exposed to a simulated coal combustion environment for up to 5000 h and microstructural analysis was performed on the corroded samples. From this work, the following conclusions can be drawn:

  • The Fe–12.5Al–5Cr alloy exhibited the best corrosion performance of the three alloys in this study; no iron sulfides were observed at any test duration. Corrosion resistance of this alloy was attributed to the formation of aluminum and oxygen rich layers, and the ability of the chromium to form chromium sulfide, thus preventing reaction of iron and sulfur and the rapid growth of iron sulfide.

  • The binary Fe–12.5Al alloy exhibited the worst corrosion behavior with extensive iron sulfide formation after 1000 h. Transport of iron and sulfur through an initially formed aluminum oxide was the cause of iron sulfide formation at this time. Subsequent weight gain for this alloy was limited, however, by the development of a secondary aluminum oxide-type rich layer at the metal/scale interface that acted as a diffusion barrier.

  • The Fe–10Al–5Cr alloy began to develop internal iron sulfide protrusions between 1000 and 5000 h of exposure; chromium was observed to form only oxygen-rich corrosion products. In direct contrast to the Fe–12.5Al–5Cr alloy, chromium sulfides did not form on the Fe–10Al–5Cr alloy.

  • The results presented in this work indicate that while chromium additions improve the corrosion resistance of iron aluminum alloys in a simulated coal combustion environment, such improvements are only achieved when the aluminum concentration is high enough to form an initial aluminum oxide layer. Additions of chromium to low aluminum iron alloys may be beneficial only for short term exposures. Longer durations of corrosion exposure result in the formation of deleterious iron sulfide products.

  • Despite the significant weight gain of the binary alloy, no depletion of iron or aluminum was found even after 5000 h of exposure to a corrosive environment. Formation of an aluminum and oxygen rich layer internally on this alloy most likely resulted in the protective behavior observed after 1000 h. Subsequent exposure between 1000 and 5000 h allowed enough time for diffusion to eliminate any depletion zone that may have developed.

  • Encapsulation of external iron sulfide corrosion products with an aluminum and oxygen rich layer was observed for all three alloys. This layer may serve as a secondary barrier to diffusion and may improve corrosion resistance by re-healing any protruding sulfide nodules that develop through an otherwise protective scale.