Introduction

Heat-resistant alloys used in steam cracker furnaces are designed for creep resistance and high-temperature corrosion resistance. Creep resistance is traditionally achieved by stabilizing an austenitic matrix (with 35 wt%–45 wt% of Ni) together with primary and secondary Cr-rich, Nb-rich and Ti–rich carbides (with 0.4 wt%–0.5 wt% of C). In the 1970′s, HP grades containing 25 wt% of Cr and 35 wt% of Ni were traditionally used for cracking applications [1, 2]. The first evolution of this family has been Nb additions to improve creep resistance [3, 4], followed by Ti and few other micro-alloying strategies [4]. In order to increase the corrosion resistance, alloys with Cr contents up to 30 wt%–35 wt% were developed. The corrosion resistance of such alloys relies on the surface formation of a protective chromia layer. Nevertheless, oxide may volatilize and spall especially at temperature above 800 °C in air–steam atmospheres typically used during decoking steps [5, 6]. Alumina-forming alloys are known to be stable at higher temperatures than chromia-forming alloys, especially under water vapor conditions [5,6,7]. A first patent has been published in 1981 for an alloy based on HP grade [8], but the development was stopped due to creep resistance problems [9]. Better understanding of creep mechanisms in Al-containing austenitic stainless alloys up to 1000 °C–1050 °C [10] has contributed to develop a new generation of alloys for steam cracking applications. It has been shown indeed that 3 wt%–5 wt% of aluminum could be successfully added leading to the formation of an Al2O3 subscale. However, some impurities, like sulfur, are known to significantly affect alumina formation [5,6,7].

Besides, sulfur is known to have a substantial impact on the spalling resistance of alumina-forming alloys, especially during cyclic oxidation. Only a few ppm of sulfur increase mass losses during oxidation [11]. It has been shown that sulfur segregates at matrix / oxide interfaces at high temperatures [12,13,14,15,16,17,18,19,20,21,22,23], making such interface weaker [24, 25]. This effect has been attributed to a chemical modification of the interface. This sulfur segregation also promotes the growth of porosities at these interfaces [26].

A detrimental synergy between sulfur and oxidation under water vapor is often reported for the spalling resistance of alumina layers [14, 27,28,29,30,31,32]. It has been hypothesized that it could be due to segregated elements [14], such as hydrogen that would segregate at matrix / oxide interfaces like sulfur, further decreasing the oxide layer resilience [6].

More generally, sulfur influence on oxidation kinetics remains a controversial subject. Some authors have reported a significant increase of the oxide layer thickness and of its initial growth rate [33, 34], while others have shown only a decrease of the spalling resistance [5,6,7, 35].

Under water vapor conditions, the growth rate of metastable phases, like transition aluminas (θ and γ) or spinels, is higher in the transient stage [5, 6, 28, 32]. This phenomenon could have an impact on the α-Al2O3 growth as shown on FeCrAl [36, 37], on Fe-Al [5], and NiCrAl [27, 30, 31, 38] alloys. However, oxidation kinetics under water vapor are strongly dependent on the alloy composition, the atmosphere or the temperature and the mechanisms are not fully clarified yet.

In order to explain the combined effect of sulfur and water vapor influence on oxidation kinetics in alloys used in steam cracking applications, the oxidation behavior of an alumina-forming industrial alloy was studied both in air and in steam at 900 °C. Alloys were specially cast with sulfur contents ranging from 1 to 82 ppm in wt%. The oxidized surfaces were characterized down to the nanoscale using Raman spectroscopy, NanoSIMS, SEM and TEM. A special emphasis was given on alumina nanoscaled layers, to identify crystalline defects and chemical modifications.

Material and Methods

Dedicated alumina-forming austenitic alloys were centrifugally cast especially for this study. Their composition is based on the composition of Manaurite XAl4® shown in Table 1, a commercial alloy of Manoir Industries for steam cracker applications. Six different sulfur contents have been considered: 1, 12, 24, 38, 48 and 82 wt ppm, measured by a nondispersive infrared spectrometer (ND-IR in a LECO analyzer). Tested alloys are named Al-X, with X = wt ppm of S.

Table 1 Nominal compositions of the Manaurite XAl4 ® (in wt%)
Full size table

Figure 1 shows the typical as cast microstructures formed in alloys Al-1 and Al-82. Large grains of austenitic matrix (grey contrast) are dendrites grown during solidification. They are surrounded by two kinds of carbides: Cr-rich M7C3 (darkly imaged) and (Nb, Ti)-rich MC (brightly imaged). More information on microstructures and phase identification in such alloys can be found in [10, 39, 40].

Fig. 1
figure 1

Typical as cast microstructures of Manaurite XAl4 (Al-1 and Al-82 samples, SEM images)

Full size image

10 X 15 X 2 mm3 coupons were machined out of the 6 alloys in order to perform oxidation tests. Prior to the oxidation tests, surfaces were prepared by grit polishing using SiC paper down to 800 grit, and samples were cleaned in an acetone ultrasonic bath. Mass measurements were carried out before and after oxidation with three samples of each alloy using a balance with a reliability of 0.1 mg.

Oxidation tests were performed for 24 h at 900 °C in a tubular furnace in laboratory air and in a mixture of 2/5 Ar- 3/5 H2O by volume including some N2 impurities. Samples were introduced in a cold furnace and heated up at a rate of 900 °C/h followed by the isothermal oxidation treatment. Samples were cooled down slowly (180 °C/h) in the furnace to limit spalling.

In order to identify crystal structures of the main phases formed during oxidation tests, Raman spectroscopy was performed with a Horiba Jobin–Yvon LabRam Aramis Raman spectrometer, equipped with two lasers of wavelength 632.8 nm and 532 nm. The laser beam was focused on the sample using a 100X objective giving a spot size smaller than 1 μm.

Prior to cross-sectional SEM investigations, samples were first coated with a thin gold layer (100 nm–200 nm) in a vacuum metallizer and then a thicker electrolytic copper layer (about 20 µm) was deposited for an optimal protection of oxidized surfaces. These samples were then embedded in a conductive resin and mechanically polished with ¼ µm diamond paste for the final step.

SEM observations were carried out with a ZEISS LEO 1530 XB operating at 15 kV, with images recorded using a secondary electron detector.

Transmission electron microscopy (TEM) samples were prepared using a FEI Helios dual-beam SEM-focused ion beam, using Xe ions. Samples were protected through the application of an in situ deposited platinum layer; then trenches were dug and a lamella was extracted using a micromanipulator. Samples were then thinned to electron transparency at 30 kV and 12 kV.

TEM observations were carried out with a probe corrected JEOL ARM-200F operated at 200 kV. Elemental mapping was carried out using energy-dispersive X-ray spectroscopy (EDX) with an Oxford Instruments X-max detector (collection angle 0.7 sr). Grain size quantification was done from scanning transmission electron microscopy (STEM) images on at least 30 different grains on each samples.

High-resolution nanoSIMS analyses were carried out using a CAMECA NanoSIMS 50 with a 16 keV Cs+ primary beam. 52Cr16O−, 27Al16O−, and 32S− ions were simultaneously collected for mapping.

Results

The mass gain (relative to sample surfaces) of samples oxidized in air and steam is plotted as a function of sulfur content in Fig. 2. From 1 to 38 wt ppm of sulfur, mass changes under both atmospheres are extremely small and remain near the detection limit (about 0.1 ± 0.1 mg/cm2). At sulfur contents higher than 40 wt ppm, mass variations are more significant and a strong correlation with sulfur level is exhibited. For the Al-82, the mass gain in air (up to 0.2 ± 0.1 mg/cm2) is indeed twice larger than for Al-1 and it raises up to 0.54 ± 0.1 mg/cm2 in steam. This data indicates that there is a strong combined effect of the sulfur content and of the oxidation atmosphere.

Fig. 2
figure 2

Influence of sulfur on the mass variations of samples oxidized at 900 °C during 24 h under dry air and water vapor + Ar mixture

Full size image

Three representative alloys oxidized in both atmospheres were selected for further analysis: a low sulfur content (Al-1), an intermediate case (Al-48) and a high sulfur content (Al-82).

Figure 3 shows the cross-sectional samples of Al-1, Al-48 and Al-82 after 24 h isothermal oxidation at 900 °C in dry air and in water vapor. On the top surface, the thick electrolytic Cu and thin gold protective layers are clearly exhibited. Below these layers, two kinds of features appear: thin oxide layers (zone A circled in Fig. 3c), and protuberated oxides (zone B circled in Fig. 3f) always combined with internal oxidation (zone C circled in Fig. 3f).

Fig. 3
figure 3

Cross-sectional SEM observations of Al-1 (a, b), Al-48 (c, d) and Al-82 (e, f) oxidized 24 h at 900 °C under dry air (a, c, e) and under water vapor mixture (b, d, f)

Full size image

The identification of phases located in these different zones was performed using Raman spectroscopy on cross-sectional samples. Figure 4 shows representative Raman spectra recorded, respectively, on zones A, B and C. In zone A, two phases could be clearly identified. Raman shifts located at 302 cm−1, 348 cm−1 and 555 cm−1 are attributed to Cr2O3 [41], and those at 401 cm−1, 656 cm−1 and 694 cm−1 to spinel structure [42]. In the spinel structures, the position of the highest wavenumber Raman peak, which is also the most intense, has a great dependence on the nature of the cation in the octahedral sites. In the present case, the strong and broad peak at 694 cm−1 is consistent with the presence of mainly Cr3+ cations in the octahedrally coordinated sites. Unfortunately, others Raman peaks related to this phase are not strong enough to determine the exact stoichiometry of the oxide [42, 43]. The alloys of the present study contain a significant amount of Al. Thus, aluminum oxide is expected to form. However, probably because of a low volume fraction and overlaps with other oxides, Raman modes of alumina phases cannot be unambiguously identified. Some authors have shown, however, that polymorphs of alumina could be identified when they contain Cr3+ impurities, thanks to the photo-induced fluorescence peaks of Cr3+ ions occurring in the Raman shift range [1000–1800] cm−1. Thus, Raman spectra were recorded in zone A using a He–Ne laser (Fig. 4b) and they show indeed two well-defined peaks at 1370 cm−1 and 1400 cm−1 which are characteristic of Cr3+ fluorescence in α-Al2O3 [44,45,46], proving also the presence of this later phase in the oxide scale.

Fig. 4
figure 4

Raman spectra collected in zones A, B and C indicated in Fig. 3. Spectra recorded in the low-wavenumbers range [200 to 1000] cm−1 were obtained using a 532 nm radiation and laser power of ~ 13 mW (a) In the high-wavenumbers range [1000 to 1800] cm−1, spectra were recorded with a 632.8 nm radiation and laser power of 0.2 mW was used (b)

Full size image

Similar oxides were detected in zone B (protuberated oxides), but interestingly the main Raman mode of Cr2O3 (at 551 cm−1) was often much more intense than the main spinel’ peak, suggesting a larger volume fraction of Cr2O3 in this thick external oxide layer as compared to the thinner of zone A (Fig. 4a). Besides, in zone C (internal oxidation area), two different polymorphs of alumina were identified with a pair of fluorescence peaks located at 1179 cm−1 and 1239 cm−1 and another at 1374 cm−1 and 1404 cm−1 that could be attributed, respectively, to θ-Al2O3 [45, 46], and α-Al2O3. AlN is also detected locally with Raman peaks located at 617 cm−1 and 657 cm−1[47], on platelet-like morphologies, shown in Figs. 3d, e and f.

In summary, thin oxide layers (zone A) are composed of a spinel phase, Cr2O3 and α-Al2O3. Thick oxide layers (zone B) contain mainly Cr2O3, with a small amount of spinel and α-Al2O3 too. Last, the internal oxidation zone located under zone B (zone C) contains AlN and two alumina phases, α-Al2O3 and θ-Al2O3.

A more precise localization of the different compounds identified by Raman spectroscopy was established based on SEM image contrasts (Figs. 3). For the lowest sulfur content (Figs. 3a and b), after oxidation in air and steam, the oxide structured layer is typical of zone A: the alloy matrix is covered by a thin layer of α-Al2O3, then Cr2O3 and spinel on the top. At medium sulfur content (Figs. 3c and d), some protuberated oxides with internal oxidation (such as zone B and C) were also formed locally, and this phenomenon is more pronounced after oxidation under vapor conditions. At the highest sulfur content (Figs. 3e and f), the growth of protuberated oxides with underneath internal oxidation is generalized and more especially under water vapor. On protuberated oxides (zone B), non-oxidized matrix grains were also observed in the chromia layer above the alumina layer. Thus, it turns out that the structure, the thickness and the morphology of oxidation affected layers strongly depend both on the sulfur content and the oxidation atmosphere. Both highest sulfur levels and water vapor conditions give rise to more numerous protuberated oxides with underneath internal oxidation zones. These observations are fully consistent with mass variation evolutions displayed in Fig. 2.

A thin α-alumina layer was formed on the surface of samples, in thin oxide layers and in protuberated oxides zones. However, the α-alumina layer formed in zones B is apparently less protective. Indeed, these zones contained a thick spinel + Cr2O3 layer with matrix grains which were not oxidized, and (α + θ)-Al2O3 internal oxidation, with some AlN particles. In order to explain this phenomenon, structures and compositions of alumina layers were analyzed by transmission electron microscopy. Nanoscale secondary ion mass spectrometry was also used for sulfur localization.

A typical image of the structure of the oxide layer in zone A is shown in Fig. 5. On the cross-sectional STEM image recorded with a high angle annular dark field (HADDF) detector (Fig. 5a), the different oxides are revealed with a Z contrast. Additional information about the distribution of alloying elements was collected thanks to EDX mapping (Figs. 5b, c, d, e and f). Phase identification was also carried out using electron diffraction, however, due to the nanoscaled structure of oxide layers and the large number of possible oxide diffraction spot overlaps, it has not been possible to collect reliable and quantitative information, therefore these data are not shown here.

Fig. 5
figure 5

Cross-sectional STEM-HAADF image (a) and EDX elemental maps of Al (b), Cr (c), Fe (d), Ni (e) and O (f) (Kα1 rays) of Al-1 oxidized 24 h at 900 °C under dry air

Full size image

The matrix is located at the bottom, it contains a high Cr, Fe and Ni concentration and a low O level. The Cr-rich zone on the right-hand side is attributed to a chromium-rich M7C3 carbide. Three different layers with a high amount of O are stacked above the matrix. The top layer contains mainly Cr, Fe, Ni, Al and O which is consistent with the spinel phase. Below, a thin layer with a high amount of Cr and O is exhibited, consistent with the Cr2O3 phase. In contact with the austenitic matrix, the last layer is exclusively enriched in Al and O, corresponding to the α-Al2O3 phase.

The mean thickness and the grain size of Al2O3 layers have been estimated from STEM images in regions corresponding to zone A (thinnest oxide layers) of Al-1 and Al-82 alloys oxidized under both atmospheres. In the Al2O3 layer, grains exhibit a columnar structure, thus, two parameters have been chosen, namely the layer thickness and the apparent grain width, as depicted in Figs. 6a and b. The thickness is similar for both samples, with a thickness of about 90 nm whatever the oxidation conditions (Fig. 6c). Meanwhile, the grain width appears relatively similar for both alloys but strongly depends on the oxidation atmosphere. Indeed, after oxidation under dry air, alumina grain width is 90 ± 19 and 84 ± 12 nm for Al-1 and Al-82 alloys, respectively, while after oxidation under water vapor, alumina grain width is 57 ± 13 and 52 ± 4 nm for Al-1 and Al-82 alloys, respectively (Fig. 6d). Thus, the oxidation atmosphere seems to significantly affect the nucleation and growth mechanisms of alumina grains.

Fig. 6
figure 6

STEM-HAADF images of Al-82 oxidized under dry air (a) and under water vapor (b) showing alumina grain size, measured thickness (c) and width (d) of alumina grains in samples Al-1 and Al-82 oxidized 24 h at 900 °C under dry air and under water vapor

Full size image

Since mass gain measurements and SEM observations revealed also a strong combined effect of the sulfur content and of the oxidation atmosphere, further analyses were carried out to localize the sulfur within the complex multilayer oxide structure.

The NanoSIMS technique is known to be particularly sensitive to light elements and is well adapted to the localization of sulfur traces [48, 49]. NanoSIMS elemental maps of 52Cr16O−, 27Al16O− and 32S− ions collected in a cross-sectional sample of the Al-82 alloy oxidized under water vapor are displayed in Fig. 7a. The few micrometers thick internal oxidation zone is clearly exhibited with Al-rich particles, corresponding most probably to α and θ alumina phases. On the top right corner, a 3 μm Cr-rich layer corresponding to the Cr2O3 layer appears. Inside this layer, zones with low signals correspond most probably to non-oxidized metallic particles. Last, the thin continuous Al2O3 layer located between Cr2O3 and the matrix is highlighted by the strong 27Al16O− ion signal. 32S− enrichments are exhibited in the continuous α alumina layer, around metallic particles inside the chromia layer and at interfaces between internally nucleated Al2O3 particles and the austenitic matrix. Figure 7b shows a NanoSIMS elemental map combining 52Cr16O−, 27Al16O− and 32S− signals obtained in the cross section of the Al-82 alloy oxidized under dry air. The thin oxide layer that is exhibited is typical of zone A (Fig. 3). 32S− enrichments are observed in the oxide scale but this layer is too thin to obtain a relevant contrast on elemental maps. So for a better localization of these different signals, a linear profile was computed from this map (Fig. 8) starting in the austenitic matrix at the bottom, up to the spinel layer.

Fig. 7
figure 7

Superposition of NanoSIMS elemental maps of 52Cr16O−, 27Al16O− and 32S− ions collected in Al-82 oxidized for 24 h at 900 °C under water vapor (7.a) and under dry air (7.b)

Full size image
Fig. 8
figure 8

Line profiles of 52Cr16O−, 27Al16O− and 32S− extracted from map (Fig. 7b) through the oxide layer along the direction indicated by the red arrow

Full size image

The signal of 52Cr16O− exhibits two maxima, corresponding to Cr2O3 and spinel phases. A large zone, with a thickness ranging from 0.35 to 0.75 μm, exhibits a strong 27Al16O− signal. According to Fig. 5, the spinel phase located on the top contains some Al and the Cr2O3 layer below is very thin. Thus, this Al-rich zone should contain almost the entire oxide layer. A maximum of the 32S− signal clearly appears between the matrix and the α-alumina layer, however, because of the limited spatial resolution (about 50 nm), it is not possible to conclude about the exact location of sulfur within the complex nanoscale structure of the oxide layers revealed by NanoSIMS.

To collect more detailed information about the exact location of sulfur, some STEM-EDX line profiles were collected on the Al-82 alloy oxidized under water vapor (Fig. 9). The EDX line profile recorded inside the external alumina layer (Fig. 9b), shows as expected, a strong oxygen and aluminum enrichment, but also periodic spatial fluctuations of the chromium concentration with a typical length scale of about 100 nm. The sulfur content is very low (local maximum up to 1 at%) but it is obviously correlated with local Cr enrichment.

Fig. 9
figure 9

Al-82 oxidized under water vapor mixture, (a) cross-sectional STEM image; (b) EDX line profile inside the external alumina layer along the black line indicated on (a)

Full size image

In summary, after oxidation under dry air and under water vapor at 900 °C of Al-82, sulfur is located at all alumina interfaces, in internal and external oxides. Sulfur is also found in the alumina layer, in chromium-rich zones, periodically distributed inside the α-Al2O3 layer. Last some sulfur was also detected around metal particles in protuberated oxides (zone B). It is important to note that similar investigations were carried out in the alumina layer formed on the Al-1 alloy oxidized under both atmospheres. As expected, sulfur was never detected (data not shown here) and chromium concentration fluctuations inside the Al2O3 layer were not found neither.

Discussion

Sulfur presence in alumina-forming alloys is known to have a negative impact on spalling resistance [12,13,14,15,16,17,18,19, 21,22,23, 25]. In this study, no spalling of oxide scale was observed. This is most probably the result of the specific oxidation treatment conducted in the present study (isothermal oxidation and slow cooling of the samples in the furnace).

Mass gain measurements (Fig. 2) demonstrated two very important features for the considered alumina-forming austenitic alloy: (i) higher sulfur contents lead to faster oxidation kinetics; (ii) at sulfur concentrations above 40 ppm, the oxidation kinetics is significantly accelerated under water vapor conditions.

To understand the underlying mechanisms, oxidized surfaces have been investigated to identify the different phases and the related microstructures. Similar oxides have always been detected from the top surface to the core material, namely spinel phases, Cr2O3 and Al2O3. This sequence is fully consistent with the respective formation enthalpies of these three oxides [5, 6]. On a microstructural point of view, large differences have, however, been revealed. SEM data (Fig. 3) revealed that slow oxidation kinetics are always connected with a thin (less than a micrometer in the tested conditions), continuous and regular multilayer oxide structure. While faster oxidation rates (at medium sulfur content under water vapor or for high sulfur content in air or water vapor conditions) are connected to the growth of thick (few microns) protuberated oxides made of spinel and Cr2O3 standing over a thin continuous Al2O3 layer and a relatively thick region (up to 10 µm) exhibiting internal oxidation and nitridation of Al. Thus, it seems that under the tougher conditions (high sulfur and water vapor), the continuous Al2O3 layer is not protective anymore, allowing faster downward diffusion of O2− and upward diffusion of Cr3+.

The continuous nanoscaled Al2O3 layers for all alloys and all conditions exhibit the same α-alumina phase and similar thicknesses (about 90 ± 10 nm, Fig. 6). A significant microstructural difference was found, however, between samples oxidized in air and in water vapor conditions: columnar alumina grains are almost half the thickness under water vapor than under dry air (55 against 90 ± 10 nm, respectively, Fig. 6). Water vapor is known to affect the oxidation mechanisms of alumina-forming alloys leading usually to more transient oxides [5, 6, 27, 28, 32, 36,37,38]. More generally, Gulbransen and Copan [50] demonstrated also in experiments carried out with pure Fe at 450 °C that water vapor facilitates oxide nucleation by promoting surface diffusion. Thus, in the present case, more metastable Al2O3 nucleation sites probably appeared in the early stage, leading to a larger number density of growing alumina grains with thinner lateral dimensions.

Since the mobility of O2− and Al3+ in α-alumina is dominated by grain boundaries diffusion [51,52,53,54], then it could be expected that thinner alumina grains would create a weaker barrier and then give rise to faster oxidation kinetics in long-term exposures. A pre-oxidation treatment under dry air before utilization under steam conditions would probably be beneficial even when the sulfur is below 30 ppm. However, after only 24 h oxidation at 900 °C, our experiments carried out on the Al-1 alloy show that the smaller alumina grains that have nucleated under water vapor (Fig. 6) do not give rise to any other detectable microstructural differences as compared to air oxidation (Fig. 3). This clearly demonstrates that sulfur plays a predominant role and that water vapor simply acts as an accelerating agent in the case of high S alloys.

Thanks to NanoSIMS (Figs. 7, 8) and STEM-EDS (Fig. 9), sulfur was localized in various location of the complex oxide surface microstructure: (i) at interfaces between internally nucleated Al2O3 particles and the austenitic matrix and around metallic particles located inside the chromia layer, both corresponding to grain boundaries where sulfur segregation is well known [55,56,57]; (ii) inside the nanoscaled continuous α alumina layer. In this layer, TEM-EDS data revealed also significant fluctuations of the S concentration spatially correlated to Cr concentration fluctuations. The co-segregation of sulfur and chromium in metallic alloys has already been experimentally observed [33, 58, 59] Chromium-sulfur compounds can be expected, or some authors have hypothesized that it could result from the formation of nanoscaled chromium sulfides at the interface between the matrix and the oxide layer in FeCrAl alloys [33, 60]. Similar mechanisms could occur in the alloy of the present study where chromium sulfides could nucleate inside the alumina layer. Considering relative stabilities of chromium sulfides [34], CrS formation is the most likely possible in our case. Thus, it is fully consistent with our STEM-EDS data (Fig. 9) but EDS measurements are far from the expected concentrations and stoichiometry (only about 10 at% of Cr and 1 at.% of S maximum). In addition to chromium sulfides, other chromium-rich products could also be present: metallic chromium, Cr2O3, nitrides or carbides. Unfortunately, phase identification using electron diffraction did not allow the identification of all phases located in the complex oxide layers.

The inhomogeneous alumina layers formed are less protective and could explain the formation of protuberated oxides and internal aluminum oxidation in the samples Al-48 and Al-82. A schematic illustration of oxide scales and their respective protectiveness formed with high sulfur content under steam conditions and low sulfur content under dry air are shown in Fig. 10.

Fig. 10
figure 10

Schemes of oxide scales formed after oxidation under steam conditions of a high sulfur content alloy and after oxidation under dry air of a low sulfur content alloy

Full size image

Furthermore, oxidation products formed are not the only problem in this application. Indeed metal particles were found in thick chromia layers when sulfur content isn’t negligible (Figs.7a, 9a). They might be the result of a fast initial chromium oxidation along matrix grain boundaries before the formation of the alumina layer. The number density of such metal particles is enhanced under steam conditions and could act as catalyst [61, 62] and promote coke formation during vapocracking applications. Since coke layers gradually cover surfaces and block pipes, decocking operation [63, 64] must be carried out which is deleterious for material integrity.

Conclusions

  • Higher sulfur contents lead to faster oxidation kinetics at 900 °C under dry air. At sulfur concentrations above 40 ppm, the oxidation kinetic is significantly accelerated under water vapor conditions. A thin α alumina layer always covers the alloy but it is protective against oxidation only at lower sulfur concentrations. Then, it is concluded that a sulfur content below about 30 ppm in wt% is necessary to achieve a protective alumina layer.

  • Sulfur was detected along alumina interfaces, and inside the thin and continuous α alumina layer where it is correlated with local chromium enrichments. The exact structure of these Cr- and S-rich features could not be identified (metal particles, oxides, sulfides, carbides) but these inhomogeneities clearly act as short-circuits diffusion through the alumina layer and affect its permeability.

  • The thickness of the thin and continuous α alumina layer is relatively similar under water vapor and air conditions, but alumina grains exhibit significantly smaller lateral dimensions when they grow under water vapor.