II. EXPERIMENTAL

The specimens used in the oxidation experiments were 8-mm-diameter disks of 1 mm thickness cut from a polycrystalline Fe₃Al binary alloy ingot (Fe–26 at. % Al). The as-cast samples had grain sizes of a few hundred μm. Prior to oxidation, the samples were mechanically ground and polished to a 1 μm finish and thoroughly cleaned in ethanol. All oxidation experiments were performed at 700 °C in atmospheric air at ambient pressure.

The chemical composition in the oxide scales grown on Fe–26 at. % Al after 300 min was characterized by applying X-ray photoelectron spectroscopy (XPS). XPS spectra and sputter depth profiles were taken using a monochromatic Al K α X-ray source, step size of 0.8 eV, pass energy of 93.90 eV, 20 sweeps, and a spot size of 100 μm. The spectra were recorded within the binding energy range of 0 to 1200 eV, sputter depth profiles of the Fe 2p, Al 2p, and O 1s photolines were measured, and sputtering was performed with 2 keV ^Ar+ ions in steps of approximately 10 nm (calibrated according to a _SiO2 standard). The XPS data were analyzed using the CASA-XPS software.

In order to assess the conditions for the appearance of metastable Al₂O₃ polymorphs, ex situ phase analyses applying grazing-incidence X-ray diffraction (XRD) at an incidence angle (ω) of 1° were carried out using a laboratory Co K α source for samples produced after different oxidation times, ranging from 2 to 300 min at 700 °C. Under these conditions, the penetration depth is estimated to be approximately 400 nm.

In situ oxidation experiments for stress analyses were carried out at the beamline for energy-dispersive diffraction (EDDI) of the Helmholtz-Zentrum Berlin, situated at the Berlin synchrotron radiation facility Berliner Elektronenspeicherring-Gesellschaft für Synchrotronstrahlung. The samples were oxidized in atmospheric air using an air-cooled resistance furnace. The time required to reach the desired oxidation temperature of 700 °C was approximately 10 min. The ED diffraction technique was applied (Genzel et al., Reference Genzel, Denks, Gibmeier, Klaus and Wagener2007) within an energy range of 15 to 85 keV and the internal stresses in the growing oxide scale were determined using the sin²ψ method. In contrast to time-consuming angle-dispersive diffraction with monochromatic radiation, ED diffraction uses a polychromatic X-ray beam, thus allowing for the acquisition of complete ED diffractograms at a fixed 2θ angle. This enables the determination of stress values for several diffraction lines simultaneously (Juricic et al., Reference Juricic, Pinto, Cardinali, Klaus, Genzel and Pyzalla2010).

In the present work the 2θ angle was set to 5°. The acquisition time for an individual ED diffractogram amounted to 5 min. The sin²ψ measurements were performed using seven ψ tilts which resulted in a total measurement time, including sample movements, of 36 min/stress analysis. An ex situ texture analysis was also performed at the EDDI beamline on an Fe–26 at. % Al sample previously oxidized for 300 min at 700 °C. Pole-figure data were recorded for a ψ range of 0° to 75° in 5° steps and for an azimuthal φ range of 0° to 345° in 15° steps. Corrections of the pole-figure intensities were performed for absorption and geometrical aberrations (Welzel and Leoni, Reference Welzel and Leoni2002) by measuring a randomly oriented W powder sample deposited on the oxidized sample. Complete pole figures and inverse pole figures were computed using the BEARTEX software, version 3.4 (Wenk et al., Reference Wenk, Mathhies, Donovan and Chateigner1998).

Figure 1. XPS sputter depth profile in the oxide scale grown on Fe–26 at. % Al substrate at 700 °C after 5 h oxidation time

III. RESULTS AND DISCUSSION

The chemical composition gradient in the scale grown on Fe–26 at. % Al after 5 h oxidation at 700 °C is presented in Fig. 1 (sputter depth values increase in the direction of the substrate). Scale thickness ranges between 80 and 100 nm and the scale appears to be formed exclusively by Al oxides except for a small concentration (lower than 10%) of Fe oxides which is restricted to the first 30 nm at the oxide outer surface. It is worth noticing that the concentration of Al⁺³ in the oxide scale, approximately 35%, is well above the nominal Al content in the base material (26%). This is an indication that oxide-scale growth occurs by outward diffusion of Al⁺³ cations as well as by the inward diffusion of O⁻², as it has been previously reported for the oxidation of other binary iron aluminides (Pöter et al., Reference Pöter, Stein, Wirth and Spiegel2005).

Phase analyses by grazing-incidence XRD were performed on samples oxidized for 2, 5, 10, 30, and 300 min at 700 °C. The diffraction data obtained for oxidation times of 5, 10, and 30 min are shown in Figs. 2a – 2c, respectively. The clear (220) diffraction line of the substrate, which appears in some of the samples, results from the coarse-grained substrates, which were prepared from different cast ingots. In spite of the low oxidation temperature applied, the XRD data show that even at the onset of the oxidation process α-Al₂O₃ is the predominant oxide phase in the scale, with only minor quantities of α-Fe₂O₃ and metastable θ-Al₂O₃. θ-Al₂O₃ appears to vanish already after 30 min of oxidation even though the development of the stable α-Al₂O₃ is usually associated with oxidation temperatures of over 1000 °C (Levin and Brandon, Reference Levin and Brandon1998; Pöter et al., Reference Pöter, Stein, Wirth and Spiegel2005). However, it has been noted that other oxides with trigonal crystal lattices, such as α-Fe₂O₃ and α _-Cr2O₃, may form at earlier stages and act as templates for the growth of α-Al₂O₃, thus favoring the earlier development of stable corundum at lower temperatures (Renusch et al., Reference Renusch, Grimsditch, Koshelev, Veal and Hou1997; Asteman and Spiegel, Reference Asteman and Spiegel2008). The results suggest, therefore, that the formation of α-Al₂O₃ occurs not only by transformation from θ-Al₂O₃ but also spontaneously as a consequence of the presence of α-Fe₂O₃ in the oxide scale.

Experimental pole figures were determined for the (104), (110), (113), and (116) diffraction lines of the α-Al₂O₃ developed after 5 h oxidation at 700 °C. These data were used for calculating the orientation distribution function (ODF) and inverse pole figures. Owing to the lower volume fraction of α-Fe₂O₃ in the oxide scale, the only pole figure that could be measured for this phase was the one corresponding to the

Figure 2. Grazing-incidence XRD patterns showing the oxide phases developed on Fe–26 at. % Al after (a) 5, (b) 10, and (c) 30 min of oxidation.

(104) reflection, which represents the strongest diffraction line. For this reason, it was not possible to determine the ODF or inverse pole figures for α-Fe₂O₃.

The experimental pole figure measured for the (104) diffraction line of α-Al₂O₃ is presented in Fig. 3a. The corresponding recalculated pole figure obtained from the ODF is shown in Fig. 3b. Figure 3c displays the experimental pole figure measured for the (104) diffraction line of α-Fe₂O₃. The texture strength is given in terms of multiples of a random distribution (mrd). The α-Fe₂O₃ pole figure is similar to the one of α-Al₂O₃ in the sense that in both cases the maximum texture strength occurs between polar distances of 20° to 40°. This suggests that the texture formation of both oxides is equivalent and could be a consequence of α-Fe₂O₃ acting as a crystallographic template for the formation of α-Al₂O₃. This observation is in agreement with studies of the epitaxy relation between α-Fe₂O₃ films grown on

Figure 3. (a) Experimental (104) pole figure of α-Al₂O₃, (b) recalculated (104) pole figure of α _-Al3O₃, and (c) experimental (104) pole figure of α-Fe₂O₃ after 5 h oxidation of a Fe–26 at. % Al polycrystal at 700 °C. Texture strength is given in terms of mrd.

α-Al₂O₃, which show that the (0001) surfaces of both phases remain parallel to each other (Wang et al., Reference Wang, Thevuthasan, Gao, McCready and Chambers2002; Lee et al., Reference Lee, Kim, Yu, Chang, Joo, Lee, Hur and Kim2005). Furthermore, Sun et al. (Reference Sun, Stirner and Matthews2006) suggested that the (0001) plane of another trigonal structured oxide, α _-Cr2O₃, provides favorable conditions for the template growth of α-Al₂O₃. Eklund et al. (Reference Eklund, Sridharan, Sillassen and Bøttiger2008), while observing that this template effect is stronger for α-Al₂O₃ developing on the (10-14) of α _-Cr2O₃, also reported the growth of textured (0001) α-Al₂O₃ on α _-Cr2O₃ (0001).

The inverse pole figure of α-Al₂O₃ in the normal direction (ND) of the scale surface is plotted in Fig. 4. It can be noticed that the (0001) basal planes of the corundum structure grow with a certain preferential orientation parallel to the surface of the sample. These results are in agreement with a previous electron backscattered diffraction study which also revealed a (0001) fiber texture in the ND of thermally grown α-Al₂O₃ scales (Karadge et al., Reference Karadge, Zhao, Preuss and Xiao2006). The texture strength of the oxide scale is, however, weak, thus allowing, e.g., for unrestricted sin²ψ measurements. This

Figure 4. ND inverse pole figure of the α-Al₂O₃ grown on Fe–26 at. % Al after 5 h oxidation at 700 °C. Texture strength is given in terms of mrd.

agrees with the predictions of Blachère et al. (Reference Blachère, Schumann, Meier and Pettit2003) that the ion mobility inside an oxide is low for oxidation temperatures below T _m/2 (T _m is the melting point of the oxide, here 2054 °C for α-Al₂O₃), causing a competitive oxide grain growth in the direction of the ion flux, i.e., parallel to the direction of the surface normal. This produces oxide scales which consist of stacks of α-Al₂O₃ grains without a strong crystallographic preferred orientation.

Figure 5 shows the evolution of the internal stresses in the oxide scale growing on Fe–26 at. % Al. The stresses were determined for the (012) and (104) diffraction lines. From the first stress analysis performed, at 45 min, to approximately 200 min of oxidation, the scale undergoes a continuous increase in the level of compressive stresses until a maximum average value of −630±200 MPa [−720±70 MPa, as determined for the (012) reflection, and −550±200 MPa, as determined for the (104) reflection] is reached. The occurrence of compressive growth stresses during oxidation is attributed to the lateral growth mechanism of the oxide scale (Clarke, Reference Clarke2003; Rhines and Wolf, Reference Rhines and Wolf1970). According to this model, Al⁺³ and O⁻² diffusing in opposite directions combine to form new oxide grains along grain boundaries of the scale, which are oriented vertically to the surface. Since this new oxide is laterally constrained by the surrounding oxide grains as well as by the underlying substrate, high compressive stresses develop. After a maximum compressive stress level is reached in the scale, creep-induced stress relief begins to balance the intrinsic oxide growth stresses (Veal and

Figure 5. Evolution of growth stresses, determined for the (012) and (104) diffraction lines of α-Al₂O₃, with increasing oxidation times at 700 °C.

Paulikas, Reference Veal and Paulikas2008; Panicaud et al., Reference Panicaud, Grosseau-Poussard and Dinhut2006; Limarga et al., Reference Limarga, Wilkinson and Weatherly2004), and the stress level in the oxide scale approaches a steady state condition which, in the present case, is close to −300 MPa.

The strain evolution during the first 45 min of oxidation could not be characterized because of the long measurement times, which were necessary given the reduced thickness of these oxide scales. Thus, the impact of the appearance of θ-Al₂O₃ on the growth stress level could not be directly verified. However, the results obtained here appear to be consistent with previous studies (Veal and Paulikas, Reference Veal and Paulikas2008; Hou et al., Reference Hou, Paulikas, Veal and Smialek2007) in which it was observed that the first formed α-Al₂O₃ grains are under tensile stresses due to the development of metastable θ-Al₂O₃ and subsequent transformation of θ-Al₂O₃ to α-Al₂O₃. The extrapolation of the initial stress values (represented by the dotted line in Fig. 5) suggests the possibility of tensile stress formation during the first minutes of oxidation since it is rather unlikely that the oxide scale grows initially nearly stress-free, as the first determined stress values, ranging between −10±100 and −70±100 MPa, show. As displayed in Fig. 1c, the θ-α transformation is completed before 30 min of oxidation time. After the metastable θ-Al₂O₃ is consumed, the compressive stress generation mechanism via new oxide growth inside the scale becomes predominant and the internal stresses change from tensile to compressive. The absence of tensile stresses in the scale during the later stages of oxidation further corroborates that metastable Al₂O₃ formation on the binary Fe–26 at. % Al alloy is restricted to the beginning of the oxidation process.