Controlled Growth of α-Al₂O₃ Nanofilm on FeCrAl Alloy as an Effective Cr Barrier for Solid Oxide Fuel Cell (SOFC) Cathode Air Pre-Heaters

Abstract

Solid oxide fuel cell (SOFC) systems often employ metallic cathode air pre-heaters (CAPHs), frequently made from alloys with high chromium (Cr) content, to recover thermal energy from exhaust gases and pre-heat incoming air and fuel. Cr evaporation from metallic CAPHs can poison SOFC cathodes, reducing their durability. To mitigate this, we investigated controlled pre-oxidation of a FeCrAl alloy (alloy 318) to form a protective alumina scale by self-growing, assessing its impact on and oxidation resistance and Cr retention capability for CAPH applications. The effects of pre-oxidation were investigated across a temperature range of 800 to 1100 °C and dwelling times of 0.5 to 4 h. The formed oxide scales were characterised using gravimetry in combination with advanced analytic techniques, such as SEM/EDX, STEM/EDX, TEM, and XRD. Subsequently, the pre-oxidised FeCrAl alloys were characterised with respect to the oxidation rate and Cr₂O₃ evaporation in a tubular furnace at 850 °C, with 6.0 L/min air flow and 3 vol% H₂O to simulate the SOFC cathode environment. TEM analysis confirmed that the FeCrAl alloys formed alumina scales with 10 nm and 34 nm thickness after 1 h of pre-oxidation at 900 and 1100 °C, respectively. The corrosion and Cr₂O₃ evaporation rates of the FeCrAl alloy at 850 °C in humidified air were shown to be dramatically decreased by pre-oxidation. It was found that the mechanisms of oxidation and Cr₂O₃ evaporation were found to be controlled by the formation of different alumina phases during the pre-oxidation. Measurements of Cr₂O₃ evaporation and weight gain revealed that the alloy 318 pre-treated at 1100 °C for 1 h will form an α-Al₂O₃ scale, leading to a 98% reduction of the oxidation rate and 90% reduction of Cr₂O₃ evaporation compared to the non-oxidised alloy 318 under simulated SOFC cathode conditions.

1. Introduction

Combined heat and power (CHP) systems utilising solid oxide fuel cells (SOFCs) offer a promising approach for future electricity and heat generation in residential and commercial buildings, owing to their low emissions and high energy efficiency [1,2]. Due to the high operating temperature of the SOFC stack and to avoid thermal shock from cold gas supply, the temperature of incoming air must be raised to a temperature close to that of the SOFC stack [3]. To achieve this, counter-current cathode air pre-heaters (CAPHs) are generally employed to recover the thermal energy from the afterburner of the SOFC exhaust gases to heat the air flowing towards the cathode. Remarkable progress has been achieved in the reduction of the operating temperature of SOFCs to 700 °C through the development of novel electrode materials and the decrease of the electrolyte thickness [4,5,6]. Furthermore, by lowering the operating temperature, a wider range of lower-cost materials can be used, especially in relation to the balance of plant (BOP) and stack interconnects [4,7,8].

CAPHs made from metallic materials have received substantial attention because of their high thermal conductivity, formability, manufacturability, and superior mechanical properties for use in SOFC systems. However, at SOFC cathode input conditions (air with 3 vol% of steam), the typical chromia-forming alloys used in commercial heat exchangers allow Cr to evaporate, causing SOFC cathode poisoning and degradation, and therefore reducing the SOFC performance and lifetime [9]. Significant research have been devoted to addressing chromium poisoning from metallic interconnects by developing suitable coatings [10,11]. However, the coatings developed for metallic interconnects may not be completely suitable for the BoP components because of their complex geometry [12].

It is worth noting the slower growth rate and higher water vapour resistance of the alumina scale compared to the chromia scale, making it a promising candidate for use in Cr retention in water-vapour-containing environments [13,14]. A broad range of techniques have been investigated to produce protective Al₂O₃ layers on Fe-based alloys. Grishina et al. produced a 0.3 to 1 µm Al₂O₃ coating on low-carbon steel by Sol-gel coating, lowering the oxidation rate by 70%. However, each coat–dry–fire cycle adds processing time and organic burnout steps [15]. Gong et al. developed 20–100 nm Al₂O₃ films by atomic layer deposition (ALD) on the top of stainless steel 316L, which significantly improved the corrosion resistance compared to wrought SS316L. Atomic-layer deposition (ALD) offers high-level thickness control and pin-hole-free Al₂O₃ even inside complex porosity, but it has drawbacks, like sub-nm growth rates and high capital cost [16]. Krumdieck et al. employed pulsed-pressure metalorganic chemical vapour deposition (PP-MOCVD) to deposit ≈200 nm amorphous Al₂O₃ at 450–800 °C with excellent conformity, but the technique required a specialised vacuum-reactor infrastructure and pyrophoric precursors [17]. Wu et al. produced 25–180 µm double-layer aluminised coatings on Inconel 625 by pack cementation, aluminising at 850 °C, which formed a protective 10 µm α-Al₂O₃ after super-critical oxidation at 500 °C for 72 h, but the 3–5 h dwell at 850 °C and large Al uptake partly depleted the Al reservoir and added processing time [18]. Compared with these multistep techniques, the present one-hour pre-oxidation approach aims to leverage the alloy’s own aluminium content, offering a simpler and lower capital cost method, and it is compatible with complex CAPH geometries. Thus, the development of a self-growing alumina scale on Al-containing alloys could be a promising solution for this case. Previous work of ours has proven that the alumina-forming alloy (alloy 318) forms an alumina scale on its surface, which acts as a protective layer and reduces the Cr evaporation rate to an order of magnitude lower than chromia-forming alloys [8,19]. Zhou et al. confirmed the formation of a continuous alumina scale on the alumina-forming alloys after 5000 h of exposure at 900 °C in air with 10% water vapour [20]. The European HEATSTACK project [21] implemented a field test of a customised micro-CHP integrated with an alloy 318 CAPH at a customer site for a hot-operation time of 25,000 h. However, the longer-term operation of the alloy 318 heat exchanger in the SOFC cathode environment showed a large amount of Cr₂O₃ formation around the exhaust outlet (cold zone) and a fast Al oxidation rate around the exhaust inlet (hot zone; Figure 1). Apparently, the cold zone was too cold to allow the alloy to form a sufficient alumina scale, leaving the chromium exposed. Or, in other words, in the cold zone the chromia formation rate (and hence the potential to form volatile compounds) was faster than the alumina formation rate, thus lacking the inhibiting alumina layer. Therefore, pre-oxidation was proposed to allow sufficient formation of the protective alumina scale on the alloy 318 CAPH prior to the application in SOFC environments.

Studies on the effects of pre-oxidation on alumina-forming alloys have been reported in the literature [22,23], showing huge improvement in corrosion resistance. Stanislowski et al. [22] found an improvement in the corrosion resistance and Cr retention ability of alumina formers, such as AluChrom YHf, AluChrom YB, Kanthal AF, and PM 2000, after a pre-oxidation at 1200 °C in air for 24 h by forming an alumina scale with thickness around 7 µm. Similarly, Gomez-Vidal et al. [23] explored pre-oxidation of alumina-forming alloys (e.g., Inconel 702 and Kanthal APMT) at high temperatures, such as 1050 °C, for 4 h to form protective alumina scales against molten chloride attack, demonstrating good performance. While these studies show the efficacy of high-temperature pre-oxidation, such extreme conditions (e.g., 1050–1200 °C, extended durations) present considerable drawbacks for practical CAPH manufacturing, primarily due to significantly increased energy costs, processing complexity, and the substantial risk of excessive aluminium reservoir depletion within the alloy. This rapid Al consumption could, paradoxically, shorten the protective lifespan of the component during subsequent high-temperature operation in a SOFC system. These limitations underscore the need for pre-oxidation strategies that can achieve a sufficiently protective alumina scale under more industrially viable conditions, i.e., lower temperatures and shorter durations. Furthermore, the corrosion protection of FeCrAl alloys is provided by the slowly growing α-Al₂O₃ scale that forms at high temperatures (>1000 °C). The oxidation rate will rapidly slow down once a continuous layer of α-Al₂O₃ is formed [24,25]. Thus, the formation of a continuous α-Al₂O₃ scale is highly desirable for alumina-forming alloys used in CAPH applications. However, initially and at lower temperatures, outward growth of metastable alumina forms, which are less protective because of higher growth rates and higher defect concentrations, is observed [24]. The most frequently reported transient alumina forms on alumina-forming alloys are γ-, δ-, and θ-Al₂O₃ [26]. At temperatures higher than 1000 °C, mainly α-Al₂O₃ is formed [27,28,29]. However, many studies concerning oxidation of FeCrAl alloys also observed the formation of α-Al₂O₃ in the relatively low temperature range of 650 to 900 °C [30,31,32,33,34,35,36]. Mutual results from these studies are that the transient oxides (α-Cr₂O₃ and α-Fe₂O₃), which are isostructural with α-Al₂O₃, act as crystallographic templates for the nucleation and growth of α-Al₂O₃. The simultaneously formed metastable alumina will, after a certain time, transform into the stable α-Al₂O₃. However, the phase transformation rate of such metastable alumina strongly depends on the exposure temperature and hence is relatively slow at temperatures below 1000 °C [37]. It can be inferred that both the pre-oxidation time and temperature will have a significant influence on the phase composition of the alumina scale formed on FeCrAl alloys.

In the current work, the pre-oxidation of alloy 318 is suggested to be performed directly in air atmosphere at relatively low temperatures (800 to 1100 °C) and short dwelling times (0.5 to 4 h) to form the most stable alumina phase (α-Al₂O₃) with an ultrathin thickness on both inner and outer surfaces of the single CAPH cell, whilst requiring as little time as possible to minimise the consumption of Al reservoir within the alloy. This would make the pre-oxidation process cost-effective and suitable for industrialisation. While pre-oxidation is a known strategy, the specific novelty of this work lies in the systematic investigation of the impact of lower-temperature and short-duration pre-oxidation treatments on the alumina phase evolution, microstructure, oxidation resistance, and Cr retention capabilities of a commercial FeCrAl alloy, aiming to identify a cost-effective and industrially scalable process for enhancing SOFC CAPH durability. The aim of this research is to investigate the effect of the pre-oxidation temperature and time on the phase formation of the alumina scale and measure the chromium release from the alloy 318 pre-oxidised at different conditions by means of the denuder technique. The suitability of pre-oxidation for alloy 318 in SOFC CAPH applications in terms of oxidation resistance and Cr₂O₃ retention capability will be discussed.

2. Materials and Methods

2.1. Materials

The commercial AluChrom 318 (FeCrAl alloy) with 0.3 mm thickness was selected for this research. The samples were cut into coupons of the dimensions 15 × 15 mm². The alloy composition for the alloy 318 in wt.% is shown in

Table 1. Nominal compositions of the alloy 318 in wt.%.

(wt.%) Nominal	Fe	Cr	Mn	Al	Ni	Si	Nb	Others
Alloy 318	74.06	18.8	0.21	3.58	0.2	0.32	0.73	Hf 0.06; Y 0.07; Zr 0.03; Cu 0.03; C 0.01; N 0.01; S 0.002; W 2.02

. The surface roughness (Ra) of alloy 318 prior to exposure was 0.11 ± 0.02 µm, as determined using an atomic force microscope (AFM) (cli Dimension 3100, Veeco Instruments Inc., Plainview, NY, USA) equipped with a Nanoscope 4 controller. The samples were exposed in as-received condition after having been cleaned in acetone and ethanol using an ultrasonic bath. Moreover, a 50-cell alloy 318 CPAH operated under the SOFC cathode environment for 25,000 h was provided by Vaillant GmbH (Remscheid, Germany). Coupons were cut from both the hot zone and cold zone on one of the single cells (marked in Figure 1a) to analyse the microstructure evolution of both inner and outer surfaces of the alloy 318 plate exposed to two extreme conditions from a real-life operating system.

2.2. Pre-Oxidation Percedures

The effect of pre-oxidation on the alloy 318 was investigated with regards to two variables: temperature and dwelling time. All the pre-heat oxidations were conducted in a tubular furnace (Vecstar HZ ST 1100, Vecstar, Chesterfield, UK) with the samples standing vertically on an alumina plate in air atmosphere (with zero gas flow). The samples were heated to the target temperature with a heating rate of 5 °C/min, held at the target temperature for the programmed duration, and subsequently cooled down to room temperature inside the furnace with a cooling rate of 5 °C/min. The matrix of pre-oxidation conditions is shown in

Table 2. The matrix for the pre-oxidation conditions of the alloy 318.

Temperature	Dwelling Time
800 °C	1 h	2 h	4 h
900 °C	1 h	2 h	4 h
1000 °C	1 h	2 h	4 h
1100 °C	30 min	1 h	2 h

.

2.3. Exposure Tests

Measurement of volatile Cr(VI) species for the Cr₂O₃ evaporation assessment was achieved by exposing three identical samples continuously for 168 h at 850 °C. This process employed a denuder technique within a Vecstar HZ ST 1100 tubular furnace, under an atmosphere of 6.0 L/min air with 3 vol% H₂O. The denuder apparatus employed a quartz glass tube internally coated with Na₂CO₃ for capturing Cr(VI). The chromium-containing gas, CrO₂(OH)₂, generated in the simulated SOFC conditions, interacted with this Na₂CO₃ layer, resulting in the formation of Na₂CrO₄, as detailed in Equation (1) [38]:

CrO₂(OH)₂ (g) + Na₂CO₃ (s) → Na₂CrO₄ (s) + H₂O (g) + CO₂ (g).

(1)

After the Cr(VI) exposure test, the reacted denuder tube was subsequently flushed with 20 mL of deionised (DI) water, and the Cr-containing solution was carefully collected. Continuous and uninterrupted assessment of Cr₂O₃ evaporation during the test period was achieved by replacing used denuder tubes with freshly Na₂CO₃-coated ones without reducing the furnace temperature. A Perkin Elmer Optima 8000 inductively coupled plasma–optical emission spectrometer (ICP-OES) was then utilised to quantify the total evaporated Cr(VI) species from the collected Na₂CrO₄-containing solutions. All Cr concentrations obtained from ICP-OES in this investigation were subsequently converted to their equivalent Cr₂O₃ mass. Thus, the accumulated mass loss by Cr evaporation was calibrated to the specific mass of Cr₂O₃, and this re-calculated mass was plotted as a function of time. Details about the denuder technique can be found in Refs. [38,39]. To further investigate their behaviour, various pre-oxidised samples were also subjected to a long-term high-temperature oxidation test. This involved a simultaneous 500 h exposure in a tubular furnace at 850 °C, with air (6.0 L/min) containing 3 vol% H₂O. Samples from this oxidation test were periodically withdrawn after furnace cooling at extending intervals to facilitate gravimetric analysis and physical characterisations, aimed at determining oxidation rates and monitoring oxide layer formation.

2.4. Microstructural Analysis

A high-resolution balance with a capacity resolution of 2.1 g × 0.1 μg (Sartorius Cubis Ultra-Micro Balance, MSA2.7S0TRDM, Sartorius AG GmbH, Göttingen, Germany) was used for all the gravimetric measurements. The mass change of the coupons was recorded before and after each pre-oxidation to evaluate the effect of temperature and dwelling time on the oxidation rate. For high-temperature oxidation tests, the mass change of the pre-oxidised samples was recorded at regular time intervals. To determine the crystalline phases within the oxide layers on the steel, X-ray diffraction (XRD) was performed using a Bruker D2 PHASER (2nd generation) instrument, employing Co-Kα radiation (λ = 0.179026). Surface morphological features of both untreated and oxidised samples were examined with a Hitachi TM3030 Plus scanning electron microscope integrated with an energy-dispersive X-ray spectrometer (SEM/EDX), primarily operating in backscattered electron (BSE) mode.

For cross-sectional evaluation, coupons sectioned from a CAPH stack’s single cell were embedded in epoxy resin and subsequently mechanically polished to a 1 μm finish, then analysed with the identical SEM/EDX setup. The detailed cross-sectional microstructures of oxide scales on pre-oxidised alloy 318 were scrutinised using analytical scanning transmission electron microscopy (STEM) combined with EDX capabilities on a Philips Tecnai F20 microscope (200 kV) (Amsterdam, The Netherlands) outfitted with an Oxford Instruments ISIS EDX system (Abingdon-on-Thames, UK). Furthermore, a JEOL 2100 TEM (Tokyo, Japan) was utilised with convergent beam electron diffraction (CBED) to ascertain the oxide phase identity on the pre-oxidised alloy 318. Preparation of cross-sectional TEM samples from these pre-oxidised specimens involved an in situ lift-out method performed with an FEI Quanta 3D dual-beam focused ion beam/scanning electron microscope (FIB/SEM) (Hillsboro, Oregon, USA), and this procedure is extensively detailed in other literature [40].

3. Results

3.1. Microstructural Analysis of an Alloy 318 CAPH

Figure 1 illustrates the condition of a CAPH cell after extended operation and the typical heat distribution. Figure 1a shows a single counter-current alloy 318 CAPH cell operated under the SOFC cathode environment for 25,000 h. Figure 1b shows a schematic of the heat distribution on a single CAPH cell during operation. Due to the convective flow between the hot exhaust gas and the incoming cold air, the temperature gradient across the heat exchanger plate caused different oxidation states from high- to low-temperature regions. The oxidation stage of the CAPH cell under the different temperature regimes was evident by the difference in surface colour, as shown in Figure 1a. Figure 2 shows the cross-sectional SEM images of both inner and outer surfaces of a single CPAH cell exposed in the hot zone and cold zone, respectively. In the hot zone, both inner and outer surfaces of the CAPH cells were protected by alumina layers since the EDX elemental maps and line scans showed high intensity of Al in the oxide. However, the alumina scale was found to be thicker on the outer surface (4.38 ± 0.28 μm) than on the inner surface (3.41 ± 0.21 μm), indicating a faster oxidation rate on the exhaust side than on the air side. In the cold zone, a thick Cr₂O₃ layer (2.15 ± 0.20 μm) was formed on the outer surface of the single CAPH cell (exhaust side) owing to the extremely low growth rate of Al₂O₃ at low temperatures. This could result in a large amount of Cr(VI) leakage to the environment. Besides, no oxide layer was formed on the inner surface of the cell exposed to the cold zone due to the extremely low temperature of the inlet air (room temperature). Based on these findings, a pre-oxidation on the alloy 318 was suggested in order to enable the heat exchanger plates to form an alumina scale prior to the application in the SOFC system. The pre-oxidation is expected to produce an α-Al₂O₃ scale on the entire heat exchanger surface, which would reduce the growth rate of the alumina scale in the hot zone and subdue the formation of the Cr₂O₃ scale in the cold zone.

3.2. Pre-Oxidation

3.2.1. Mass Measurement

Figure 3 shows the effect of pre-oxidation temperature and dwelling time on the oxidation behaviour of alloy 318. The mass increase obtained by pre-oxidation corresponded to the reaction between oxygen and the alloy elements, which was mainly the establishment of an alumina layer in the case of an alumina-forming alloy. It can be clearly seen that the mass gain values increased with the pre-oxidation temperature and time. The mass gain showed a steady increase with the pre-oxidation dwelling time within the tested range (up to 4 h; Figure 3a), while it had an exponential relation to the pre-oxidation temperature (Figure 3b). Moreover, depending on the pre-oxidation temperature, the slope of each linear relationship increased with the temperature, as shown in Figure 3a.

3.2.2. Surface Charicaterisation

In Figure 4, photographs of the unoxidised alloy 318 and the alloy 318 pre-oxidised at 800, 900, 1000, and 1100 °C in dry air for 1 h are shown. The unoxidised alloy 318 showed a shiny metallic surface. The pre-oxidised samples at 800, 900, 1000, and 1100 °C exhibited dark blue, green, gold, and grey colours, respectively. The different colours presented by the alloy surfaces indicate the oxidation degree of the alloy 318 at different temperatures. Figure 5 shows the concentration of four critical elements (O, Al, Cr, and Fe) present on the alloy surface after pre-oxidation at different temperatures from 800 to 1100 °C. For all four pre-oxidation temperatures, the concentrations of Al and O increased with the ascending dwelling time, which corresponded to a continuous growth of the alumina scale on the alloy surface. The decreased concentration of Fe and Cr could be attributed to the increased coverage by the alumina scale on the alloy surface, since aluminium will preferentially oxidise prior to Cr and Fe when the concentration of Al is higher than 3 wt.% in the alumina-forming alloy [41]. The pre-oxidised sample at 800 °C showed the highest surface concentration of Cr (12 at.%), in comparison with the lowest Cr surface concentration of 2.75 at.% for the sample pre-oxidised at 1100 °C for 2 h. It is also interesting to note that the surface concentrations of the four elements on the samples pre-oxidised at 800 °C for 4 h were comparable to those on the samples pre-oxidised at 900 °C for 1 h. In addition, the samples pre-oxidised at 900 °C for 4 h and 1000 °C for 1 h demonstrated quite similar levels of surface concentration of O, Al, Cr, and Fe. The element concentrations of the surfaces of samples pre-oxidised at 1000 °C for 4 h were also found to be close to those of the samples pre-oxidised at 1100 °C for 30 min.

The alloy 318 pre-oxidised at 800, 900, 1000, and 1100 °C for 1 h was characterised by XRD to examine the effect of temperature on the alumina crystalline structure. The diffractograms obtained are displayed in Figure 6. Peak pattern matching of the XRD spectra showed the presence of Fe₂Nb for the samples pre-oxidised at all four temperatures. For the samples pre-oxidised at 1000 and 1100 °C, XRD confirmed the formation of an α-Al₂O₃ phase on the alloy surface. However, it is worth noting that the phase of alumina formed on the samples after being pre-oxidised at 800 and 900 °C could not be detected by XRD because the oxide layers were extremely thin, resulting in a very low intensity of the diffraction peaks, which did not allow identification with sufficient precision [34,39].

The SEM micrographs of all pre-oxidised samples are displayed in Figure 7. Figure 8 compares the EDX line scan intensity of Al and Cr along the line shown from the samples pre-oxidised for 1 h at the four different temperatures. After pre-oxidation at 800 °C for 1, 2, and 4 h, the surfaces still looked metallic, and the original scratches left from manufacturing were still visible on the surface. As indicated by the EDX line scan, the surface of the alloy 318 after being oxidised at 800 °C for 1 h exhibited the lowest intensity of Al and the highest intensity of Cr among all the 1 h pre-oxidised samples. The samples pre-oxidised at 900 °C had built more alumina on the alloy surface in comparison. However, the distribution of alumina on the surface was considered to be inhomogeneous. As indicated by the EDX line scan for the sample pre-oxidised at 900 °C for 1 h, the dark region showed an extremely high intensity of Al and low intensity of Cr corresponding to areas covered with a thick alumina scale. Furthermore, more areas on the surface were observed to be covered with a thick alumina scale with the pre-oxidation dwelling time increased to 4 h at 900 °C. In contrast, the grey region with lower intensity of Al and slightly higher intensity of Cr corresponded to the formation of a thinner alumina scale on the alloy surface. The alumina scale developed on the samples after being pre-oxidised at 1000 °C had an inhomogeneous thickness. The inhomogeneity of the alumina distribution at 900 °C and 1000 °C (Figure 7d–i) was qualitatively assessed based on variations in SEM backscattered electron (BSE) image contrast and features observed in the EDX line scans (Figure 8). This can be confirmed by our previous detailed investigation on the same FeCrAl [39], where it was proved by STEM with EDX mapping and TEM analysis that the darker grey, Al-rich regions (described as ‘ridged alumina’) in BSE images corresponded to significantly thicker alumina layers compared to lighter grey ‘plain’ alumina regions. The partially covered thick alumina formed on the samples pre-oxidised at 1000 °C exhibited a platelet-like structure, which was completely different from that formed on the samples pre-oxidised at 900 °C. The EDX line scan confirmed that the platelet-like alumina showed the highest intensity of Al and the lowest intensity of Cr. With the increase in dwelling time, the platelets considerably expanded in size. However, the thin alumina area in the sample pre-oxidised at 1000 °C for 1 h still showed a lower Al intensity and higher intensity of Cr in comparison with the sample pre-heated at 1100 °C for 1 h. Unlike the samples pre-heated at 900 and 1000 °C, the distribution of the alumina scale on the samples pre-oxidised at 1100 °C was much more homogeneous, with the highest intensity of Al and the lowest intensity of Cr. It is noted that the platelet-like alumina formed on the alloy 318 pre-oxidised at 1100 °C was much less compared to that formed at 900 °C. Furthermore, when the dwelling time increased to 2 h at 1100 °C, the platelet structure almost disappeared, and only a dense oxide layer was observed. The TEM images for samples pre-oxidised at 1100 °C for 1 h were prepared from the region covered by both a dense oxide layer and the platelet-like alumina scale.

3.2.3. Cross-Sectional Analysis

Figure 9 shows the cross-sectional TEM images of the pre-oxidised samples. Cross-sectional TEM specimens were prepared by FIB lift-out from regions on the pre-oxidised samples that, based on prior SEM imaging (Figure 7), appeared representative of the overall scale coverage and morphology for each condition. Figure 9a shows a cross-section through the oxide scale formed after 1 h of exposure in dry air at 900 °C. It can be seen that the oxide scale was extremely thin and had a thickness around 10 nm. Both Liu et al. [30] and Gotlind et al. [34] reported the formation of a double-layered alumina scale with 100 nm thickness on Kanthal AF (Al 5.1 wt.%) after exposure in dry O₂ at 900 °C for 1 h. However, the duplex structure of the scale formed on the alloy 318 sample after 1 h of pre-oxidation at 900 °C was difficult to distinguish even in a TEM micrograph at high magnification due to the extremely low thickness. This may also be the reason for the failure of XRD to detect the alumina phase for samples pre-oxidised at 800 and 900 °C, where the amount of alumina formed was too low and below the detection limit of the XRD technique. The TEM images in Figure 9b display a cross-section through the oxide scale formed on alloy 318 after 1 h of pre-oxidation in air at 1100 °C. The oxide scale was around 34 nm thick and had a clear double-layered structure with an outer and inner layer, with the inner layer making up approximately two-thirds of the total thickness. As can be seen, this oxide thickness was about three times higher than that formed at 900 °C. This was consistent with the mass measurements. The mass gain of the alloy 318 pre-oxidised at 1100 °C for 1 h was three times higher than that pre-oxidised at 900 °C (Figure 3a). A STEM cross-sectional image of the oxide scale formed on the alloy 318 pre-oxidised at 1100 °C for 1 h is shown in Figure 9c, accompanied by EDX elemental maps. The Cr map detected a narrow Cr-rich band located at the interface between outer and inner alumina layers. In the Fe map, the distribution of Fe on the oxide scale only showed a very weak signal in the outer layer. EDX line scans (Figure 9d) were performed in order to further analyse the elemental distribution on the oxide scale. As can be seen, the compositional profiles of the oxide scale exhibited three different regions. Region I was the inner α-A₂O₃ layer and contained the highest level of Al, with no Cr and Fe. Region II had a high concentration of Cr, indicating the Cr-rich band detected in the Cr map. Region III was the outer alumina layer and had a relatively lower Al concentration compared to region I. In addition, a small amount of Fe and Cr were also observed in region III.

3.3. Exposure Test

3.3.1. Mass Measurements

The mass gain of the pre-oxidised samples after 500 h of exposure at 850 °C in 3 vol% humidified air (6.0 L/min) is plotted in Figure 10a, in comparison with the non-oxidised sample. It can be clearly seen that all the pre-oxidised samples showed a lower mass gain compared to the unoxidised sample (0.130 mg/cm²) over the 500 h of exposure test. In this study, a low mass gain corresponded to a low oxidation rate for the pre-oxidised samples exposed to the simulated SOFC cathode conditions. The samples pre-oxidised at 800 °C for 1 h showed the highest mass gain (0.072 mg/cm²) among all the pre-oxidised samples after 500 h of exposure, while the mass gain was found to dramatically decrease with the pre-oxidised samples when the dwelling time increased to 4 h (0.035 mg/cm²) at the same pre-oxidation temperature. By comparing the same dwelling time and varying the temperature, the samples pre-oxidised at 900 °C for 1 h (0.059 mg/cm²) showed a lower mass gain than the samples oxidised at 800 °C for 1 h, and the samples pre-oxidised at 900 °C for 4 h (0.027 mg/cm²) showed a lower mass gain than the samples pre-oxidised at 800 °C for 4 h, following the exposure tests. The overall mass gain of the samples pre-oxidised at 1100 °C for 30 min (3.36 × 10⁻³ mg/cm²) was slightly lower than the samples pre-oxidised at 1000 °C for 4 h (4.44 × 10⁻³ mg/cm²) after 500 h of exposure. The best corrosion resistance was observed for the samples pre-heated at 1100 °C for 1 h (4.37 × 10⁻⁴ mg/cm²), with a 98% reduction of the oxidation rate compared to the unoxidised alloy 318 owing to the formation of the α-Al₂O₃ layer.

3.3.2. Cr₂O₃ Evaporation

The evaporation of Cr₂O₃ from the oxidised and unoxidised alloy 318 is plotted in Figure 10b. It is interesting to note that the samples pre-oxidised at 800 °C for 1 h showed the highest amount of Cr₂O₃ evaporation (3.18 × 10⁻³ mg/cm²). In contrast, the amount of Cr₂O₃ evaporated from the samples pre-oxidised at 800 °C for 4 h (1.67 × 10⁻³ mg/cm²) was lower than the unoxidised samples (2.84 × 10⁻³ mg/cm²). The samples pre-heated at 900 °C for 1 h had a similar level of Cr₂O₃ evaporation (1.69 × 10⁻³ mg/cm²) as those pre-heated at 800 °C for 4 h. Whereas, with the pre-oxidation time increasing to 4 h at 900 °C, the amount of Cr₂O₃ evaporated from the alloys had decreased to 9.88 × 10⁻⁴ mg/cm², which is 35% of that from unoxidised samples. The samples pre-oxidised at 1000 °C for 1 h showed an amount of Cr₂O₃ vaporisation of 1.13 × 10⁻³ mg/cm², which was comparable to that from the samples pre-oxidised at 900 °C for 4 h. It is worth noting that the samples pre-heated at 1100 °C displayed the lowest level of Cr₂O₃ evaporation among all the pre-oxidation temperatures investigated in this research. The quantity of Cr₂O₃ evaporated from the unoxidised sample (2.84 × 10⁻³ mg/cm²) was 5 times higher than that from the sample pre-oxidised at 1100 °C for 30 min (5.97 × 10⁻⁴ mg/cm²) and 10 times higher than that from the sample pre-oxidised at 1100 °C for 1 h (3.03 × 10⁻⁴ mg/cm²) over the 168 h of collection.

3.3.3. Surface Analysis

Figure 11 compares the surface morphology of unoxidised alloy 318 and pre-oxidised alloy 318 for 1 h at 800, 900, 1000, and 1100 °C before and after the 500 h exposure at 850 °C in 3% humidified air. Following a 500 h exposure period, the unoxidised alloy 318 revealed two distinct zones of alumina formation on its surface. These comprised areas with a dark grey appearance, featuring both globular and long-ridged alumina, alongside regions of flat, plain alumina that appeared light grey. The formation mechanism of the long-ridged alumina on the alloy 318 surface has been elaborated in previous work [39]. However, this surface morphology was more noticeable in the SEM images of samples pre-oxidised at 800 and 900 °C than the unoxidised one after 500 h of exposure. For the samples pre-oxidised at 1000 °C for 1 h, the partially covered platelet-like alumina had grown larger. It is worth noting that the morphology of the alumina scale formed on the samples pre-oxidised at 1100 °C did not show any change before and after 500 h of exposure in the simulated SOFC atmosphere.

4. Discussion

4.1. Pre-Heat Oxidation

For the pre-oxidised samples in Figure 3, the mass gain showed a linear increase with dwelling time and an exponential growth with temperature. For samples pre-oxidised at the same temperature, the Al diffusion rate from the alloy base was constant for short-term pre-oxidation, which resulted in the growth of the alumina scale at a constant rate. Whereas, for the samples pre-oxidised for the same dwelling time but with different temperatures, the exponentially increased Al diffusion rate at elevated temperatures led to much faster growth of alumina on the surface. This was further confirmed in the EDX analysis in Figure 5. The surface concentration of Al for the samples pre-oxidised for the longest dwelling time was nearly equivalent to that pre-oxidised for the shortest dwelling time with a 100 °C increase in temperature. These findings confirmed the high effectiveness of the pre-oxidation process with a high temperature and short dwelling time compared to that with a low temperature and long dwelling time.

As indicated by XRD, the oxide scales formed at 1000 and 1100 °C consisted exclusively of α-Al₂O₃. Since the alumina-forming alloys generally developed multi-layered oxide scales with different crystalline structures, it was not possible to identify the phase composition of each oxide layer with XRD. The work performed by the Swedish High-Temperature Corrosion Centre employed CBED to determine the phase composition of the top and bottom alumina scales formed on Kanthal AF exposed at 900 °C for 1 h [30,34,42]. In their studies, the CBED patterns obtained from both the inner and outer alumina scales indicated that the double-layered alumina formed on Kanthal AF consisted of a bottom layer of α-Al₂O₃ and an upper layer of γ-Al₂O₃ after 1 h of exposure at 900 °C in dry oxygen. CBED analysis was also conducted on the double-layered oxide scales formed on the alloy 318 pre-oxidised at 900 and 1100 °C in order to identify the phase composition of both inner and outer alumina scales. However, in this study, the CBED technique failed to identify the phase composition of both inner and outer alumina on the alloy 318 pre-oxidised at 900 and 1100 °C due to the extremely thin oxide scale and small oxide grains.

Most researchers in this field agree that the outer alumina scale grows by Al outward diffusion, while the inner α-Al₂O₃ grows mainly by oxygen inward diffusion [43,44,45]. The Cr-rich oxide band is suggested to have been produced during the initial stage of oxidation to provide corrosion protection prior to the formation of an alumina scale, indicating the original metal/gas interface [34]. In addition, several researchers also reported the coexistence of γ-Al₂O₃ and α-Al₂O₃ on FeCrAl alloy oxidised at low temperatures (650 to 700 °C) [45,46]. Therefore, both γ-Al₂O₃ and α-Al₂O₃ were expected to form simultaneously on the surface of the alloy 318 pre-oxidised at all four temperatures in the dry air environment in this study. It has been widely reported that the initially formed corundum-type oxides (α-Cr₂O₃ and α-Fe₂O₃), which are isostructural with α-Al₂O₃, act as crystallographic templates for the nucleation and growth of α-Al₂O₃, facilitating the easy formation of α-Al₂O₃ in the beginning of oxidation [31,47,48]. The location of the Cr-rich band suggests that the inner α-Al₂O₃ scale nucleated near the original alloy surface and grew primarily inwards [34]. For the alloy 318 pre-oxidised at 1100 °C, the presence of the α-Al₂O₃ phase confirmed by XRD must indicate the inner scale detected below the Cr-rich band.

Both Liu et al. [30] and Gotlind et al. [34] found that the thickness of the outer alumina layer (80 nm) was much higher than that of the inner layer (20 nm). In contrast, the alloy 318 pre-oxidised at 1100 °C for 1 h developed a thicker inner alumina scale than the outer alumina. This means that the outer alumina scale did not have significant growth during pre-oxidation at 1100 °C. Thus, the oxidation of the alloy 318 pre-oxidised at 1100 °C was mainly driven by oxygen inward diffusion for the growth of the inner alumina layer. Liu et al. [30] also found that the outer γ-Al₂O₃ on Kanthal AF transformed to α-Al₂O₃ with the exposure time extended from 1 h to 24 h at 900 °C in dry air. Rybicki and Smialek [49] reported that the rate of the phase transformation from metastable alumina to α-Al₂O₃ was faster at higher temperatures. Thus, the phase composition of the outer oxide layer on the alloy 318 formed at 1100 °C was assumed to be composed of two different cases: (1) a mixture of metastable alumina and α-Al₂O₃ and (2) only α-Al₂O₃, formed by phase transformation from the metastable alumina at 1100 °C [30,42]. As shown in Figure 9d, both Cr and Fe showed higher concentrations in the outer layer than in the inner layer. The concentration of Fe was found to be higher than that of Cr in the outer layer. Engkvist et al. [42] stated that the higher mobility of Fe in the outer Al₂O₃ was attributed to it being in the divalent state (Fe²⁺). Furthermore, the growth of the outer alumina scale provided an outwards transport path for divalent cations, causing a higher concentration of Fe on the outer alumina scale [34].

For the alloy 318 pre-oxidised at 1000 °C, the surface SEM images revealed a different alumina phase composition, which was confirmed by XRD analysis. It should be noted that the platelet-like alumina (Figure 7g,h,i) had a similar morphology to the γ-Al₂O₃ discovered by Kadiri et al. [27]. Therefore, we can infer that the scale growth on the alloy 318 pre-oxidised at 1000 °C mainly occurred in the outer alumina layer by cation outward diffusion. In addition, the pre-oxidation temperature of 1000 °C was not high enough to support the complete phase transformation to the stable α-Al₂O₃ [50], thereby leaving the platelet-like metastable alumina on the surface. Moreover, the amount of platelet-like alumina on the alloy 318 pre-oxidised at 1000 °C remarkably increased as the dwelling time increased from 1 to 4 h. The observed inhomogeneity of the alumina scale formed at 1000 °C, characterised by regions of platelet-like metastable alumina alongside thinner areas, raises concerns about its long-term stability, particularly under the fluctuating temperature conditions inherent in SOFC operations. Such multiphase or structurally non-uniform scales can experience significant internal stresses due to differential thermal expansion coefficients between alumina polymorphs or between the oxide and the alloy. These stresses, exacerbated by thermal cycling, can promote micro-cracking, delamination, or localised spallation, potentially leading to non-uniform degradation and premature loss of Cr retention capability. This further underscores the advantage of the dense, uniform, and predominantly α-Al₂O₃ scale formed at 1100 °C, which is expected to offer superior thermomechanical stability and thus better long-term protection. However, for the samples pre-oxidised at 1100 °C, the very small amount of platelet-like alumina formed during the temperature ramping stage had dramatically decreased as the pre-oxidation dwelling time increased from 30 min to 1 h. In addition, with the pre-oxidation time increasing to 2 h, the surface had been completely covered with a relatively dense and even α-Al₂O₃ layer, with no visible platelet-like alumina. This can be explained by the fact that the 1100 °C used for pre-oxidation promoted faster phase transformation of metastable alumina formed during the temperature ramping stage to the stable α-Al₂O₃ phase compared to the other lower temperatures used [49]. It can be inferred that there must be a transition temperature between 1000 and 1100 °C in which only α-Al₂O₃ is formed [51]. Thus, the samples pre-oxidised at 1100 °C for 30 min and 1 h are considered to constitute promising pre-oxidation conditions for industrial application in a cost-effective process.

4.2. Exposure Tests

The relatively low mass gain and Cr₂O₃ evaporation of the pre-oxidised samples compared to the unoxidised ones was due to the alumina scales established on the alloy surface during pre-oxidation effectively slowing down the outward diffusion of aluminium and chromium in the exposure tests. In addition, the corrosion behaviour and Cr retention ability of the pre-heated samples under the simulated SOFC atmosphere were found to strongly depend on the pre-oxidation time and temperature. This can be explained by the influence of different types of alumina phases formed under different pre-oxidation conditions. However, the higher Cr₂O₃ evaporation of the samples pre-oxidised at 800 °C for 1 h compared to the unoxidised samples was due to the fact that the Cr₂O₃ formed in the initial stage of pre-oxidation could not be completely covered by the simultaneously formed alumina scale under low pre-oxidation temperature and short dwelling time conditions. The average thickness of the alumina scale developed on the alloy 318 pre-oxidised at 900 °C for 1 h was about 10 nm. A drop of the temperature by 100 °C will reduce the scale thickness of the alloy 318 pre-oxidised at 800 °C since a reduction of temperature dramatically reduces the diffusion rates [51]. Therefore, a large amount of Cr₂O₃ evaporation from the already built Cr₂O₃ scale was observed for the samples pre-oxidised at 800 °C for 1 h. However, with the dwelling time being extended to 4 h at 800 °C, a decreased Cr₂O₃ evaporation was detected due to the increased surface coverage by the alumina scale. The higher mass gain for the samples pre-oxidised at 800 and 900 °C corresponded to a faster Al outward diffusion rate for the development of an alumina scale on the outer surface, in comparison with the samples pre-oxidised at 1000 and 1100 °C. It should be noted that a fast Al consumption rate will cause the exhaustion of the Al reservoir in the alloy base, which will eventually result in material failure at an accelerated rate [37]. In this research, one of the main objectives to carry out the pre-oxidation for the alloy 318 was to decrease the Al oxidation rate and make the Al reservoir last longer in the CAPH application. The gross mass gain of the sample pre-oxidised at 1100 °C for 1 h can be calculated by adding the mass gain from the pre-oxidation and the mass of evaporated Cr₂O₃ to the mass gain obtained from the 500 h exposure test. It was found that the calculated gross mass gain of the sample pre-oxidised at 1100 °C for 1 h was 0.094 mg/cm², which is much smaller compared to the gross mass gain of 0.133 mg/cm² for unoxidised samples after the 500 h exposure test in the simulated SOFC cathode environment. This indicates that the total Al diffused from the alloy base to the scale/gas interface for alumina formation was less for the sample pre-oxidised at 1100 °C for 1 h (0.094 mg/cm²) compared to the unoxidised sample (0.133 mg/cm²) because of the extremely low diffusivity of Al through the α-Al₂O₃ built on the sample pre-oxidised at 1100 °C for 1 h. Therefore, the sample pre-oxidised at 1100 °C for 1 h with the lowest oxidation rate amongst all the tested pre-oxidised samples could potentially extend material life much longer for the CAPH application. To investigate whether this low mass gain could be associated with unobserved scale loss, post-exposure surface SEM analysis was conducted on the sample pre-oxidised at 1100 °C for 1 h. This examination confirmed that the alumina scale formed on these samples remained intact and well adhered, with no signs of spallation or delamination. The lowest mass gain and Cr₂O₃ evaporation of the sample pre-oxidised at 1100 °C for 1 h were due to the double-layered alumina scale with a thick inner α-Al₂O₃, which had a more compact lattice structure, in comparison with the metastable alumina phase developed on the samples pre-oxidised at lower temperatures [52]. Metastable alumina types are cation-deficient and are, therefore, considered to possess a lower Cr retention capability than α-Al₂O₃ [22]. Furthermore, the extremely low amount of Cr₂O₃ evaporation from the sample pre-oxidised at 1100 °C was considered to be from the Cr outward diffusion from the Cr-rich band between the upper and lower oxide scale [34], instead of from the alloy base. The observation that no other cations were detected in the inner α-Al₂O₃ is suggested to reflect the excellent barrier effect of α-Al₂O₃ in Cr₂O₃ retention.

The measured oxidation rates of the samples pre-oxidised at different temperatures were consistent with the morphology change detected by surface SEM analysis. As shown in Figure 11, the expansion of the size and area of the long-ridged alumina before and after the 500 h of exposure was detected as follows: 800 °C_1 h > 900 °C_1 h > 1000 °C_1 h. However, it is worth noting that no significant change of surface morphology could be detected for the sample pre-oxidised at 1100 °C for 1 h even after the 500 h exposure test. It has been confirmed by several studies that γ-Al₂O₃ does not transform to α-Al₂O₃ in humid environments during high-temperature exposure since water vapour stabilises γ-Al₂O₃ by hydroxylation of the surface [30,34,42,43]. The metastable alumina is known to be prevalent at low temperature, due to its easier nucleation than α-Al₂O₃ [43]. Therefore, in the current research, the growth of the alumina scale on the pre-oxidised samples under these exposure conditions (850 °C, 6.0 L/min air, and 3% water) was dominated by the Al outward diffusion for the outward growth of metastable alumina, rather than the oxygen inward diffusion for the growth of an inner α-Al₂O₃ layer. As mentioned before, the difference in the alumina growth rate for the samples pre-oxidised at the four temperatures during the exposure tests was caused by the difference in the Al outward diffusion rate throughout the alumina scale developed during the pre-oxidation. The alumina scale established at low pre-oxidation temperatures was thin and dominated by alumina in the metastable phase. However, the pre-oxidation for alloy 318 conducted at high temperatures facilitated the growth of a thicker, compact α-Al₂O₃ scale. Therefore, the faster Al outward diffusion rate via the thin metastable alumina scale developed at 800 °C for 1 h led to the formation of the largest amount of long-ridged alumina on the surface. On the contrary, the α-Al₂O₃ formed a dense and continuous layer on the samples pre-oxidised at 1100 °C for 1 h, restricting the supply of cations for the growth of the outer γ-Al₂O₃ layer. This resulted in a much slower oxidation rate in the SOFC cathode environment.

5. Conclusions

In this study, we investigated the effect of pre-oxidation conditions on the oxidation resistance and Cr evaporation behaviour of alloy 318 for SOFC CAPH applications. The main conclusions derived from this research were as follows:

• A pre-oxidation treatment dramatically decreased the subsequent oxidation rate and Cr₂O₃ evaporation of alloy 318 when exposed to humidified air at 850 °C.
• The kinetics of oxidation and Cr(VI) release for the pre-oxidised alloy 318 were governed by the phase composition of the alumina scale formed during the pre-oxidation step, which was dependent on temperature and time.
• Pre-oxidation at 800 °C and 900 °C yielded less protective scales, likely containing metastable alumina phases, which permitted relatively faster Al and Cr outward diffusion compared to higher-temperature treatments.
• Optimal corrosion resistance and Cr retention were achieved with pre-oxidation at 1100 °C for 1 h, resulting in a 98% reduction in the oxidation rate and a 90% reduction in Cr₂O₃ evaporation compared to the unoxidised alloy under the tested exposure conditions.
• The superior performance of the 1100 °C/1 h treatment was attributed to the formation of a compact and homogenous α-Al₂O₃ scale that effectively prevented outward diffusion of Al and Cr.
• The formation of this stable α-Al₂O₃ scale via pre-oxidation is expected to mitigate key degradation issues observed in alloy 318 CAPHs during operation, specifically by slowing down Al oxidation in hot zones and preventing Cr₂O₃ formation in cold zones.

These findings indicate that controlled pre-oxidation at 1100 °C for 1 h is a highly effective and potentially industrially scalable strategy for enhancing the durability and extending the lifespan of alloy 318 CAPHs in SOFC systems. Further investigations could focus on evaluating the long-term thermomechanical stability of the α-Al₂O₃ nanofilm under realistic thermal cycling conditions and validating its performance in prototype CAPH components.