2018 年 59 巻 4 号 p. 563-567
Alumina-forming oxide dispersion strengthened (ODS) superalloys are favorable oxidation-resistant materials for extremely high temperature applications. In order to develop the advanced Co-based superalloys with high strength and superior oxidation resistance at elevated temperature of 1000℃, a new series of Co-Cr-Al ODS superalloys were designed and fabricated by mechanical alloying (MA) and spark plasma sintering (SPS), and then followed by hot rolling and annealing at 1200℃. In this work, the oxidation behavior of Co-10Al (mass%) ODS superalloys with/without 20Cr was investigated at 1000℃ in air to understand the effect of Cr on oxidation resistance. The results indicate that the addition of Cr improves the oxidation resistance significantly through optimizing the oxide scales from the multilevel scales with an external CoO/CoAl2O4 and an internal Al2O3 to a single layer of Al2O3. The alumina-forming Co-20Cr-10Al (mass%) ODS superalloys are expected to be applicable at 1000℃.
Co-based alloys are used to produce the components for gas turbine engine and power generation system due to their inherent advantage of outstanding hot corrosion resistance1,2). However, the application of Co-based alloys has been significantly restricted, since a shortage of valid strengthening methods at high temperature. In general, conventional Co-based superalloys are strengthened by solid solution strengthening and carbide precipitation strengthening, but both of these strengthening methods would be ineffective due to dissolution or coarsening of carbides at the temperature above 800℃3). Even though a γ'-Co3(Al, W) precipitate has been discovered latterly for strengthening the Co-based alloys4), the γ' phase is unstable at the temperature above 900℃5). It's significant to increase the service temperature of Co-based alloys, since it enables to remarkably increase the thermal efficiency of the industrial gas turbine. In order to further improve their applicable temperature and simultaneously retain the strength at elevated temperature, the advanced strengthening approaches for developing Co-based alloys are necessary. Meanwhile, with increasing temperature, a favorable oxidation resistance also should be key issue to be considered. The best high-temperature oxidation resistance could be achieved through developing outer/single Al2O3 scale because of its excellent stability at elevated temperature, but this alumina former has high-demanding on alloys composition with high Al content. The single Al2O3 formation has been proved to be inapplicable for the conventional carbides and γ' strengthened Co-based alloys because of their narrow composition range.
Based on the superiority of oxide dispersion strengthening (ODS) with stable oxide particles up to 1200℃ and flexibly controlled composition6), developing Co-based ODS superalloys for extremely high temperature applications has been focused, in which the strength and oxidation resistance are expected to be guaranteed by a reliable oxides pinning effect and the single Al2O3 former at elevated temperature, respectively. In fact, the Ni-based and Fe-based ODS alloys have been developed and already partially used as industrial materials6,7). In addition to their superior creep resistance and favorable high-temperature strength, ODS superalloys show better oxidation resistance than the ODS-free alloys with the same basic composition, which has been frequently proved through the research of Fe- and Ni-based ODS superalloys8). However, the Co-based ODS system has received much less attention than Ni- and Fe-based ODS superalloys, and even no reports on the high-temperature oxidation of Co-based ODS superalloys can be referenced.
Since Hf was claimed to be fairly effective to refine dispersoids through forming small Y2Hf2O7 particles9), a Co-3Al-1.5Y2O3-1.2Hf ODS alloy was discovered by our previous work. Takezawa et al.10) studied the composition and found that the Y2Hf2O7 oxide particles are extremely stable at the temperature above 1000℃, suggesting that the high temperature strength of this Co-based ODS superalloy can be enhanced dramatically as compared with the conventional Co based superalloys. Besides, Sasaki et al.11) also announced that the above Co-based ODS alloys were able to maintain high hardness with an outstanding microstructure stability after annealing at 1000℃ for 240 h, owing to the pinning of grain boundaries by the dispersed oxide particles.
To develop advanced Co-based ODS superalloys for extremely high temperature applications with favorable strength and superior oxidation resistance at elevated temperature, in addition to the dispersion of fine Y2Hf2O7 dispersoids, the formation of a single alumina scale is another key target. Therefore, a higher Al content was added in the present Co-based ODS system, and chromium was also introduced to promote the formation of alumina scale according to the reports claimed by G.N. Irving et al. that the addition of Cr to Co-Al base alloys markedly reduced the level of Al required in the ternary alloy below that required in the binary to form a continuous external scale of Al2O3 12).
Aimed at discovering alumina-forming Co-based ODS superalloys and making clear the influence of Cr on high-temperature oxidation behavior of the Co-based ODS system, Co-(20Cr-)10Al-2.4Hf-1.5Y2O3 (mass%) ODS superalloys with and without Cr addition were designed and fabricated. In this work, the investigation of oxidation resistance at 1000℃ in air was performed through an isothermal oxidation. Other oxidation results of the Cr and Y2O3 influence at a low temperature of 900℃ and the effect of Al content on oxidation behavior will be published elsewhere13,14).
The Co-20Cr-10Al-2.4Hf-1.5Y2O3 (mass%) ODS samples and Co-10Al-2.4Hf-1.5Y2O3 (mass%) ODS samples were prepared and referred to as 10AlCr and 10Al, respectively. Both samples are composed of two different phases, i.e. Co solid solution of fcc structure, and B2 phase highly enriched with aluminum, as shown in the phase diagram (Fig. 1) computed with the thermodynamic calculation software Pandat15). All the elemental powders were mixed and mechanically alloyed (MAed) for 48 h under argon gas atmosphere using a planetary type ball mill (Fritsch P-5). The mechanically alloyed powders were consolidated in a graphite mold by spark plasma sintering (SPS) at 1100℃ and 45 MPa for 2 h, an then followed by a hot-rolling at 1200℃ into samples of 3 mm thickness. The final annealing was conducted at 1200℃ for 1 h under vacuum of 10−4 Torr.
The ternary Co-Cr-Al phase diagram calculated with Pandat software at 1100℃.
The samples for oxidation test were cut into a cuboidal shape, and all the surfaces of investigated samples were wet ground using SiC paper up to 2400-grit, and then polished with 3 μm and 1 μm-diamond paste. After confirming that no residual scratches can be observed in the surfaces by an optical microscopy, the samples were ultrasonically cleaned in alcohol, and dried in hot air. Isothermal oxidation test at 1000℃ was carried out inside a muffle furnace in air. In order to determine the oxidation kinetics, the oxidation weight gains in different exposure durations (1, 4, 16, 25, 49, 100, 225 h) were measured at room temperature via a Sartorius precision electronic balance with a resolution of 1 μg.
The crystalline structures of oxide layers were detected using X-ray diffraction (XRD, Philips X' Pert PRO), and the 3º incident angle has been adopted to emphasize the semaphore of thin oxide scales. The surface morphologies of oxide scales were investigated by scanning electron microscope (SEM, Carl Zeiss Cross Beam 1540 EsB). The polished cross sections of oxidized samples were analyzed by the electron probe micro-analyzer (EPMA, JEOL JXA-8530F), in which the elemental mapping and quantitative point analysis by wavelength dispersive spectrometer (WDS) were carried out.
Figure 2 shows the X-ray diffraction pattern of the specimens after final annealing at 1200℃ for 1 h. Both the Co-based ODS superalloys with and without Cr addition are mainly composed of two phases; the Co solid solution phase of fcc structure, and the CoAl phase of B2 structure.
XRD results for the 10Al and 10AlCr alloys after annealing 1 h at 1200℃.
Figure 3 shows the backscattered electron microscopy (BSE) images of the metallographic analysis. Due to higher aluminum content in the CoAl (B2) phase than that in the Co solid solution phase (fcc), it's evident to distinguish the B2 and fcc phase through the image contrast at the BSE mode. As shown in the Fig. 3 (a) and (b), both the samples basically consist of two regions; the dark region marked with dashed box 1 and the light grey matrix noted with dashed box 2. Our previous work16) has proved that the dark region is CoAl phase with B2 structure and the grey matrix is cobalt solid solution with fcc structure. With the participation of Cr, a larger area of dark regions could be found in the sample of 10AlCr, as shown in Fig. 3 (b). The distribution of those dark and grey areas in the 10Al and the 10AlCr samples is qualitatively consistent with the volume fractions of the B2 and the fcc in the XRD results.
SEM images of 10Al (a) and 10AlCr (b) at the manufactured condition.
Volume fraction of the fcc matrix and the B2 phase in the 10Al and the 10AlCr alloys were measured through analyzing the BSE images with ImageJ software, as shown in Fig. 4, in which the corresponding results predicted with the Pandat software has been introduced, as well. Experiment and analyses show quite good agreement. In addition, the concentration of primary elements in the fcc and B2 phases measured by EPMA is listed in Table 1. It has been verified by Yu et al.16) that the actual concentrations of those Co, Cr, Al in the B2 and the fcc phase are close to the values predicted by Pandat software, even though the Y2O3 powder was added during MA process.
Volume fractions of fcc and B2 phase in the 10Al and 10AlCr alloys obtained from experimental measurement and thermodynamic calculation software Pandat, respectively.
The observed weight changes for the 10Al and 10AlCr samples as a function of time during exposure at 1000℃ in air are shown in Fig. 5. It's noticeable that the Co-based ODS superalloys with and without Cr addition have presented totally different oxidation kinetics, namely, a drastic mass changing happened in the 10Al samples, whilst a smooth oxidation process was retained in the 10AlCr alloys. Firstly, the oxidation rate of 10Al sample in the initial stage up to 1 h was much faster than that in the 10AlCr sample, resulting in a large mass gain of 0.5 mg/cm2 in 1 h, which is twice more than that in the 10AlCr. Subsequently, after a transient mass growth, an evident spallation occurred and resulted in the evident mass loss, as shown in the Fig. 5. However, by the addition of Cr, in addition to a good spalling resistance, the 10AlCr samples possess an extremely slow oxidation rate because of the low mass gain of less than 0.6 mg/cm2 even up to exposure times to 225 h at 1000℃.
Mass gain vs time curves of the 10Al and the 10AlCr samples oxidized at 1000℃ in air.
In order to understand the growth of oxide scales during the oxidation process, X-ray diffraction has been conducted to detect the samples in different exposure time. Figure 6 shows the scales crystalline structures of the 10Al and 10AlCr alloys oxidized at 1000℃. Based on the characteristic peaks of CoO/CoAl2O4 and A2O3 in X-ray diffraction pattern, as shown in the Fig. 6 (a), the attendance of CoO/CoAl2O4 in the 10Al oxide scales was confirmed and should dominate the scales. Besides, one point should be noted that the diffraction peaks of CoO were faint suddenly in the XRD pattern of 25 h, as marked by a triangle. By contrast, in the case of the 10AlCr sample, only A2O3 diffraction peaks were confirmed, and no obvious peak fluctuation could be observed even up to 225 h, as shown in the Fig. 6 (b).
XRD results of the 10Al (a) and the 10AlCr (b) alloys oxidized at 1000℃ in air with different exposure time.
In addition, the cross sections of samples oxidized at 1000℃ also have been observed and the results are exhibited in Fig. 7. In comparison with a flat and smooth scale in the sample of 10AlCr, the scales of 10Al are rough and one incomplete film at outer layer was observed. In addition, based on the contrast difference in Fig. 7 (a), it could be determined that the oxide scales in the 10Al alloys are composed of three layers.
Scales morphologies of 10Al (a) and 10AlCr (b) oxidized at 1000℃ with 100 h.
In order to further identify the distribution of elements in the oxide scale, the EPMA elemental mapping was carried out for the specimens oxidized at 1000℃ in 100 h, and the results of the 10Al and the 10AlCr samples are displayed in Fig. 8 (a) and (b), respectively. According to the Fig. 8 (a), the multiply scale in the 10Al could be evidently separated, i.e., in addition to an external scale containing Co and an inner scale including Al, there are another intermediate layer enriched Co and Al. According to the XRD diffraction pattern in Fig. 6 (a), those scales along the direction from air to substrate were considered to be CoO, CoAl2O4 and Al2O3, respectively. In respect of the 10AlCr sample, as shown in Fig. 8 (b), the homogeneous oxide layer was found to contain aluminum and oxygen elements, which could be easily defined as alumina according to the results of XRD in Fig. 6 (b).
EPMA elemental mappings obtained from the 10Al (a) sample oxidized at 1000℃ with 100 h and 10AlCr (b) at 1000℃ with 225 h.
It is well known that the presence of Cr in Fe- and Ni-based alloys could reduce the Al level needed to form a protective Al2O3 layer17,18), defined as the third-element effect (TEE)19). This indicates that the addition of a third-element B to binary A–C alloys can decrease the critical C concentration needed to establish an external scale of the C oxide, in which A is the most inert element and C is the most reactive one in the ternary system17). The TEE seems to also work in the Co-based ODS superalloys, since the Cr possesses an oxygen affinity falling in between those of Co and Al, and also the addition of Cr reduces the concentration of Al required to establish an external alumina scale, as shown in the Fig. 7 where an exclusive alumina scale could be formed through adding Cr to the 10Al samples. The mechanism of TEE is not yet completely clear, but based on the suggestion claimed by Wagner19,20), it was supposed that an initially formed external Cr2O3 scale promotes Al2O3 formation. Yoneda et al. proposed that the outward diffusion flux of Al could be enhanced by Cr addition through analyzing the transient stage oxidation of Fe-6Al with different Cr content21). In this study we could not observe Cr2O3 formation at the very early stage of oxidation. However, Cr2O3 formation is expected, since Co-rich oxide scale formation was completely suppressed by Cr addition.
In addition, based on the phase diagram (Fig. 1), due to the addition of 20Cr (mass%), the volume fraction of CoAl (B2) is significantly increased from 39% in the 10Al to 56% in the 10AlCr, which is consistent with Fig. 4. It should be noted that the aluminum concentration in the B2 phase is almost double than that in the fcc phase. Thus, due to the high Al supply with more B2 phase volume fraction, the diffusion flux of aluminum is sufficient to maintain the growth rate of outer alumina scale in 10AlCr. The EPMA WDS analysis was conducted for the cross-section of the 10AlCr sample, as shown in Fig. 9. It's noticeable to find that the disappearance of B2 phase beneath the alumina scale, which is consistent with the “B2 free zone” in the Fig. 9. The reason why the B2 phase disappeared near the scale is that the alumina formation consumed Al in the B2 phase. Figure 9(b) shows that the Al content in the “B2 free zone” is low and similar to that in the fcc matrix. The alumina formation is governed by the inward transport of oxygen and outward transport of Al in the Co substrate (fcc and B2). The higher Al content in the B2 phase would provide a larger tendency for the outward diffusion of aluminum, and then the alumina preferentially formed near the B2 site since a rich diffusion flux of aluminum from the B2 phase. Even the Al concentration in the fcc phase is quite lower than that in the B2 phase, which is not beneficial for Al2O3 scale formation, a high Cr concentration in the fcc matrix could promote external Al2O3 scale formation by TEE as mentioned above.
The BSE image of cross-section (a) and corresponding EPMA WDS point analysis (b) for the 10AlCr sample oxidized at 1000℃ in 225 h.
Oppositely, the 10Al alloys is preferred to form Co oxides due to a high percentage of cobalt solid solution regions, resulting in a continuous Co-oxide scale at external layer. Even though initially small amounts of alumina could generate at the sites of B2 phases in the 10Al sample, it is insufficient to evolve a complete outer Al2O3 scale because of the limited aluminum flux. Consequently, due to a higher growth rate of CoO, a thick CoO scale could be formed on the surface of the 10Al alloy at initial oxidation, which is consistent with the large mass gain of 0.5 mg/cm2 in 1 h and the strong diffraction CoO peaks in the XRD pattern of 4 h. The mass gain would persistently increase with the exposure time. However, the thermal stress was also enhanced with the increasing of CoO scale thickness. Due to a high growth rate of CoO and the large thermal stress caused by the mismatch of the coefficients of thermal expansion between the CoO and matrix, the detachment of CoO layer is usually observed in the Co-based alloys22,23). Therefore, it was proposed that the CoO scale spalled on cooling when it oxidized to a certain thickness with the 25 h oxidation, which is fit with the abrupt reduction of CoO diffraction peaks in the XRD pattern of 25 h. However, the CoO is not completely detached from samples, as proved with the faint CoO diffraction peaks in the XRD pattern of 25 h, some residual CoO parts still attached at the surface. After mass gain measurement, the samples were sent back to furnace for further oxidation, and in this process some new CoO would form again through healing process. In the subsequent oxidation (40 h or 100 h), since the newly formed CoO scale is incomplete, as shown with the island CoO scale in the Fig. 8 (a), the thermal stress was easier to relieve and resulted in less spallation. Therefore, CoO diffraction peaks were revived at 40 h and 100 h even though the weight loss was continuously occurred.
In order to develop advance alumina-forming Co-based ODS superalloys for extremely high temperature applications, an innovative Co-20Cr-10Al-2.4Hf-1.5Y2O3 (mass%) superalloy has been proposed. The effect of Cr on the microstructure and the oxidation resistance (at 1000℃) of the novel composition system has been investigated. Even both the alloys with and without Cr addition are composed of the Co-based solid solution with the fcc structure and the CoAl phase with B2 structure, the addition of Cr markedly increased the volume fraction of B2 phase from 39% to 56%. Based on the oxidation behavior characterization, the addition of Cr into Co-based ODS superalloys has proven to effectively improve high-temperature oxidation resistance at 1000℃ by accelerating the formation of an exclusive alumina scale instead of the multilayered scale with an external CoO/CoAl2O4 and an internal Al2O3. The spallation of CoO scale during cooling process gives rise to a weight loss in the Cr-free samples. The Co-20Cr-10Al ODS superalloys possess far better oxidation resistance with the formation of a stable alumina scale and was expected to be available at elevated temperature up to 1000℃.
This work is supported by Grant-in-Aid for Scientific Research (B), 24360282, Japan Society for the Promotion of Science (JSPS). The authors would like to thank the laboratory of Nano-Micro Material Analysis in Hokkaido University for the utilizing of EPMA. H. Yu is grateful to the China Scholarship Council (CSC) for the provision of a scholarship.
