KCl-Induced Corrosion of the FeCrAl Alloy Kanthal^® AF at 600 °C and the Effect of H₂O

Abstract

The present study investigates the influence of H₂O and KCl on the high-temperature corrosion of the FeCrAl alloy Kanthal^® AF. Polished samples, with and without applied KCl, were exposed isothermally to O₂ or O₂ + H₂O at 600 °C. The samples were investigated using TGA, XRD, SEM/EDX, AES and IC. It was found that KCl accelerates corrosion and that a rapidly growing iron, chromium-rich oxide forms in both environments. Chromate formation and alloy chlorination are shown to initiate the formation of non-protective oxide scales. In addition, aluminium nitrides form in the alloy substrate in both environments.

Introduction

It is well known that the high-temperature corrosion of chromia-forming steels in an oxidising environment is accelerated by alkali chlorides and by high concentrations of water vapour [1–14]. The accelerated effect of water vapour has been shown to be connected to chromia evaporation in the form of CrO₂(OH)₂ [1, 2]. The resulting chromium-depleted (Fe,Cr)₂O₃ mixture has poor protective properties in comparison to the chromium-rich mixed oxide. This corrosion mechanism of stainless steel is significant at temperatures as low as 600 °C.

The high corrosivity of alkali chlorides on stainless steel at elevated temperatures has been attributed to an “active corrosion” (chlorine cycle) mechanism [3–5, 9–12], where alkali chlorides primarily are regarded as sources for molecular chlorine. However, recent reports show that the alkali cation also plays an important role in the corrosion attack [6, 7, 15]. Initially, alkali reacts with chromia in the protective oxide to form a solid alkali chromate (VI). The reactions are rapid at 600 °C and also result in the conversion of the protective oxide into a poorly protective iron-rich scale. Subsequently, chlorine or chloride ions can penetrate the scale, forming transition metal chlorides at the scale metal interface. The formation of sub-scale chlorides further accelerates corrosion due, for example, to decreased scale adhesion.

In contrast to chromia (and the solid-solution (Cr_xFe_1−x)₂O₃), alumina (Al₂O₃) is not expected to form compounds with alkali chlorides at intermediate temperature. Consequently, it appears worthwhile to investigate the usefulness of alumina-forming alloys (e.g., FeCrAl) in environments where chromia-forming stainless steels suffer from rapid corrosion. The literature on the corrosion properties of FeCrAl alloys at temperatures lower than 700 °C is quite scarce, however some studies have been performed [13, 16, 17]. The common view is that the temperature is too low to obtain a protective oxide scale on FeCrAl alloys. This work addresses the corrosion behaviour of an FeCrAl alloy (Kanthal^® AF) in the presence of KCl in dry and wet environments.

Experimental Methods

Material and Preparation

A commercial FeCrAl alloy, Kanthal^® AF containing nominally 21 wt% Cr, 5 wt% Al balanced with Fe (for nominal composition, see Table 1) was used in this study. In addition to the main alloying elements, very small amounts of trace elements (Si, Mn and Mg) and reactive elements (Y, Zr) were present in the alloy. A detailed description of the oxidation behaviour of the reactive elements can be found in Ref. [18]. The samples were ground and polished down to 1 µm finish, thereafter they were ultrasonically cleaned in water, acetone and ethanol. A saturated solution of KCl in water/ethanol was used for applying KCl to the surfaces. The specimen was alternately sprayed and dried with warm air (~35 °C) in order to avoid the formation of droplets on the surface. Each sample was applied with 0.05 mg KCl/cm². The samples were then placed in a desiccator to cool, and their weight was recorded using an analytical balance. The size of the salt crystals was in the range of 10–50 µm, see Fig. 1. The corrosion exposure was started immediately after recording the sample mass.

Table 1 Nominal chemical composition of Kanthal^® AF

Fig. 1

SEM-BSE image of unexposed Kanthal^® AF with KCl applied to the surface

Exposures

The exposures were performed in a horizontal tube furnace and a Setaram instrument with $5\,\% \;{\text{O}}_{ 2} + \, 9 5\,\% \;{\text{N}}_{ 2} \;{\text{or}}\; 5\,\% \;{\text{O}}_{ 2} + \, 40\,\% \;{\text{H}}_{ 2} {\text{O}} \, + \,5 5\,\% \;{\text{N}}_{ 2}$ at 600 °C. The flow rate in the tube furnace was 1,000 ml/min, which corresponds to 3.2 cm/s, and the exposure times were 1, 24, 72 and 168 h. The TGA measurements were performed up to 72 h with a flow rate of 15 ml/min, which corresponds to 0.3 cm/s. Humidification was achieved by saturating the exposure gas with water vapour at the desired dew point (76.4 °C), equivalent to 40 vol% water vapour. The samples were mounted on an alumina holder and introduced to the tube furnace or hung in the Setaram system. After exposure the samples were allowed to cool in dry air in a desiccator. Reference exposures were performed for 168 h in both dry and humid atmosphere in the absence of KCl.

Analytical Techniques

X-Ray Diffraction, XRD

A Siemens D5000 powder diffractometer was used to determine the crystalline corrosion products. The instrument was equipped with a grazing-incidence-beam attachment together with a Göbel mirror. The samples were exposed to a source of CuK_α radiation (λ = 1.5418 Å) with an incident angle of 0.5°–1.5°. The moving detector collected data in the range of 20° < 2θ < 65° with step size of 0.05°. Silicon powder was added to the sample surfaces for calibration. The background was subtracted from the diffraction measurements.

Thermo Gravimetric Analysis, TGA

A Setaram TAG thermobalance was used to study the oxidation kinetics at 600 °C for up to 72 h. An alumina reference sample with the same geometry as the exposed sample was used to diminish the buoyancy effect. In addition, the ex situ weight gains were recorded using an analytical balance. The data were plotted from the time when the isothermal exposure temperature was reached. The time to reach 600 °C was about 6 min.

Ion Chromatography, IC

A Dionex ICS-90 system was used to establish the amount of water-soluble anions (CrO₄ ²⁻, Cl⁻) on the surface after exposure. The anions were analysed with an IonPac AS4A-SC analytic column and with 1.8 mM Na₂CO₃/1.7 mM NaHCO₃ as elution. A Dionex OnGuard II H was used to prevent metal ions from entering the column. The samples were leached in 5 ml Milli-Q water using ultrasonic agitation for 10 + 10 min. The elution was 20 mM sulfonic acid and the flow rate was 2 ml/min. The detection limits for the different species were: Cl⁻ = 0.03 and CrO₄ ²⁻ = 0.01 µmol.

Scanning Electron Microscopy with Energy Dispersive X-ray Spectroscopy, SEM/EDX

The microscopy analysis was done using an FEI Quanta 200 FEG ESEM operated in high vacuum mode. The instrument was equipped with an Oxford Inca EDX system, which was used for chemical quantification and elemental mapping. The ESEM was operated at 8–10 kV for imaging and EDX analysis.

A Gatan Ilion+ broad ion beam system was used to prepare cross sections by sputtering argon ions. The ion gun was operated at 6 kV and the sputter time was 2–3 h.

Auger Electron Spectroscopy, AES

Auger electron spectroscopy was used to estimate oxide thickness and to determine elemental depth distribution. The AES analyses were performed with a PHI 660 scanning auger microprobe (SAM) using an accelerating voltage of 10 kV and a beam current of 200 nA. AES depth profiling was performed using ion sputtering with 3.5 keV Ar+. Two areas were analysed; one smaller 10 × 10 µm² and one larger 90 × 120 µm². They exhibited almost the same oxide thickness and elemental depth distribution.

Quantitative analyses were performed using the peak-to-peak height of the Auger transitions of a specific element together with sensitivity factors provided by PHI, except for the Al and O, which were calibrated against pure Al₂O₃. Since the sensitivity factors for oxygen in Cr₂O₃ and Fe₂O₃ are different from that for Al₂O₃ the oxygen signal will be slightly off when mixtures of the oxides are present [19]. The computer software PHI-Matlab and linear least square (LLS) routines were used to separate the oxide and metal components in the depth profiles.

Results

Oxidation in Dry O₂ and O₂ + H₂O

Exposure of Kanthal^® AF for 168 h in O₂ and O₂ + H₂O at 600 °C in the absence of KCl resulted in very low mass gains, about 7 µg/cm² in dry O₂ and 9 µg/cm² in O₂ + H₂O. After exposure, the surface was almost featureless, being covered by a thin and smooth oxide film. In addition, RE-rich oxide particles (1–3 μm in diameter) were scattered over the surface. After 168 h exposure the oxide film was analysed using AES, showing a thickness of the base oxide of about 45 nm in O₂ (not shown) and 55 nm in O₂ + H₂O (Fig. 2), in good agreement with the recorded mass gains. The analyses showed essentially the same oxide composition in the two environments. The oxide film was dominated by aluminium but significant amounts of iron and chromium were also present. Chromia was enriched in a band in the middle of the oxide film while iron oxide was consistently found outside the chromia band. The bottom part of the film consisted of relatively pure aluminium oxide. Thus the oxide film was dominated by aluminium (30–40 at.% Al) and also contained 0–10 at.% of Fe and Cr. XRD showed clear evidence of chromium-rich corundum-type Me₂O₃ while no aluminium-containing crystalline product was identified. The film formed after 24 h (not shown) was similar to the one formed after 168 h, except that it was slightly thinner and less aluminium-rich.

Fig. 2

AES depth profile of Kanthal^® AF exposed to O₂ + H₂O at 600 °C for 168 h

Oxidation in Dry O₂ in the Presence of KCl

Gravimetry

Figure 3 shows mass gain in O₂ at 600 °C in the presence of KCl. After 72 h, the mass gain was about 40 times higher than the samples exposed in the absence of KCl (not shown). The mass gains were large compared to the amount of KCl added before exposure (0.05 mg/cm²). The mass gain was relatively rapid during the first 24 h and then continued at a slower rate. The samples subjected to 168 h of exposure in the tube furnace exhibited breakaway corrosion and scale spallation during cooling. The in situ TG curves in Fig. 4 provide much more detail on the reaction kinetics than Fig. 3. They are described in the O₂ + H₂O + KCl section below.

Fig. 3

Mass gain versus exposure time for Kanthal^® AF exposed in a horizontal tube furnace at 600 °C in O₂ and O₂ + H₂O in the presence of KCl

Fig. 4

In-situ TGA (mass gain vs. exposure time) for Kanthal^® AF at 600 °C in O₂ and O₂ + H₂O in the presence of KCl

XRD

Figure 5 shows XRD patterns acquired in dry O₂ + KCl. It may be noted that KCl was only detected after one and three hours, indicating that, with time, KCl is completely consumed by reaction and/or vaporization. K₂CrO₄ was detected after 1, 3 and 72 h. Initially, the chromium-rich corundum-type oxide (FeCr)₂O₃ formed on the surface. With exposure time, the corresponding peak positions shifted slightly toward the lower diffraction angle, indicating an increase in cell volume, which is attributed to the increase in the Fe content. After 72 h the positions of the main peaks corresponded to hematite (α-Fe₂O₃), representing the iron-rich end point of the solid solution (FeCr)₂O₃. However, after 72 h there was still some diffraction from chromium-rich (FeCr)₂O₃. After 3 and 24 h of exposure, a few additional weak diffraction peaks appeared that could not be attributed to specific compounds.

Fig. 5

XRD diffractograms for Kanthal^® AF after 1, 3, 24 and 72 h at 600 °C in O₂ + KCl (Si was used as standard). The symbols indicate: K₂CrO₄ (filled diamond), (Fe₂O₃ (filled triangle), Cr₂O₃ (+), KCl (filled circle) and substrate (filled square)

IC

The IC results in Fig. 6 show that the amount of chloride on the samples decreases with exposure time in dry O₂. The chloride detected corresponds to 70, 54, and 20 % of the added amount after 1, 24 and 168 h, respectively. Figure 6 also shows the formation of chromate (VI) on the samples. The maximum amount of CrO₄ ²⁻ that can form on the surface (considering that all KCl added before exposure is converted to potassium chromate, see Reaction (1)) is 0.33 μmol/cm². The amount of chromate detected on the surface was 33 % of the theoretical yield after 1 h and 52 % after 24 h. The amount of chromate decreased after longer exposure times.

Fig. 6

Amount of water soluble ionic substances on the sample surface versus exposure time for Kanthal^® AF exposed at 600 °C in O₂ and in O₂ + H₂O in the presence of KCl

SEM/EDX

SEM/EDX analysis after 1 h of exposure to O₂ + KCl (Fig. 7) showed a thin oxide that covered most of the surface. Much of the thin oxide features 1–10 µm irregularly shaped corrosion product particles on top of the surface. EDX showed chlorine but no potassium in these particles, suggesting that they consist of chlorides formed by the alloying elements (Fe, Cr, Al). Corrosion product agglomerates with 10–100 µm diameters are assumed to have replaced the KCl crystals present before exposure because they showed the same distribution and size and similar shape. They were dominated by iron oxide with some Cr and Al. The shell-like structures that mimic the shape of the original salt crystals are assumed to correspond to partly reacted KCl crystallites (the latter type is shown in the BSE image in Fig. 7). In addition there were rounded corrosion product agglomerations with a diameter of 5–20 μm. The EDX analyses showed that they were dominated by O, Fe, Cr and K and contained very little Cl.

Fig. 7

SEM images (SE left and BSE right) of Kanthal^® AF exposed at 600 °C for 1 h in O₂ with KCl

After 3 h in O₂, most of the thin oxide had been replaced by a rough base oxide that was dominated by iron and also contained considerable amounts of Cr and some Al (Fig. 8). Notably, the Cl-containing corrosion product particles that were so conspicuous after 1 h are rarely seen at this stage. However the corrosion product agglomerates and shell-like structures seen after 1 h remained on the surface after 3 h of exposure. EDX analyses showed that these features have a composition similar to that of the rough base oxide. In the middle of Fig. 8 there are a small number of unreacted KCl crystals. Spherical 5–20 µm diameter corrosion product agglomerations were also present at this stage. They are similar to the corresponding features seen after 1 h but poorer in potassium. In addition, there was a small amount of particles in the size range of 1–3 µm on the scale surface (not shown). EDX point analyses showed that they mainly contained K, Cr and O, which lead to the inference that they consisted of K₂CrO₄, which was supported by XRD (see Fig. 5). The presence of chromate was also verified by the IC analysis (Fig. 6).

Fig. 8

SEM-BSE image of Kanthal^® AF exposed at 600 °C for 3 h in O₂ with KCl

Figure 9 shows EDX maps of the corroded surface after 24 h in dry O₂. The oxide morphology is similar to the one described after 3 h. The main difference is that patches of iron-rich oxide have formed around the former KCl crystals (not shown). In some cases several of these patches have merged, forming larger areas. Al and Cl can be seen in the areas with a thinner oxide scale in Fig. 9, while iron is the dominant cation in the areas covered by a thick scale. The EDX maps also show a few small particles mainly containing K, Cr and O, which have been concluded to be K₂CrO₄. The corrosion product agglomerates contain O, Cr, Fe and Al (about 4 at.% Al, 16 at.% Cr and 22 at.% Fe). The Cl detected by EDX is not correlated to K, which shows that it does not correspond to KCl. Figure 10 is a high-magnification image of the boxed area in Fig. 9 which contains chlorine according to EDX analysis. The Cl-enriched areas appear smooth and bright in the BSE and SE images in Fig. 10 and are also rich in Cr. Since potassium is absent it has been concluded that the Cl is present in the form of chlorides of the alloying elements. The bright angular particles in Fig. 10 consist of iron chromium oxide.

Fig. 9

SEM-BSE image and EDX elemental maps of Kanthal^® AF exposed at 600° C for 24 h in O₂ with KCl

Fig. 10

SEM images (BSE left and SE right) of Kanthal^® AF exposed at 600 °C for 24 h in O₂ with KCl

Figure 11 shows a cross section of scale and substrate after 24 h of exposure in dry O₂. The scale is porous and multi-layered and there is an oxidation-affected zone in the alloy substrate. The EDX analysis in Fig. 12 (showing a limited area of Fig. 11) reveals a succession of four sub-scales on top of each other. Each sub-scale features a more or less continuous thin layer made up of oxides of Al, Fe and Cr. It is proposed that each of these thin layers corresponds to an external scale that was once protective. According to EDX spot analysis the outermost thin layer has higher aluminium content than the thin layers closer to the substrate. The innermost thin layer is dominated by Cr and Fe. Obviously, only the innermost layer is still protective at this stage. In addition to the thin layers, there are large oxide agglomerations, especially in the outermost sub-scale, which correspond to the spherical oxide agglomerations seen in the top-view image in Fig. 9. They are dominated by relatively pure iron oxide but also feature smaller Cr-rich areas at the centre. Areas in the outer subscales exhibiting potassium but no chlorine are tentatively attributed to K₂CrO₄ (compare Fig. 9).

Fig. 11

SEM-BSE image of a BIB cross section of Kanthal^® AF exposed at 600 °C for 24 h in O₂ with KCl

Fig. 12

SEM-BSE image and EDX elemental maps of a BIB cross section of Kanthal^® AF exposed at 600° C for 24 h in O₂ with KCl

The Al and N EDX elemental maps overlap in the oxidation-affected zone immediately below the scale, indicating that aluminium nitrides have formed (see Fig. 12). Cl-enriched areas are found beneath the oxide subscale, which is still in contact with the substrate. Because of the absence of potassium, it has been deduced that these areas consist of chlorides of the alloying elements (Fe, Cr and Al). However, the data (including EDX point analyses) do not allow identification of specific compounds. It may be noted that the chlorides were formed close to the alloy grain boundaries. The same characteristic oxide morphologies (top view) were present after 72 h of exposure (not shown). As noted above, spallation occurred after 168 h.

Oxidation in O₂ + H₂O in the Presence of KCl

Gravimetry

According to Fig. 3 the mass gains in the ex situ O₂ + H₂O + KCl and the O₂ + KCl runs are similar. However, the in situ TG measurements in Fig. 4 reveal that the corrosion kinetics are rather different in the two environments. Thus, in O₂ + H₂O + KCl, the TG curve is linear between approximately 1 and 4 h. Subsequently, mass gain accelerated rapidly. After >10 h, the TG curve becomes rather flat. In dry O₂ there is no linear part of the TG curve.

XRD

The same crystalline corrosion products were identified as in the corresponding dry exposure (compare Figs. 5, 13). K₂CrO₄ was detected after all exposure times except at 168 h (not shown). The K₂CrO₄ peak intensities were higher here than in the dry exposures. Similar to the corresponding dry runs, the (FeCr)₂O₃ solid solution became more iron-rich with time and hematite (α-Fe₂O₃) and Cr rich (FeCr)₂O₃ solid solution appeared after 72 h. In contrast to the dry exposures, there were no unidentified peaks in the diffractograms.

Fig. 13

XRD diffractograms for Kanthal^® AF after 1, 3, 24 and 72 h at 600 °C in O₂ + H₂O + KCl (Si is used as standard). The symbols indicate: K₂CrO₄ (filled diamond), (Fe₂O₃ (filled triangle), Cr₂O₃ (+), KCl (filled circle) and substrate (filled square)

IC

The IC analyses showed that the chloride added before exposure was lost more rapidly than the corresponding dry O₂ runs, see Fig. 6. The formation of chromate (VI) on the surface was also much faster than in dry O₂ (see Fig. 6). Already after 1 h, 0.28 µmol of chromate was formed, corresponding to 85 % of the available potassium on the surface. After 24 h the amount of chromate formed reached 94 % of the theoretical yield. As in the dry O₂ experiment, the amount of chromate decreased >24 h. The IC analysis (see Fig. 6, Table 2) showed that the formation of CrO₄ ²⁻ and the consumption of chloride on the metal surface were, in most cases, in accordance with the stoichiometry of the reaction:

Table 2 Water-soluble anions after exposure in O₂ and O₂ + H₂O (µmol/cm²) and stoichiometry of Reaction (1)

$$2 {\text{KCl(s)}}\,{ + }\, 1 / 2 {\text{Cr}}_{ 2} {\text{O}}_{ 3} ( {\text{s)}}\,{ + }\,{\text{H}}_{ 2} {\text{O(g)}}\,{ + }\, 3 / 4 {\text{O}}_{ 2} ( {\text{g)}} \to {\text{K}}_{ 2} {\text{CrO}}_{ 4} ( {\text{s)}}\,{ + }\, 2 {\text{HCl(g)}}$$

(1)

The stoichiometric relationships in Table 2 show that other mechanisms for chloride loss from the surface, such as KCl volatilization, are comparatively slow. It can be concluded that the Cl that remained on the surface (15 % after 1 h and 6 % after 168 h) was present either as unreacted KCl or as chlorides of Al, Fe or Cr. When spallation occured (168 h) most of the chloride was already lost from the samples, and therefore spallation hardly affected the chlorine balance. Table 2 shows large deviations from the stoichiometry of Reaction (1) after long exposure times, suggesting that K₂CrO₄ decomposes with time.

SEM/EDX

After 1 and 3 h much of the surface was still covered with a thin and smooth oxide scale, the alloy grain boundaries were still visible, (Figs. 14, 15). While the thin oxide is similar to the one formed under dry conditions after 1 h (see Fig. 7), its surface has been covered with potassium chromate particles, and there is little evidence for the chloride-containing corrosion product particles formed in dry conditions. It may be noted that while the thin oxide was replaced by a thick iron-rich oxide after 3 h in dry conditions, it remained intact in the wet environment at that stage (Fig. 15). The K₂CrO₄ particles were much more numerous and no unreacted KCl crystals were observed on the surface. This is in accordance with the IC analysis (see above), which shows that the formation of CrO₄ ²⁻ is faster in O₂ + H₂O environment. The distribution of the K₂CrO₄ crystallites is clearly seen in the EDX maps of Fig. 15. Similar to the dry O₂ case, the original KCl crystals were replaced by corrosion product agglomerations and shell-like formations. In the wet environment, the latter type dominates and corrosion product agglomerates only formed in connection to the largest KCl crystals. Figure 15 also reveals that Cl is enriched at the alloy grain boundaries, indicating that part of the KCl added has formed chlorides with the alloying elements. It was not possible to identify the specific metal chlorides due to the large interaction volume of the EDX analysis.

Fig. 14

SEM-SE image of Kanthal^® AF exposed at 600 °C for 1 h in O₂ + H₂O with KCl

Fig. 15

SEM-BSE image and EDX elemental maps of Kanthal^® AF exposed at 600° C for 3 h in O₂ + H₂O with KCl

After 24 h, the surface morphology was similar to that in the dry O₂ environment, i.e. most of the surface had been covered with a rough iron-rich base oxide containing corrosion product agglomerations formed at the former KCl crystals (compare Figs. 9, 16). Thus, there are almost no areas with the thin smooth base oxide as found after 3 h (compare Figs. 15, 16). Substantial amounts of K₂CrO₄ particles remain after 24 h (see Fig. 16). At this stage, no chlorine was detected using EDX, which implies that KCl is absent. However, the presence of small amounts of chlorides of Al, Cr and Fe below the oxide scale cannot be ruled out.

Fig. 16

SEM-BSE image and EDX elemental maps of Kanthal^® AF exposed at 600° C for 24 h in O₂ + H₂O with KCl

A cross-sectional image shows that the oxide scale formed in the O₂ + H₂O environment is much more compact than that formed in the dry O₂ environment (Figs. 11, 17). Similar to the dry exposure, the oxide scale is layered. K₂CrO₄ particles are situated at the scale/gas interface, immersed in oxide. The EDX point analysis shows that the upper part of the scale mainly consists of iron oxide. It is suggested that this part corresponds to α-Fe₂O₃, which was identified using XRD (Fig. 13). In the middle of the outer (iron oxide) part of the scale there is a band enriched in aluminium and poor in chromium. This band is suggested to correspond to the initial thin and smooth base oxide observed in top-view after 3 h (Fig. 15). The lower part of the oxide scale is relatively rich in Cr and Al (roughly 18 at.% Fe, 18 at.% Cr and 4 at.% Al). Similar to the dry O₂ exposure, the substrate below the scale has been enriched in nitrogen, indicating that alloy nitridation has occurred. In accordance with the plan view SEM/EDX, no chlorine was detected in the scale. This implies that the chlorides formed by the alloy constituents (see the 3 h results above) are no longer present.

Fig. 17

SEM BSE image and EDX elemental maps of a BIB cross section of Kanthal^® AF exposed at 600° C for 24 h in O₂ + H₂O with KCl

The oxidation morphology after 72 h is similar to that seen after 24 h, except that the amount of K₂CrO₄ particles on the surface has decreased (see SE image Fig. 18). At this stage a small part of the surface is still covered by a relatively thin base oxide, see the SEM image in Fig. 18. An enlargement of the area marked by the small square (Fig. 18) reveals small patches with oxide whiskers. The distribution and shape of these patches are reminiscent of the K₂CrO₄ particles seen at an earlier stage. The EDX maps show that the patches are enriched in Cr and O. Small amounts (roughly 2 at.%) of K were also present which may correspond to unreacted K₂CrO₄.

Fig. 18

SEM images (SE left and BSE right) and EDX elemental maps of Kanthal^® AF exposed at 600° C for 72 h in O₂ + H₂O with KCl

Discussion

Oxidation in the absence of KCl resulted in an oxide film dominated by alumina, the aluminium content increasing with exposure time. There was a chromia band in the middle of the oxide film and iron was enriched in the top part (Fig. 2). The chromia band corresponds to Cr₂O₃ which was identified using XRD while the alumina in the film was apparently non-crystalline. This is in accordance with a recent study of Canovic et al. [20] who reports that the oxide film formed on Kanthal^® AF in O₂ and O₂ + H₂O at 600 °C consists of a top corundum-type layer and a bottom layer made up of non-crystalline alumina. A TEM investigation by Liu et al. [21] of the oxide film formed on Kanthal^® AF after 1 h in dry O₂ at 900 °C showed that the oxide film (<100 nm) exhibited a central, chromium enriched band and an outer Fe enriched zone. In contrast to the present study, where the only crystalline compound identified was Cr₂O₃, γ- and α-alumina were identified by Liu et al. [21]. Similar to Liu et al. [21], it is suggested that Cr and Fe enter the oxide film during transient oxidation, i.e., during sample heat-up.

The evidence from SEM/EDX, XRD and IC clearly shows that KCl reacts to form K₂CrO₄ on the alloy surface. As noted in the introduction, FeCrNi and FeCrAl alloys have been reported to react with KCl and water vapour to form K₂CrO₄ [7, 8, 13, 14] according to Reaction (1). For standard state conditions it is found that $\Delta {\text{G}}_{{873{\text{K}}}}^{^\circ }$ = 90.5 kJmol⁻¹K [22] for Reaction (1). The corresponding equilibrium partial pressure of HCl (P_eq(HCl)) is 400 × 10⁻⁶ atm under the experimental conditions (pO₂ = 0.05 atm, pH₂O = 0.40 atm). The continuous flow of gas ensures that p(HCl) << P_eq(HCl) during the experiment. Hence, Reaction (1) is expected to proceed from left to right. The formation of K₂CrO₄ in a dry O₂ environment is less favoured:

$$\, 1 / 2 {\text{Cr}}_{ 2} {\text{O}}_{ 3} ( {\text{s)}}\,{ + }\, 2 {\text{KCl(s)}}\,{ + }\,5 / 4 {\text{O}}_{ 2} ( {\text{g)}} \to {\text{K}}_{ 2} {\text{CrO}}_{ 4} ( {\text{s)}}\,{ + }\,{\text{Cl}}_{2} ( {\text{g)}}$$

(2)

for which $\Delta {\text{G}}_{{873{\text{K}}}}^{^\circ }$ = 90.8 kJmol⁻¹K [22], corresponding to P_eqCl₂ = 0.09 × 10⁻⁶ atm (in 5 % O₂). It may be noted that the “dry” O₂ environment contained on the order of 10 ppm water vapour, corresponding to a P_eq(HCl) of 1.5 × 10⁻⁶ atm by Reaction (1). This implies that K₂CrO₄ formation is dominated by Reaction (1) even in the dry exposures and that Reaction (2) is of secondary importance.

When the material is exposed in the presence of KCl, the salt is initially separated from the alloy by an air-formed film, only a few nanometres thick, consisting of oxides and hydroxides. During heat-up, this film increases in thickness and the hydroxides form oxides. It has been reported by Pettersson et al. that the formation of K₂CrO₄ in Reaction (1) proceeds at a measurable rate from 400 °C [23]. Thus, chromate is expected to form as early as during sample heat-up. At this stage, the oxide is expected to be quite thin (between 3 and 30 nm), and relatively poor in Al. It is suggested that the protective properties of this film are compromised because of the consumption of chromia through Reaction (1) and that this leads to an increase in the oxidation rate of the alloy. The corrosion-accelerating effect of KCl is illustrated by the fact that the number of moles of CrO₄ ²⁻ detected after 1 h in O₂ + H₂O was about 20 times greater than the number of moles of Cr present in the oxide film formed in the absence of KCl, after the same exposure time. Thus, chromium was oxidised more rapidly in the presence of KCl.

The observation that CrO₄ ²⁻ formation was much faster in O₂ + H₂O compared to dry O₂ (see Fig. 6) is in accordance with the K₂CrO₄ formation being dominated by Reaction (1). According to the IC analysis (Table 2), the ratio between formation of chromate (VI) and the amount of chloride lost from the surface satisfies the stoichiometry of Reaction (1) in both dry and wet environments after short exposure times. Consequently, the formation of chromate was accompanied by the loss of chlorine in the form of HCl(g). In addition, it can be concluded that the loss of Cl from the surface by other mechanisms such as evaporation of KCl and transition metal chlorides (see below), is initially of secondary importance.

Under dry conditions, chlorides of the main alloy constituents were observed after 1, 3 and 24 h. In the presence of water vapour, metal chlorination was observed after 1 and 3 h, but not after 24 and 72 h. In other words, the amount of solid chlorides formed was much less than in the dry O₂ environment, e.g. the chloride-containing corrosion product formed on top of the thin oxide in dry O₂ was absent in the presence of O₂ + H₂O. Thus, metal chlorination is much more prominent in a dry environment than in a wet one. It is argued that this is mainly because of Reaction (1), which causes Cl to be lost from the surface in the form of HCl under humid conditions. As observed, chromate formation was slow in “dry” condition, allowing KCl longer time to interact with the alloy. This argument is fully supported by the IC analyses (see Table 2 and Fig. 6).

The TG curves in Fig. 4 further illustrate the different corrosion kinetics in dry and wet environments. It may be noted that the overall mass change is the result of several processes, including oxide growth, the potassium chromate formation by Reaction (1), alloy chlorination, and volatilization of KCl and metal chlorides. The initially relatively slow mass gain in wet condition was attributed partly to the rapid loss of HCl(g) in Reaction (1), and partly by the slower corrosion attack. Considering the small amount of KCl added (corresponding to 24 μg/cm² of Cl⁻) most of the difference should be due to the faster oxidation in dry condition (compare the situation after 3 h in Fig. 8 (dry) and Fig. 15 (wet)). In wet condition mass gain was relatively slow between approximately 1 and 4 h. At this stage chromate formation (Reaction (1)) occurred on the surface which was still protected by a Cr-rich scale. The accelerated mass gain that sets in after about 5 h and then lasts a few hours was attributed to the formation of a rapidly growing iron oxide scale, i.e. breakaway oxidation. The TG curve in dry condition provided little indication of protective behaviour.

The kinetics of the corrosion process can be further elucidated by displaying the mass gain data on a ln(m) versus ln(t) plot (Fig. 19). Assuming that mass gain is proportional to exposure time, the slope in such a plot will correspond to 1/α, where α is the exponent in the kinetic Eq. (3), a constant mass gain corresponding to α = 1, and parabolic oxidation, giving α = 2.

$${\text{m}}^{\alpha } \propto {\text{t}}$$

$$\alpha \ln ({\text{m}}) \propto \ln ({\text{t}})$$

$$\ln ({\text{m}}) \propto (1/\alpha) \ln ({\text{t}})$$

(3)

Fig. 19

ln(m) versus ln(t) of in situ TGA (mass gain vs. exposure time) for Kanthal^® AF at 600 °C in O₂ and O₂ + H₂O in the presence of KCl

The two ln(m) versus ln(t) plots in Fig. 19 display several similarities. The large scatter during the first few minutes after reaching the exposure temperature is attributed to the buoyancy effect. After this stage, there follows a short period during which α was 0.6 in dry and 1.0 in the wet environment, corresponding to accelerated and linear mass gain, respectively. Then comes a stage where α was about 2.0 corresponding to a period with relatively slow mass gain in the O₂ + H₂O TG curve (Fig. 4). It may be noted that while Fig. 19 gives clear evidence of such a stage also in the dry O₂ environment, it was difficult to trace in the corresponding TG curve in Fig. 4. The subsequent breakaway behaviour in O₂ + H₂O (Fig. 4) corresponds to a sudden decrease in the value of α to 0.1. Again, Fig. 19 reveals that there occurred a similar; although less marked transition in dry O₂. During the last stage of the exposure, the value of α was greater in O₂ + H₂O than in dry condition, indicating that a more protective scale has evolved in the former case. This is supported by the cross-sectional micrographs, which showed a dense oxide scale in O₂ + H₂O while a more porous layered scale formed in dry O₂ (see Figs. 11, 17).

The results suggest that we are dealing with two main corrosion modes, namely (I), breakaway oxidation triggered by chromate formation which is dominant in the O₂ + H₂O environment and (II), chlorination of the alloy by KCl accompanied by rapid oxidation, which is predominant in dry O₂. In the O₂ + H₂O environment, chromate formation in Reaction (1) depletes the protective scale in Cr, triggering the formation of a fast-growing iron oxide scale. On the other hand, the concomitant loss of Cl from the surface in the form of HCl(g) decreases the tendency for chlorination of the substrate. In dry conditions, the KCl stays on the surface for a longer time, resulting in a higher degree of alloy chlorination. Thus, it is suggested that alloy chlorination results in the porous multi-layered oxide scales observed in dry O₂ (Fig. 11), while breakaway is caused by the formation of chromate (Reaction (1)) and results in a relatively dense scale, as observed in the presence of water vapour (Fig. 17).

The chlorination of high temperature steels by HCl, KCl and NaCl in the presence of O₂ at high temperatures has been investigated by several authors [5, 9, 13, 24]. Grabke et al. [9] proposed that alkali chlorides “catalyse” corrosion in the following way: First, the chlorine-containing species react to form Cl₂ at the scale/gas interface. Gaseous Cl₂ then penetrates the scale and reacts with the alloy substrate, forming chlorides. The (transition metal) chlorides are then transported (as gas molecules) to the scale surface via cracks and pores in the scale. At the scale surface they react with O₂ to form metal oxide and Cl₂. In this way, Cl₂ is regenerated, closing the so called “chlorine cycle”.

The “chlorine cycle” scenario has some problems. First, the reactions proposed by Grabke et al. [9] and Li et al. [13] for generating Cl₂ at the scale surface appear unable to generate the necessary Cl₂. Thus, of the reactions proposed by Li et al. [13], the most favoured one is Reaction (2) (see above) which generates an equilibrium partial pressure of only 0.09 × 10⁻⁶ atm Cl₂ (at 5 % O₂). In the present case, where the alloy may form an alumina scale, we must examine whether or not the reaction of KCl, O₂ and Al₂O₃ can generate appreciable Cl₂ concentrations. The following reaction is considered:

$$2 {\text{KCl(s)}} + \,{\text{Al}}_{ 2} {\text{O}}_{ 3} ( {\text{s)}} + 1 / 2 {\text{O}}_{ 2} ( {\text{g)}} \to 2 {\text{KAlO}}_{ 2} ( {\text{s)}} + {\text{Cl}}_{ 2} ( {\text{g)}}$$

(4)

$\Delta {\text{G}}_{{873{\text{K}}}}^{^\circ }$ = 197.091 kJmol⁻¹K [22], P_eq Cl₂(g) = 3.6 × 10⁻¹³ atm.

As the equilibrium concentrations of Cl₂ generated in (4) is quite low it is concluded that there is no viable route for generating Cl₂ for the chlorine cycle mechanism.

Another problem with the “chlorine cycle” is the inward transport of Cl₂ to the metal/scale interface where low oxygen activity is postulated to prevail. It is argued that the low oxygen activity at the metal/scale interface would be very difficult to maintain, considering that the cracks and pores open to Cl₂ diffusion must also be open to O₂ diffusion. Moreover, it is considered improbable that the highly reactive chlorine molecule would be able to reach the metal/scale interface without being reduced by the oxide en route (e.g. by Fe²⁺ ions in the scale). Further, because the chlorine cycle mechanism predicts that the metal chloride will form below the scale, it cannot account for the observation in the present paper that metal chlorides also form at the scale/gas interface. Similar observations have been made in the HCl-induced corrosion of FeCrNi steel at 500 °C in 5 % O₂ by Folkesson et al. [25].

It is argued that the problems inherent in the chlorine cycle mechanism can be resolved by applying an electrochemical mechanism for metal chlorination. Since Wagner [26, 27], the formation of external oxide scales on metals at high temperature is usually described in terms of an electrochemical mechanism where the oxidant (e.g. O₂) is reduced at the scale/gas interface while the metal is oxidized at the scale/metal interface, the two electrodes being connected by electronic and ionic currents:

$$1 / 2 {\text{O}}_{ 2} + 2 {\text{e}}^{ - } \to {\text{O}}^{ 2- }$$

(5)

$${\text{Me}} \to {\text{Me}}^{{ 2 { + }}} + 2 {\text{e}}^{ - }$$

.

In the case of the electrochemical chlorination of an alumina-forming alloy by KCl in the presence of O₂, the cathodic process on the scale surface should generate “free” chloride ions at the surface as well as incorporating potassium in a stable form. The analytical results show that as KCl is consumed, potassium becomes associated to Cr. The association of K and Cr is attributed to K₂CrO₄ which is mainly considered to be formed in Reaction (1). However, it is proposed that K₂CrO₄, also may form as part of the following electrochemical reaction:

Scale surface:

$$2 {\text{KCl + 5}}/ 4 {\text{O}}_{ 2} + 1/ 2 {\text{Cr}}_{ 2} {\text{O}}_{ 3} \left( {\text{s}} \right) + 2 {\text{e}}^{ - } \to {\text{K}}_{ 2} {\text{CrO}}_{ 4} \left( {\text{s}} \right) + 2 {\text{Cl}}^{ - }$$

_.

Scale/alloy interface:

$${\text{Fe(s)}} \to {\text{Fe}}^{{ 2 { + }}} ( {\text{oxide)}} + 2 {\text{e}}^{ - }$$

$${\text{Cr(s)}} \to {\text{Cr}}^{{ 2 { + }}} ( {\text{oxide)}} + 2 {\text{e}}^{ - }$$

$${\text{Al(s)}} \to {\text{Al}}^{{ 3 { + }}} ( {\text{oxide)}} + 3 {\text{e}}^{ - }$$

The sum reaction becomes (for the case of Fe oxidation):

$$2 {\text{KCl(s)}}\,{ + }\, 5 / 4 {\text{O}}_{ 2} ( {\text{g)}}\,{ + }\, 1 / 2 {\text{Cr}}_{ 2} {\text{O}}_{ 3} ( {\text{s)}}\,{ + }\,{\text{Fe(alloy)}} \to {\text{K}}_{ 2} {\text{CrO}}_{ 4} ( {\text{s)}}\,{ + }\, 2 {\text{FeCl}}_{ 2} ( {\text{s)}}$$

(6)

For Reaction (6), $\Delta {\text{G}}_{{873{\text{K}}}}^{^\circ }$ = −141.458 kJmol⁻¹K [22] which corresponds to a standard cell potential E^o of +0.73 V, thus showing that the reaction is strongly favoured by thermodynamics. The corresponding anodic reactions with Cr and Al provide even higher thermodynamic driving forces.

The electrochemical mechanism for metal chlorination by alkali chloride requires that the scale can conduct both ions and electrons. Cationic conduction through the scale is known to be dominated by the grain boundary transport at the experimental temperature [28]. Chloride ions do not dissolve in the bulk oxides of interest here. However, similar to hydroxide ions, chloride ions can exist in the adsorbed form, both on the scale surface and in the oxide grain boundaries. Thus, the ionic transport in the electrochemical cell is suggested to occur in the grain boundaries. In this scenario, the relative transport rates of chloride ions and metal ions would determine whether the solid chloride forms on top of the scale or at the scale/alloy interface.

The presence of metal chloride on the metal surface (see Figs. 7, 10) implies that non-equilibrium conditions prevail. This is not unexpected, considering the rapid corrosion process. Because of the high equilibrium vapour pressures at 600 °C (P_eq FeCl₂(g) = 1 × 10⁻³ atm, P_eq CrCl₂(g) = 0.5 × 10⁻⁶ atm) these compounds are expected to vaporize [22]. Accordingly, the transition metal chlorides on the surface are only seen after short exposure times. The observed metal chlorides on the surface (e.g. Figure 10) after 24 h might have a thin oxide layer closest to the oxidising atmosphere. This has been observed by Jonsson et al. at 500 °C on a FeCrNi steel in the presence of HCl(g) [29].

The scale morphology after breakaway in the O₂ + H₂O + KCl environment (see Fig. 17) is similar to the one observed on chromia-forming stainless steels exposed to KCl in the presence of water vapour [14]. Jonsson et al. [14] showed that the scale consisted of an outward growing hematite and an inward growing FeCr spinel oxide. However, further investigation is needed to determine the structure and composition of the oxide scale formed on Kanthal^® AF. Breakaway oxidation triggered by alkali chromate formation has been described for chromia-forming steels [7, 8, 14]. It was concluded that K₂CrO₄ acts as a sink for chromia in the oxide, thereby transforming the protective Cr-rich oxide to rapidly growing iron oxide. The present work indicates that KCl has a similar, corrosion accelerating effect on FeCrAl alloys at 600 °C. The sub-scale formation of aluminium nitrides also shows that the scale formed after breakaway is permeable to N₂. The formation of AlN may obstruct the ability of the alloy to form a protective alumina scale since the reservoir of aluminium close to the metal/scale interface decreases. Aluminum nitride precipitates have been reported to form in FeCrAl alloys at 1,100–1,200 °C when the alumina scale failed during oxidising conditions [30].

Table 2 shows that the amount of chromate on the surface decreases >24 h and that the stoichiometry of Reaction (1) is only satisfied up to 24 h. As spallation is not an issue, this must be due to the decomposition of K₂CrO₄. The thermal decomposition of K₂CrO₄ according to:

$${\text{K}}_{ 2} {\text{CrO}}_{ 4} ( {\text{s)}}\,{ + }\,{\text{H}}_{ 2} {\text{O(g)}}\, \to \, 1 / 2 {\text{Cr}}_{ 2} {\text{O}}_{ 3} ( {\text{s)}}\,{ + }\, 2 {\text{KOH(g)}}\,{ + }\, 3 / 4 {\text{O}}_{ 2} ( {\text{g)}}$$

(7)

is considered insignificant at the relevant temperature ($\Delta {\text{G}}_{{873{\text{K}}}}^{^\circ }$ = 357.7 kJmol⁻¹K [22], corresponding to P_eq KOH = 1.5 × 10⁻²¹ atm (5 % O₂, 40 % H₂O)). In addition, no decomposition was observed when a small amount of K₂CrO₄ was placed on gold foil and exposed for 168 h under the experimental conditions. Instead it is proposed that K₂CrO₄ is cathodically reduced on the scale surface, the electrons being supplied by alloy oxidation:

$${\text{K}}_{ 2} {\text{CrO}}_{ 4} ( {\text{s) + H}}_{ 2} {\text{O(g) + 3e}}^{ - } \to 1 / 2 {\text{Cr}}_{ 2} {\text{O}}_{ 3} ( {\text{s) + 2KOH(l) + 3/2O}}^{2 - }$$

$${\text{Cr(alloy)}} \to {\text{Cr}}^{{ 3 { + }}} + 3 {\text{e}}^{ - }$$

$$\sum :\;{\text{K}}_{ 2} {\text{CrO}}_{ 4} ( {\text{s)}}\,{ + }\,{\text{H}}_{ 2} {\text{O(g)}}\,{ + }\,{\text{Cr(alloy)}}\, \to \,{\text{Cr}}_{ 2} {\text{O}}_{ 3} ( {\text{s)}}\,{ + }\, 2 {\text{KOH(l)}}$$

(8)

$\Delta {\text{G}}_{{873{\text{K}}}}^{^\circ }$ = −231.4 kJmol⁻¹K [22], corresponding to a standard cell potential E^o of +0.80 V. Here, the reduction of K₂CrO₄ on the scale surface is coupled to the oxidation of metallic chromium. The latter reaction can of course be replaced by the oxidation of iron or aluminium. Because O₂ is a stronger oxidant than K₂CrO₄ (thermodynamically speaking), chromate reduction on the scale surface through Reaction (8) is only possible if the oxygen electrode (5) is far from equilibrium.

It may be noted that the reduction of K₂CrO₄ at cathodically active sites at the surface causes alkaline conditions that allow chromate formation to occur through chemical oxidation (9) elsewhere in the scale. In this way, K₂CrO₄ catalyses alloy oxidation.

$$2 {\text{KOH(l)}} + 3 / 4 {\text{O}}_{ 2} ( {\text{g)}} + 1 / 2 {\text{Cr}}_{ 2} {\text{O}}_{ 3} ( {\text{s)}} \to {\text{K}}_{ 2} {\text{CrO}}_{ 4} ( {\text{s)}} + {\text{H}}_{ 2} {\text{O(g)}}$$

(9)

$\Delta {\text{G}}_{{873{\text{K}}}}^{^\circ }$ = − 217.9 kJmol⁻¹K [22].

Conclusions

• KCl strongly accelerated the corrosion of Kanthal^® AF in O₂ and O₂ + H₂O environments at 600 °C. The protective alumina-rich oxide formed in the absence of KCl was replaced with a rapidly growing oxide scale dominated by iron and chromium when KCl was present.

• In an O₂ + H₂O environment, chromia in the scale rapidly reacted with KCl, oxygen and H₂O forming K₂CrO₄ and gaseous HCl. Chromate formation depleted the protective scale in Cr, triggering the formation of a fast-growing iron-rich scale.

• Chromate formation also occurred in a dry O₂ environment but was slower.

• Parallel to chromate formation, the main alloying elements formed chlorides, both on top of and below the oxide scale.

• Alloy chlorination was more pronounced in a dry O₂ than in an O₂ + H₂O environment. This is because chromate formation rapidly converts KCl to HCl(g) in the latter environment. Thus, KCl stays on the surface for a longer time in a dry O₂ environment, resulting in a higher degree of alloy chlorination.

• It is proposed that the chlorination reaction is electrochemical, involving the migration of chloride ions in the grain boundaries of the scale.

• KCl-induced corrosion in a dry O₂ environment resulted in the formation of a porous multi layered oxide scale. In contrast, KCl-induced corrosion in an O₂ + H₂O environment resulted in a relatively dense oxide scale. The difference in behaviour was attributed to metal chlorination being more prominent under dry conditions.

• KCl-induced corrosion was accompanied by the formation of aluminium nitride precipitates in the alloy substrate.