Effects of Cr, W, and Mo on the High Temperature Oxidation of Ni-Based Superalloys

The oxidation behavior of Ni–9.5Co–(8~12)Cr–(2.5~5.5)Mo–(4~8)W–3Al–5Ti–3Ta–0.1C–0.01B alloys was investigated at 850 °C and 1000 °C The mass change, the phase of oxides, and the cross-sectional structure of specimens were analyzed after cyclic oxidation tests. The oxide scale was composed mainly of Cr2O3 and NiCr2O4, but NiO, TiO2, and CrTaO4 were also found. Al2O3 was formed beneath the Cr oxide layer. The Cr oxide layer and internal Al oxide acted as barriers to oxidation at 850 °C, while Al oxide was predominantly protective at 1000 °C. Cr increased the mass gain after oxidation test at both temperatures. Mo increased the oxidation rate at 850 °C but decreased the oxidation rate at 1000 °C. W slightly increased the mass gain at 850 °C but did not produce a significant effect at 1000 °C. The effects of Cr, Mo, W, and the temperature were discussed as well as the volatilization of oxides, the valence number of elements, and diffusion retardation.


Introduction
Ni-based superalloys are used for heat exchangers, gas turbine blades, and chemical reactors because of their excellent oxidation resistance and mechanical strength at high temperatures. The oxidation behavior of superalloys is influenced by their chemical compositions. Cr and Al are the representatives of protecting alloying elements [1][2][3][4][5][6][7]. Cr forms a dense Cr 2 O 3 layer on the surface of Ni-based superalloys and reduces the oxidation rate. It is very effective in protecting the alloy at temperatures up to about 871 • C (1600 • F). Above this temperature, Cr 2 O 3 becomes unstable and volatile CrO 3 forms. The role of Al is crucial at higher temperatures. When Al forms a dense and continuous layer of Al 2 O 3 , oxidation of the alloy is significantly reduced. The formation of the Al 2 O 3 layer depends on the temperature and alloy composition [8][9][10][11][12].
Recent studies show that an alloying element can change the role of another element so that contradictory results are obtained depending on the alloy composition and temperature [8,28,29]. For example, increasing the Ta content in alloys with a low concentration of Al reduces the oxidation rate, but there is almost no effect in alloys with a high Al content [8]. Cr increases the mass gain of alloys with W during oxidation, but decreases the mass gain of the alloys without W [29]. Furthermore, The oxidation test was performed for six cycles. Each cycle consisted of the following procedures: The specimens were put into the furnace when the temperature was 400 • C. The temperature was increased up to 850 or 1000 • C over 2 h and kept constant for 15 h. After thermostatic oxidation, the specimens were slowly cooled in the closed furnace with the power off for 2 h, and then in the furnace door open for about 2 h until the temperature reached 400 • C. The samples were retrieved at 400 • C and fully cooled in air.
The weight of the oxidized samples, including the fallen scales, was measured by a precision balance to an accuracy of five digits after the decimal point. The phase of the oxide scale was analyzed by using X-ray diffraction (XRD, X'pert Pro MPD, Malvern Panalytical, UK) on the oxide layer. The structure and the composition of the cross-section were examined by using scanning electron microscopy (SEM, S-4800, Hitachi, Japan) equipped with energy dispersive spectroscopy (EDS). Figure 1 shows pictures of the specimens after oxidation test. The surfaces of the samples were dark gray in color. Oxidation was more severe at the higher temperature, as evidenced by spalled oxide particles around the specimens tested at 1000 • C. The weight of the alloys increased with oxidation, as shown in Figure 2. The mass gain after six cycles of oxidation at 850 °C was 0.46~0.62 mg/cm 2 . The oxidation rate was relatively high in the initial cycles and then gradually reduced with successive cycles. The weight of a specimen increased quickly when the oxide layer formed on the bare surface. The oxidation rate slows down after the initial stage, because the rate of oxide growth is controlled by the diffusion of oxygen and alloying elements in the oxide scale, as implied by the parabolic trend of the curves. The final mass gain at 1000 °C was 2.0~3.4 mg/cm 2 , much higher than that measured at 850 °C. The mass gain curves indicate continued growth of the oxide scale, except in the case for the 8Cr-3.5Mo-6W alloy at 1000 °C. The mass gain of 8Cr-3.5Mo-6W increased up to the third cycle but thereafter it became nearly constant. This is thought to be related to the volatilization of Cr hexavalent oxide. Severe spallation of scale on 8Cr-3.5Mo-6W, as seen in Figure 1, may also be a reason that no further increase of mass gain. Because we did not use any cover on the alumina crucible, part of the spalled oxide could have been lost during the tests. The weight of the alloys increased with oxidation, as shown in Figure 2. The mass gain after six cycles of oxidation at 850 • C was 0.46~0.62 mg/cm 2 . The oxidation rate was relatively high in the initial cycles and then gradually reduced with successive cycles. The weight of a specimen increased quickly when the oxide layer formed on the bare surface. The oxidation rate slows down after the initial stage, because the rate of oxide growth is controlled by the diffusion of oxygen and alloying elements in the oxide scale, as implied by the parabolic trend of the curves. The final mass gain at 1000 • C was 2.0~3.4 mg/cm 2 , much higher than that measured at 850 • C. The mass gain curves indicate continued growth of the oxide scale, except in the case for the 8Cr-3.5Mo-6W alloy at 1000 • C. The mass gain of 8Cr-3.5Mo-6W increased up to the third cycle but thereafter it became nearly constant. This is thought to be related to the volatilization of Cr hexavalent oxide. Severe spallation of scale on 8Cr-3.5Mo-6W, as seen in Figure 1, may also be a reason that no further increase of mass gain. Because we did not use any cover on the alumina crucible, part of the spalled oxide could have been lost during the tests.  The effects of the alloy composition on the mass gain by oxidation are summarized in plots in Figure 3. Figure 3 shows the averaged mass gain of the samples as a function of the content of each element. The mass gain slightly decreased when the Cr content increased from 8 wt.% to 10 wt.% but it increased when Cr the content was increased to 12 wt.% (Figure 3a). Mo shows similar behavior in that the mass gain increased with an increase of Mo to over 4.5 wt.%. W increased the mass gain very slightly. The plots for 1000 °C (Figure 3b) suggest more clearly that Cr, Mo, and W increase the mass gain and the dependence is consistent over the composition range of this study. This result suggests that the three alloying elements do not suppress the oxidation of Ni-based superalloys effectively. The effects of the alloy composition on the mass gain by oxidation are summarized in plots in Figure 3. Figure 3 shows the averaged mass gain of the samples as a function of the content of each element. The mass gain slightly decreased when the Cr content increased from 8 wt.% to 10 wt.% but it increased when Cr the content was increased to 12 wt.% (Figure 3a). Mo shows similar behavior in that the mass gain increased with an increase of Mo to over 4.5 wt.%. W increased the mass gain very slightly. The plots for 1000 • C (Figure 3b) suggest more clearly that Cr, Mo, and W increase the mass gain and the dependence is consistent over the composition range of this study. This result suggests that the three alloying elements do not suppress the oxidation of Ni-based superalloys effectively. The oxide scale formed at 850 °C appeared to mainly consist of Cr2O3 and NiCr2O4 from the XRD pattern ( Figure 4a). NiO, TiO2, and CrTaO4 were also detected. The peaks for Al2O3 were barely measured. At 1000 °C (Figure 4b), peaks for metallic phases (matrix and Ni3Al) were not measured, implying that the oxide scale is thick. The peaks for NiCr2O4 were the strongest among the studied oxide peaks at 1000 °C. The number and intensity of the Cr2O3, CrTaO4, and NiO peaks decreased significantly, while the peak for TiO2 seemed to be similar to the XRD patterns for 850 °C. Al2O3 was detected more frequently in the data for 1000 °C than in the data for 850 °C. The difference in the XRD patterns confirms that Cr2O3 is predominant in the scale formed at 850 °C and becomes unstable at 1000 °C. The oxide scale formed at 850 • C appeared to mainly consist of Cr 2 O 3 and NiCr 2 O 4 from the XRD pattern ( Figure 4a). NiO, TiO 2 , and CrTaO 4 were also detected. The peaks for Al 2 O 3 were barely measured. At 1000 • C (Figure 4b), peaks for metallic phases (matrix and Ni 3 Al) were not measured, implying that the oxide scale is thick. The peaks for NiCr 2 O 4 were the strongest among the studied oxide peaks at 1000 • C. The number and intensity of the Cr 2 O 3 , CrTaO 4 , and NiO peaks decreased significantly, while the peak for TiO 2 seemed to be similar to the XRD patterns for 850 • C. Al 2 O 3 was detected more frequently in the data for 1000 • C than in the data for 850 • C. The difference in the XRD patterns confirms that Cr 2 O 3 is predominant in the scale formed at 850 • C and becomes unstable at 1000 • C. (a)

Results
10Cr-3.5Mo-4W The thickness of the oxide scale for the samples oxidized at 850 °C was several micrometers, as shown in Figure 5. The scale layer was mainly composed of Cr2O3 and discontinuous particles of Al2O3 were present just beneath the Cr2O3 layer. TiO2 was often found at the outermost part of the scale. The thickness of the oxide scale for the samples oxidized at 850 • C was several micrometers, as shown in Figure 5. The scale layer was mainly composed of Cr 2 O 3 and discontinuous particles of Al 2 O 3 were present just beneath the Cr 2 O 3 layer. TiO 2 was often found at the outermost part of the scale. seem to be so different for the three alloys, except that Al2O3 is a little continuous or large in the alloy with low W content (10Cr-3.5Mo-4W in Figure 5a). This result can be understood similarly to the case of Mo, in that W is also a high valance element that becomes W 4+ or W 6+ ion. Therefore, W can inhibit the diffusion of oxygen through the scale effectively. In addition, W is large in size and can retard the diffusion of other atoms nearby and the formation of internal Al2O3 is reduced, causing a relatively low oxidation rate of the alloy.  From the images of the (8-12)Cr-3.5Mo-6W alloys shown in Figure 5b, f and h, it is noted that the size of Al 2 O 3 particles was larger in the alloys with lower Cr content. The high Cr content and the relatively low temperature as 850 • C are thought to be favorable for preferential formation of a stable Cr 2 O 3 layer and hence Al 2 O 3 might grow slowly due to the low diffusion of oxygen. The mass gain ( Figure 3a) decreased with the increase in Cr content up to 10 wt.%, beyond which it then increased. These results imply that the oxidation resistance depends largely on the protectiveness of the Cr oxide layer at 850 • C but Al oxide plays an important part as well.
Figure 5c-f show the images for 10Cr-(2.5-5.5)Mo-6W alloys. The distribution depth of Al oxides tended to be smaller for the alloys with a higher Mo content, though it was not a linear function. It is anticipated that inward diffusion of oxygen is impeded by Mo, therefore the penetration depth of oxygen is limited in high Mo alloys. Mo is a high valence ion of Mo 4+ or Mo 6+ , so it consumes more oxygen than Cr or Al. Mo is assumed to be incorporated or to form a small amounts of Mo oxide, as it was not detected by XRD (Figure 4). Since more Mo in the Cr 2 O 3 layer bonds with more oxygen, the amount of oxygen which can diffuse into the region beneath the Cr 2 O 3 layer is reduced in the alloys with higher Mo content. Consequently, decreased formation of Al 2 O 3 results in increased mass gain during oxidation.
The effect of W can be seen in Figure 5a, f and g. The thickness and morphology of oxides do not seem to be so different for the three alloys, except that Al 2 O 3 is a little continuous or large in the alloy with low W content (10Cr-3.5Mo-4W in Figure 5a). This result can be understood similarly to the case of Mo, in that W is also a high valance element that becomes W 4+ or W 6+ ion. Therefore, W can inhibit the diffusion of oxygen through the scale effectively. In addition, W is large in size and can retard the diffusion of other atoms nearby and the formation of internal Al 2 O 3 is reduced, causing a relatively low oxidation rate of the alloy. Figure 6 shows that the thickness of the oxide scale formed at 1000 • C was about 10 times thicker than that formed at 850 • C. The Cr 2 O 3 layer formed at 1000 • C was porous and some cracks or voids are present between the Cr 2 O 3 layer and TiO 2 or NiO. Internal Al 2 O 3 lumps had a more columnar shape than those formed at 850 • C. It can be interpreted that the stability of Cr 2 O 3 is poor at 1000 • C and some Cr oxide were lost by volatilization of CrO 3 , leaving pores in the oxide layer. The porosity would increase the diffusion rate of oxygen; thus, the Al oxide could be formed at a deeper location. Diffusion of Al would also be greatly promoted at 1000 • C, so the size of Al 2 O 3 would grow faster than at 850 • C.

Conclusions
The effects of Cr, Mo, and W on the high temperature oxidation of Ni-based superalloys were examined by cyclic oxidation tests at 850 ° and 1000 °C.
The oxidation scale consists mainly of Cr2O3 and NiCrO4 with some NiO, TiO2, and CrTaO4. Al2O3 was formed under the scale discontinuously.
The mass gain after 20 cycles of oxidation tests was increased by increasing Cr, Mo, and W at 850 °C. The alloys with higher Cr, Mo, or W contents showed oxide structures with less Al2O3. Al2O3 appears to have an important role in protecting the alloy from oxidation even at this temperature as the mass gain was lower for the alloys with more Al2O3, although the dense Cr2O3 layer is present.
The oxidation rates were much higher at 1000 °C than at 850 °C. The mass gain was increased by adding Cr but reduced by adding Mo. W did not show a clear effect. These phenomena are explained by the complex effects of the volatilization of Cr, Mo, and W oxides at this temperature, the high valence of Mo and W reducing metal vacancy in the Cr2O3 layer, and the retarded diffusion due to the large size of W.  We discuss the effects of the Cr content on the oxide structure, as shown in Figure 6b, [31][32][33][34][35][36]. This reaction involves the reduction of Cr 2 O 3 and NiO, as confirmed by the XRD patterns in Figure 4. The protectiveness of the oxidation scale deteriorates due to the loss of Cr 2 O 3 layer and insufficient formation of Al 2 O 3 , and the mass gain increased with the increase of Cr content as shown in Figure 3b.
For the 10Cr-(2.5-5.5)Mo-6W alloys (Figure 6c-f), the porosity of the Cr 2 O 3 layer was notably higher in the 10Cr-(3.5-5.5)Mo-6W alloys than in the 10Cr-2.5Mo-6W alloy. Cr 2 O 3 is a p-type semiconductor [37] with a high concentration of metal vacancies. When the higher valence metal is incorporated in the Cr 2 O 3 , the metal vacancies increase, and the metal diffusion rate through the Cr 2 O 3 layer is enhanced. However, if Mo in the scale forms separate oxides such as MoO 2 or MoO 3 because of an even higher Mo content, the vacancy concentration will not increase further. It should be noted that MoO 3 is also volatile [13][14][15][16][17]. The increase in metal vacancy and the increase in volatilization, as mentioned above, could be the reason for the higher porosity of the Cr oxide layer in alloys with a higher Mo content. However, it was observed that more Al 2 O 3 lumps were formed and some of them were connected, although not completely continuous, in the alloy with a higher Mo content. It is considered that the higher oxygen diffusion through the porous Cr oxide layer formed in high Mo alloys promotes the internal oxidation of Al. Al 2 O 3 provides protection against further oxidation. Consequentially, Mo decreased the mass gain at 1000 • C as shown in Figure 3b, when the Mo content increased up to 3.5 wt.%. Figure 6a, f and g show the effects of W on the oxide structure. Al 2 O 3 formed more in the 10Cr-3.5Mo-6W alloy and the thickness and porosity of Cr 2 O 3 were also increased as the W content increased. These increments are shown in the figures for the alloys with 4-6 wt.%. W is thought to exert similar but more complicated effects than Mo, as it is a high valence element which forms a volatile oxide [38] and has a low diffusion rate because of its large size. W causes low oxygen activity due to its high valence and can result in the suppression of CrO 3 formation. This proposition is also supported by Huang et al. [19], who reported that W suppresses the volatilization of Cr oxide in from Ni-Cr-W alloys at temperatures higher than 1100 • C. However, WO 3 itself has volatility, making pores in the scale. The large size of W retards the diffusion of other elements and the formation of Al 2 O 3 can also be retarded. The effects of W on various volatilizations and diffusion rates of elements are not explicitly studied in this study. We can only presume that these counteractions summed up and caused almost zero dependence of mass gain on the W content, as Figure 3b shows.
Park et al. [29] previously reported that the main effects of the alloying elements were as that Cr and Mo increased mass gain slightly, but W did not affect the mass gain by oxidation of Ni-based superalloy at 1000 • C. According to their reports for the interactions between elements, the effects of Cr and Mo were rarely affected by other elements. One exception is that the increasing effect of Cr and Mo were more prominent with alloys with 3 wt.% Al content than with alloys with higher Al content. In contrast, W slightly reduced the mass gain for the 3 wt.% alloys and slightly raised the mass gain of alloys with Mo content over 2.5 wt.%. Considering that the content of Al was 3 wt.% and that of Mo was 3.5 wt.% in the comparative study of this work, the effects of W appear to be compensated by the interactions of the Al and Mo contents.

Conclusions
The effects of Cr, Mo, and W on the high temperature oxidation of Ni-based superalloys were examined by cyclic oxidation tests at 850 • and 1000 • C.
The oxidation scale consists mainly of Cr 2 O 3 and NiCrO 4 with some NiO, TiO 2 , and CrTaO 4 . Al 2 O 3 was formed under the scale discontinuously.
The mass gain after 20 cycles of oxidation tests was increased by increasing Cr, Mo, and W at 850 • C. The alloys with higher Cr, Mo, or W contents showed oxide structures with less Al 2 O 3 . Al 2 O 3 appears to have an important role in protecting the alloy from oxidation even at this temperature as the mass gain was lower for the alloys with more Al 2 O 3 , although the dense Cr 2 O 3 layer is present.
The oxidation rates were much higher at 1000 • C than at 850 • C. The mass gain was increased by adding Cr but reduced by adding Mo. W did not show a clear effect. These phenomena are explained by the complex effects of the volatilization of Cr, Mo, and W oxides at this temperature, the high valence of Mo and W reducing metal vacancy in the Cr 2 O 3 layer, and the retarded diffusion due to the large size of W.