Regular Article
Nondestructive Identification of Internal α-alumina Scales on Heat-resistant Ni–Al Alloys Based on Cathodoluminescence Spectra
2023 Volume 63 Issue 11 Pages 1869-1875
Details
2023 Volume 63 Issue 11 Pages 1869-1875
Identifying internal α-alumina (α-Al2O3) scales is critical for evaluating the performance of heat-resistant Ni–Al alloys. This study investigated the possibility of using cathodoluminescence (CL) spectra of the alloy surface for nondestructive identification of internal α-Al2O3 scales underneath multiple other scales. The presence of internal α-Al2O3 scales on a heat-treated Ni-14Al alloy was confirmed by the detection of a CL peak at 695 nm. The scales of this alloy comprised NiO on top followed by NiAl2O4 underneath and α-Al2O3 at the bottom with thicknesses of 1.1, 0.8, and 0.5 µm, respectively. The intensity of the CL peak at 695 nm was close to the minimum detectable analyte signal intensity. These results demonstrate that surface CL spectra can be used for the nondestructive identification of internal α-Al2O3 scales on Ni–Al alloys that are underneath multiple other scales.
At high temperatures, heat-resistant alloys form continuous, adhesive, and slowly growing oxide scales on their surfaces such as α-alumina (α-Al2O3), chromia (Cr2O3), and silica1,2,3) that prevent further oxidation and corrosion, which allows these alloys to withstand extreme temperatures and adverse working conditions. Among such protective oxide scales, α-Al2O3 exhibits the best mechanical resistance and the highest thermal and chemical stability against high-temperature oxidation and corrosion.4,5,6,7,8) Thus, heat-resistant alloys that form α-Al2O3 (α-Al2O3-forming alloys) are used for applications under harsh conditions such as temperatures of 900–1350°C.4,5,6,7,8) α-Al2O3 scales normally form on the outermost surface of β-NiAl, Fe3Al, and Fe–Cr–Al alloys. However, some alloys, such as γ′-Ni3Al and Ni–Cr–Al, form α-Al2O3 scales beneath other oxide layers (e.g., NiO, FeOx, NiAl2O4, and Cr2O3).3,7,9) Such internal α-Al2O3 scales can also enhance the oxidation resistance of heat-resistant alloys.8) Therefore, the identification of internal α-Al2O3 scales is important for evaluating the performance of α-Al2O3-forming alloys.
Conventional analytical methods for identifying internal scales include cross-sectional observation using scanning electron microscopy (SEM) combined with energy- or wavelength-dispersive X-ray spectrometry (EDX or WDX), Auger electron spectroscopy, X-ray photoelectron spectroscopy, and secondary ion mass spectrometry,10,11) but these methods are destructive. To simplify the preparation process and reduce the analysis time, a nondestructive analytical method is desirable. Photostimulated luminescence spectroscopy can be used to identify internal α-Al2O3 scales in a nondestructive manner,12,13,14,15) but it is not a widely used technique.
SEM–cathodoluminescence (CL) analysis is a versatile method that can potentially be applied to the nondestructive identification of internal α-Al2O3 scales. The SEM–CL system is simple to set up: an off-axis parabolic mirror, optical fiber, and portable spectrometer are attached to a conventional SEM instrument.16) The electron gun of the SEM device is used to bombard the surface of a material with electrons, and spectra and projection images of the induced light emissions can be used for analysis of areas ranging from millimeter to micrometer scales.17,18,19,20,21,22) CL spectra of a heat-treated Ni–Cr–Al alloy surface were previously used to identify internal α-Al2O3 scales beneath Cr2O3 scales with a thickness of up to ~3-μm; a CL peak was observed at 695 nm23,24) that originates from the substitution of chromium(III) ions (Cr3+) included as impurities for aluminum(III) ions (Al3+) in α-Al2O3 scales.25,26,27) However, wider application of the SEM–CL technique requires demonstrating that it can identify internal α-Al2O3 scales beneath multiple types of scales.
This study investigated whether the surface CL spectra of heat-resistant alloys can be used to identify internal α-Al2O3 scales underneath multiple types of other scales. Heat-treated Ni–14 mass% Al (Ni–14Al) alloys were selected as samples because they form NiO scales on top followed by NiAl2O4 and then α-Al2O3 scales underneath under high-temperature oxidation conditions.3,28,29) First, NiO or NiAl2O4 films with various thicknesses were deposited via sputtering on sapphire (i.e., α-Al2O3) substrates, and the CL peak intensity at 695 nm was measured to estimate the maximum NiO and NiAl2O4 film thicknesses at which a signal from the underlying sapphire. Then, the dependences of the CL peak intensity at 695 nm on the thicknesses of the NiO and NiAl2O4 films was validated against the surface CL spectra of heat-treated Ni–14Al alloys.
NiO and NiAl2O4 films were prepared to estimate the maximum thicknesses at which a signal from the underlying sapphire substrate could be detected. NiO powder (purity: 99.9%, Wako Pure Chemical Industries, Ltd., Osaka, Japan) was pressed at 140 MPa into a pellet with a dimeter and thickness were 30 mm and 3 mm, respectively. The NiO pellet was then sintered at 1200°C for 10 h, and the sintered pellet was shaped into a disk with a diameter of 25.4 mm. The disk was used as a sputtering target. Meanwhile, NiAl2O4 powder was synthesized by a conventional solid-state reaction based on the literatures.30,31,32) Equal molar concentrations of NiO and Al2O3 (purity: 99.99%, Kojundo Chemical Laboratory Co., Ltd., Saitama, Japan) powders were ball-milled for 24 h using yttria-stabilized zirconia balls (Tosoh) in isopropyl alcohol and were then heated at 1000°C for 10 h in air. The heat-treated powder was then ball-milled for another 10 h and was then heated at 1200°C for 10 h in air. The cycle of ball-milling and the heat treatment at 1200°C was repeated, and the resulting powder contained only NiAl2O4 phase as confirmed by X-ray diffraction (Ultima IV/SG, Rigaku Corporation, Tokyo, Japan). The NiAl2O4 powder was then ball-milled for an additional 100 h. The NiAl2O4 target was prepared according to the same procedure as for the NiO target except that the sintering temperature was set to 1400°C. NiO and NiAl2O4 films were deposited on sapphire substrates using a radio frequency magnetron sputtering gun (SPG-001, Pascal Co., Ltd., Osaka, Japan) and the NiO and NiAl2O4 targets at input powers of 40 and 60 W, respectively, in a vacuum chamber pumped to a base pressure of 3 × 10−5 Pa.24,33,34) During the film deposition, argon gas (purity: 99.9999%) with a flow rate of 20 mL min−1 was introduced into the chamber by a mass flow controller, and the gate value was positioned to maintain a chamber pressure of 1 Pa. The thickness of the films was monitored using a quartz crystal microbalance monitor (STM-2, INFICON Co. Ltd., Bad Ragaz, Switzerland).
Ni–14Al alloy samples were prepared to confirm that surface CL spectra can be used to identify internal α-Al2O3 scales underneath multiple types of other scales. First, 86 mass% Ni powder (purity: 99%, Wako Pure Chemical Industries, Ltd., Osaka, Japan) and 14 mass% Al powder (purity: 99.99%, Kojundo Chemical Laboratory Co., Ltd., Saitama, Japan) were melted in an Al2O3 crucible. The powder mixtures were heated at 1560°C for 30 min before being cooled to room temperature at a rate of 5°C min−1 by passing 96 vol% argon and 4 vol% hydrogen at 200 mL min−1. The Ni–14Al alloy was annealed at 1100°C and 0.1 Pa for 12 h, and the annealed alloy was then cut into cubic slices of approximately 5 mm using a low-speed precision cutter (IsoMet™ LS, Buehler Ltd., Lake Bluff, IL, USA). One exposed surface of the Ni–14Al slice was polished using 600-, 1200-, and 2400-grit abrasive papers, and it was finished using a 1-μm water-free diamond slurry (Aka-Poly WF 1 μm, Akasel A/S, Roskilde, Denmark). The polished slice was heated at 1000°C for 4 and 6 h to create NiO, NiAl2O4 and α-Al2O3 scales. Ni–18 mass% Al (Ni–18Al) alloy samples were also prepared according to the same procedure as for the Ni–14Al alloy samples.
The surface CL spectra of the NiO and NiAl2O4 film–deposited sapphire substrates and heat-treated Ni–14Al and Ni–18Al alloys were acquired using the custom SEM–CL system previously described by the author’s group.16,35,36,37,38,39) Briefly, the CL spectra were acquired using an optical spectrometer (QE65Pro, Ocean Optics Inc., Largo, Florida, USA). An optical fiber with a plano-convex lens at its tip was connected on the opposite end to the optical spectrometer. The tip of the optical fiber with the plano-convex lens was attached to the SEM instrument (Mighty-8DXL, TECHNEX, Tokyo, Japan). The light emitted from the samples was collimated using an aluminum-coated off-axis parabolic mirror with a 0.5 mm hole at the center. The collimated light was collected by the plano-convex lens attached to the tip of the optical fiber, which transmitted it to the optical spectrometer. The electron gun in the SEM was set to an acceleration voltage of 17 kV and was used to bombard a sample area of 0.058 mm2 (0.27 × 0.21 mm) during the acquisition of CL spectra.
Observations and elemental analyses of the cross-section were performed using an SEM instrument (TM3030 Plus, Hitachi High-Technologies Co., Tokyo, Japan) equipped with a silicon drift EDX detector (Quantax70, Bruker Corp., Billerica, Massachusetts, USA). Prior to the observation, one side of the Ni–14Al alloy samples was polished using a 2400-grit abrasive paper. The cross-sections of the NiO and NiAl2O4 films were prepared by cutting the sapphire substrates and deposited films using the low-speed precision cutter.
Heat treatment of Ni–Al alloys with < 17 mass% Al forms NiO scales on top followed by NiAl2O4 scales over α-Al2O3 scales on the surfaces.3,28,29) Thus, to estimate the maximum thickness of NiO and NiAl2O4 scales that would still allow the detection of the internal α-Al2O3 scales, the individual thicknesses of the NiO and NiAl2O4 scales needed to be determined first. The most logical approach would vary the thicknesses of the NiO scales while maintaining constant thickness for the NiAl2O4 and α-Al2O3 scales. However, controlling the individual thicknesses of the NiO, NiAl2O4, and α-Al2O3 scales during heat treatment of the Ni–Al alloys is difficult because they grow independently. Thus, an alternative approach was used where CL spectra were obtained for sapphire (i.e., α-Al2O3) substrates with a thickness of 0.5 mm on which a NiO or NiAl2O4 film with various thicknesses was deposited.
Sputtering the NiO target deposited a smooth film on the sapphire substrate, as shown in Figs. 1(a)–1(c). EDX analysis confirmed that the film compositions corresponded to NiO (Fig. 2(a)). As shown in Figs. 1(d)–1(f), the CL peak at 695 nm was detected for all NiO films, which had thicknesses of 0.1–3.5 μm. Small CL peaks were also observed at 670 and 715 nm, which were attributed to Mn4+ and Fe3+ substituting for octahedrally coordinated Al3+.26,27) Mn and Fe were present in the sapphire substrate as impurities. The intensity of the peak at 695 nm decreased with increasing NiO film thickness.
Figure 3(a) shows the dependence of the normalized CL peak intensity at 695 nm on the NiO film thickness. The normalized intensity was obtained by dividing the peak intensity of the sapphire substrate with a NiO film (I) by the peak intensity of a sapphire substrate without a film (I0) (Fig. 1(g)). Up to a NiO thickness of 1.5 μm, the normalized intensity was in good agreement with the calculated values using a reported absorption coefficient of a NiO thin film at a wavelength of 700 nm (7.5 × 103 cm−1).40) This confirms that a NiO film was deposited on the sapphire substrates. Increasing the NiO film thickness above 1.5 μm caused the normalized intensity to deviate negatively from the calculated values, which is similar to the behavior observed for Cr2O3 films deposited on sapphire substrates.24) This negative deviation was attributed to the decreased number of electrons bombarding the sapphire substrate. The interaction volume between the bombarded electrons and sample is pear shaped,20,41) so the interaction volume in the sapphire substrate is much larger than that in the NiO film with a thickness of less than 1.5 μm, which corresponds to the head of the pear shape. In contrast, the interaction volume in the sapphire substrate gradually decreases when the NiO film thickness increases above 1.5 μm, which indicates that the NiO film reaches the middle region of the pear shape.24)
The maximum NiO thickness for detecting the CL peak at 695 nm was determined based on the results shown in Fig. 3(a) and the background intensity of the CL spectrum in Fig. 1(f). The background intensity was 0.30 near the peaks at 695 nm. The minimum detectable analyte signal intensity (ID) was defined as at least three times the background intensity, which corresponds to a value of 0.90. The average peak intensity of the sapphire substrate without a film (I0) was 25000; thus, the detectable normalized intensity (ID/I0) was 3.6 × 10−5. Extrapolation of the normalized intensity near the detectable normalized intensity in Fig. 3(a) indicated that the maximum NiO thickness that would still allow detection of a signal from the sapphire substrate was 3.9 μm.
The maximum NiAl2O4 thickness for detecting a signal from the sapphire substrate was similarly determined by acquiring the CL spectra with deposited films of various thicknesses. As shown in Fig. 4, smooth films were deposited on the sapphire substrates with a composition corresponding to that of NiAl2O4, which was confirmed by measuring the atomic ratio of Ni:Al in the films using X-ray fluorescence (XRF) spectrometry (ZSX Primus II, Rigaku Corporation, Tokyo, Japan) (Figs. 2(b) and 2(c)). As shown in Fig. 3(b), the dependence of the normalized CL intensity at 695 nm on the NiAl2O4 film thickness was similar to that on the NiO film thickness. The maximum NiAl2O4 thickness that would still allow detection of a signal from the sapphire substrate was estimated to be 2.8 μm based on extrapolation of the normalized intensity near the detectable normalized intensity (ID/I0) of 3.6 × 10−5.
The maximum thicknesses of the NiO and NiAl2O4 films that would allow signal detection of α-Al2O3 were estimated to be 3.9 and 2.8 μm, respectively. The validity of these values was investigated by using a Ni–14Al alloy that forms NiO and NiAl2O4 scales over internal α-Al2O3 scales. Heating the Ni–14Al alloy at 1000°C for 4 h (sample A) and 6 h (sample B) was confirmed to form three-layered oxide scales via SEM observations (Figs. 5(a) and 6(a)) and EDX elemental mappings of oxygen (Figs. 5(b) and 6(b)). Three layers with different gray colors were observed in areas where oxygen was detected: light gray for the outermost layer, gray for the intermediate layer, and dark gray for the innermost layer. Previous studies3,28,29) and the EDX elemental mapping results indicated that the outermost layer was NiO scales because only Ni and O were detected (Figs. 5 and 6). In contrast, the intermediate layer was NiAl2O4 scales because Al, Ni, and O were detected (Figs. 5 and 6). The innermost layer corresponded to α-Al2O3 scales because only Al and O were detected (Figs. 5 and 6). Table 1 summarizes the thicknesses of the three scales for samples A and B.
| Sample name | Holding time at 1000°C (h) | Thickness (μm) | ||
|---|---|---|---|---|
| NiO | NiAl2O4 | α-Al2O3 | ||
| A | 4 | 1.1 | 0.8 | 0.5 |
| B | 6 | 3.1 | 2.9 | 0.5 |
For both the samples, the α-Al2O3 scales has a thickness of 0.5 μm, which is three orders of magnitude less than the thickness of the sapphire substrate in the previous experiment (500 μm). This suggests that the peak intensity of the α-Al2O3 scales on Ni–14Al alloys without the overlying NiO and NiAl2O4 scales (I0) would differ from the peak intensity of the sapphire substrate without the deposited films. To estimate I0 for the α-Al2O3 scales, the CL spectrum was obtained for a Ni–Al alloy that forms α-Al2O3 scale on the surface. Heating the Ni–18Al alloy at 1000°C for 10 h resulted in α-Al2O3 scales with a thickness of 0.6 μm on the surface, and the CL peak intensity at 695 nm was 43700. Because previous studies showed that the CL peak intensity at 695 nm increases proportionally with the thickness of the α-Al2O3 scales,42,43) I0 for α-Al2O3 scales with a thickness of 0.5 μm can be estimated to be 34100. This is greater than the I0 value for the sapphire substrate because the latter has an extremely low chromium content at the scale of parts per million less, whereas α-Al2O3 scales on heat-resistant alloys have a chromium content of ~1 at%.44) Based on the estimated I0 and the results given in Fig. 3 and Table 1, the CL peak intensities at 695 nm for samples A and B were estimated as 463 and 3.23 × 10–4, respectively. The estimated intensity at 695 nm for sample A (I = 463) was above the minimum detectable analyte signal intensity (ID = 0.90). The peak at 695 nm was detected for the CL spectrum of sample A (Fig. 7(a)), but, the peak intensity (I = 4) was two orders of magnitudes less than the estimated intensity (I = 463). This may be because Ni2+ originating from the NiAl2O4 scales and alloys dissolved into the α-Al2O3 scales, as Ni2+ is known to cause luminescence quenching.45) An EDX line scan profile of the cross-section of sample suggests the dissolution of Ni2+ into the α-Al2O3 scales because the intensity of Ni Lα line in the area corresponding to the α-Al2O3 scale was slightly higher than that in the area without any scales, as shown in Fig. 8(a). Thus, the CL intensity at 695 nm for α-Al2O3 scales below NiAl2O4 scales should be two orders of magnitude less than that of α-Al2O3 scales on the surface of the Ni–Al alloy. In contrast, the dissolution of Ni2+, originating from NiAl2O4 film, into the sapphire substrates was not confirmed because the intensity of Ni Lα line in the area corresponding to the sapphire (α-Al2O3) was as high as that in the area without any films (Fig. 8(b)). The thicknesses of the NiO and NiAl2O4 scales in sample A were close to the maximum thickness for detecting a signal from the internal α-Al2O3 scales on the Ni–Al alloy because the peak intensity at 695 nm for sample A was close to the minimum detectable analyte signal intensity. In contrast, no CL peak at 695 nm was detected for sample B (Fig. 7(b)) because the estimated intensity at 695 nm (I = 3.23 × 10–4) was below the minimum detectable analyte signal intensity. Further research is needed on the effects of Ni2+ dissolving into α-Al2O3 scales on the CL intensity at 695 nm to determine the maximum thicknesses of NiO and NiAl2O4 scales that would allow signal detection of internal α-Al2O3 scales on a Ni–Al alloy. Nevertheless, the results show that surface CL spectra can be applied to the nondestructive identification of internal α-Al2O3 scales below other sales on Ni–Al alloys.
The results of this study demonstrate that surface CL spectra can be used to detect internal α-Al2O3 scales underneath other types of scales on heat-resistant Ni–Al alloys. The presence of α-Al2O3 scales was confirmed by detecting the CL peak at 695 nm when the thicknesses of NiO, NiAl2O4, and α-Al2O3 scales on a heat-treated Ni–Al alloy were 1.1, 0.8, and 0.5 μm, respectively. The detection performance was evaluated by comparison to the CL peak intensity at 695 nm for α-Al2O3 scales formed on the surface a of Ni–18Al alloy and evaluation of its dependence on the thickness of NiO and NiAl2O4 films on sapphire substrates. The presented method is potentially applicable to nondestructive detection of internal α-Al2O3 scales underneath multiple scales on heat-resistant alloys. However, the dissolution of Ni2+ into the α-Al2O3 scales reduced the CL peak intensity at 695 nm of the Ni-14Al alloy by two orders of magnitude compared to that expected from the results with NiO and NiAl2O4 films on sapphire substrates. Future work will involve investigating the effect of Ni2+ dissolving into α-Al2O3 scale on the CL intensity at 695 nm in more detail to precisely determine the maximum thicknesses of NiO and NiAl2O4 scales that will still allow signal detection of internal α-Al2O3 scales on Ni–Al alloys.
Financial support for the present study was provided by The Iron and Steel Institute of Japan (ISIJ) through the Research Group of Non-Destructive/On-Site Analysis.
