Effect of Y on Oxidation Behavior of Directionally Solidified Ni-Based Single-Crystal Superalloy

The effect of yttrium (Y) addition on the oxidation behavior of a Ni-based directionally solidified single-crystal superalloy is investigated in this study. Isothermal oxidation tests for samples with different levels of Y addition are conducted at 1100 °C in air. The Y content of the samples is determined by the actual pickup amount obtained from an Inductively Coupled Plasma-Atomic Emission Spectrometry test. It is discovered that the addition of Y increases the oxide resistance by the scale of an adhesive double-layer oxide, which is composed of Al 2 O 3 and spinel Ni(Cr,Al) 2 O 4 . With 70 ppm of Y addition, the oxidation mass gain decreases from 12.6 g/m 2 for the alloy without Y addition to 5.3 g/m 2 , and the oxidation rate decreases significantly. In addition, the internal nitride disappears after Y doping because of an increase in oxidation scale adherence and a decrease in oxidation products. In this study, the alloy with 660 ppm Y addition demonstrates the best oxidation resistance.


Introduction
Nickel-based superalloys have been used as the main materials in the aerospace industry and industrial gas turbines owing to their excellent properties such as high strength, relatively low density, and high environmental resistance at high operating temperatures [1][2][3].Turbines, such as high-pressure turbines or combustors, are exposed to high temperature and oxygen-enriched environments, which implies that they must be manufactured using materials of high mechanical properties [4,5].The oxidation resistance of alloys at high temperatures is primarily achieved by the formation of a passive oxide scale [6][7][8].
The addition of rare earth (RE) elements can reduce the amount of impurities in the melt form because of the active properties of RE, which significantly affect the oxidation resistance of the alloy through the formation of an adhesive protective oxide scale that has a low growth rate [9,10].In 1966, Francis and Whitlow first added yttrium (Y) to a Cr-Fe alloy, which was proven to improve the oxidation resistance [11].In the past decades, a number of studies have shown that the addition of RE elements to superalloys improved the oxidation resistance properties of alloys in hightemperature environments [12,13].However, the mechanism associated with the effect of RE elements on the oxide scale remain ambiguous.Several researchers deduced that Y doping can result in better scale adhesion because of the "tying up" effect on S by RE elements [14].However, all the sulfide-forming cases were based on the premise that the alloy involved had high sulfur activity and low oxygen activity.Furthermore, that sulfides could not be observed in the alloy were considered as well.From another viewpoint, RE elements would "site block" at the grain boundaries to inhibit the outward diffusion of Al 3+ [15].Recently, Heuer and Hovis proposed a new perspective, i.e., REs might affect the oxidation behavior by reducing both the diffusion of O 2− to the grain boundaries and the Al ionization [16].
In this study, different amounts of Y were added to a directionally solidified nickel-based single-crystal superalloy to investigate the effect of Y addition on the oxidation behavior.
[001]-oriented single-crystal bars were prepared using an industrial Bridgeman directional solidification furnace, in which the Y level was changed using different amounts of Ni-Y intermediate alloy.Singlecrystal bars were subjected to standard heat treatment.The Inductively Coupled Plasma-Atomic Emission Spectrometry results show that the final pick-up amounts of Y were 70, 190, 230, and 660 ppm, and the corresponding specimens were labeled as 0Y, 70Y, 190Y, 230Y, and 660Y, respectively.The specimens for the isothermal oxidation experiment were cut into pieces measuring 30 mm × 10 mm × 1.5 mm from these singlecrystal bars via electron spark machining, and then polished using a series of SiC paper up to 800-grit.The specimens were cleaned with alcohol, followed by dry blowing.
The tested specimens were heated in dry and static air at 1100 °C for 100 h, and during the testing, they were weighed at intervals of 25 h.An X-ray diffractometer was used to characterize the oxide scales on the sample surface.The microstructures of the oxide scales and element distributionmaps were analyzed using a fieldemission scanning electron microscope equipped with an energy dispersive spectrometer.The surface roughness was determined using a laser-scanning confocal microscope.
During the isothermal oxidation experiment, the weight of the porcelain crucible was recorded before oxidation (m 1 ) and after every oxidation interval (m 2 ).The mass gain per unit of oxidation (G + ) and the average oxidation rate ( K + ) were calculated as follows: As shown in Figure 1(a), among all the samples, sample 0Y indicated the maximum mass gain, and the value after oxidation for 100 h was 12.6 g/m 2 .Compared with sample 0Y, samples 70Y, 190Y, 230Y, and 660Y indicated a monotonically decreasing mass gain of 5.3, 5.5, 5.2, and 4.1 g/m 2 , respectively.Sample 660Y indicated the least mass gain after 100 h of oxidation, which was approximately one-third that of sample 0Y.Additionally, Figure 1(a) clearly shows that, compared with 0Y, at the earlier stage, the Y-doped alloy approached a steady state of mass gain.During the first 25 h of oxidation, the mass gain between each Y-doped specimen did not differ significantly.Compared with the Y-doped alloy, sample 0Y exhibited a high mass gain even after 25 h of testing.Based on the regression analysis results of the isothermal oxidation test data, the oxidation tendency of sample 0Y obeyed the parabolic law, and the oxidation behavior of the Y-doped samples reflected the cubic law.The change in the oxidation rules implied that the Y-doped alloy had a higher oxidation resistance than the 0Y alloy.The square of the mass gain of sample 0Y vs. the test time is plotted in Figure 1(b), and the parabolic oxidation rate constant, K p , was 1.785 g 2 /m 4 h. Figure 1(c) shows the cubic mass gain vs. test time; the K p values of samples 70Y, 190Y, 230Y, and 660Y were 1.480, 1.605, 1.445, and 0.656 g 3 /m 6 h, respectively.The K p value of sample 70Y was approximately 2.5 times that of sample 660Y, which implied that 660 ppm was the best addition amount for Y to increase the oxidation resistance of the alloy in this study.

Oxidation Kinetics
The phase substituents of the samples after oxidation are shown in Figure 2

Oxidation Scale
Figure 4(a-e) show the surface morphologies of the different Y-doped alloys after oxidation at 1100 °C for 100 h.Cubic oxides appeared on the surface of sample 0Y.For sample 70Y, the cubic morphology of the oxidation particles on the alloy surface disappeared, and the oxidation particles were ridged and arranged compactly.As the Y content increased, the ridged particles began to agglomerate into a cauliflower-like bulge.When the agglomerates grew continuously, the distance between each agglomerate became smaller.Based on the EDS analysis (Table 1), the main composition of the oxide scale at the surface of all the samples was Al 2 O 3 , whereas a small amount of Ni(Al,Cr) 2 O 4 was indicated as well.Figure 5 shows the height histogram of the oxide surface, which can intuitively reflect the roughness of the sample surface; the higher the number of particles detected, the rougher is the sample surface.
It was discovered that, compared with 0Y, the roughness of 70Y decreased.For all Y-added alloys, the roughness of the oxide surface first increased and then decreased.The variation in the roughness resulting from the height histogram was consistent with the change tendency of the surface morphologies shown in Figure 4.As more Y was added, the oxide on the surface of the samples first increased and then gradually agglomerated into a uniform and continuous state.
As shown in Figure 6, all the Y-doped specimens had a continuous Al 2 O 3 oxide scale, indicating better oxidation resistance than the alloy without Y addition.This enhancement is similarly reflected in the low dropping weight of the oxide scale compared with that of 0Y (as shown in Figure 7).
Figure 8 shows the oxide scale structure of sample 230Y, which represents the typical oxide scale structure of the Ni-based alloy investigated in this study.Based on the EDS and XRD analysis results (shown in Figure 2), it exhibited a double-layer structure; the outer and inner layers were composed of Ni(Al,Cr) 2 O 4 and Al 2 O 3 , respectively.
The difference in oxide scale between samples 0Y and 660Y is depicted clearly shown in Figure 9(a) and (b).A needle-like phase appeared near the oxide scale in sample 0Y; combined with the chemical composition of Points 3 and 4 denoted in Figure 9(c) and Table 2, these needlelike phases were identified as AlN.In sample 660Y, the  Based on the cross-section of the oxide scale, a naillike internal oxidation pattern was observed within the Y-doped specimens, as shown in Figure 10.Meanwhile, the EDS analysis results in Table 3 show that, from the inside to outside, the "nail" was composed of Al 2 O 3 , a mixture of refractory oxides/carbides, and an oxide containing Y. Furthermore, it was clear that the oxide containing Y was located at the bottom of the "nail".
Table 4 shows the change in the oxide scale thickness, spallation amount of oxide scale, and oxidation mass gain for different samples.Sample 660Y indicated the smallest oxide scale thickness, i.e., approximately one-half that of sample 0Y.Meanwhile, sample 660Y indicated the least spallation amount and oxidation mass gain among the five types of samples.Therefore, it can be concluded that the 660Y alloy samples exhibited the best oxidation resistance among all the Y-doped samples.

Discussion
The initial oxidized products were NiO and spinel (Ni(Cr,Al) 2 O 4 ), in which NiO first formed during the oxidation process but eventually disappeared as the intermediate product because it reacted with Al 2 O 3 and Cr 2 O 3 to form Ni(Cr,Al) 2 O 4 [17,18].This can be described by the following expression [19]: Owing to the addition of Y, oxide scale spallation almost disappeared, which resulted in a significant increase in the oxidation resistance of the alloy.
According to the classic Wagner theory, the oxidation process is orientational and determined by the direction of ion diffusion [20,21].Subsequently, based on a study pertaining to NiCrAls and FeCrAls, oxide scale formation involves both outward aluminum diffusion and inward oxygen diffusion, and the pathway of the oxidizing process is along the grain boundaries [22,23].
An analysis of the oxidation kinetic curves show that the oxidation resistance increased with the addition of Y, and the addition of 660 ppm yielded the best oxidation resistance.However, considering the possible adverse effect of excessive amounts of Y on the comprehensive properties of Ni-based single-crystal superalloys [24,25], Y amounts exceeding 660 ppm were not investigated in this study.
The segregation of Y ions would occur at the matrix/ oxide scale interface because of the high chemical activity of element Y [26].Because of the higher standard Gibbs free energy (ΔG θ ) of Y 2 O 3 , the affinity of Y with O was stronger than that of C with Al [24,27]; hence, Y 2 O 3 formed at the surface prior to the Al 2 O 3 layer.It was demonstrated that Y 2 O 3 is suitable as the core particle for Al 2 O 3 formation and hence promotes the formation of an oxide scale [28].The continuous oxide scale formed  [13,20,29].In this study, the oxide particles containing Y, to some extent, should exhibit functions similar to Y 2 O 3 , as shown in Figure 11.In this case, the diffusion behavior was weakened owing to the obstruction of the oxide containing Y at the diffusion pathway.Consequently, the Al 2 O 3 thickness reduced, but no evidence of significant segregation between Al 3+ and O 2− was discovered.This might cause both the spinel layer and Al 2 O 3 layer of the 660Y sample to be thinner than those of sample 0Y.
For sample 0Y, needle-like aluminum nitrides appeared at the near-surface region of the oxidation scale in the matrix.Owing to the volume increase, which was caused by the formation of the oxides (Al 2 O 3 28% increase) [30], cracks and spallation occurred [31].Cracks were observed across the oxide scales, and N further penetrated through the cracks to form AlN. However, owing to Y doping, the aluminum nitrides disappeared because of the greater oxide scale adherence and fewer oxidation products, as well as a lower volume increase below the oxide scale, and the cracks disappeared [32].By contrast, as reported in other published studies [33,34], this phenomenon is associated with the purification of the alloy after the addition of Y.

Conclusions
Based on the investigation of the oxidation kinetics and surface/cross-sectional morphologies of the oxidation scale, some key points can be summarized as follows:     Consequently, the oxide volume decreased, and cracks disappeared in the oxide scale.
. The oxidation products primarily comprised Al 2 O 3, Ni(Cr,Al) 2 O 4 , and TaO 2 .The Al 2 O 3 and TaO 2 peaks between the different samples did not change significantly; however, the peak strengths of the Ni(Cr,Al) 2 O 4 phases weakened in the Y-doped samples, signifying a possible decrease in the amount of formed Ni(Cr,Al) 2 O 4 compared with the 0Y sample.The EDS mapping results of the 0Y and 70Y oxide scales proved this hypothesis.A Ni(Cr,Al) 2 O 4 layer is evident in Figure 3(a), whereas such a layer is not shown in Figure 3(b).

Figure 1 Figure 2
Figure 1 Oxidation kinetics of isothermal oxidation at 1100 °C for 100 h for Ni-based single-crystal superalloy samples with different amounts of Y addition: (a) Change in oxidation mass gain, (b) (Oxidation mass gain) 2 vs. test time for sample 0Y, (c) (Oxidation mass gain) 3 vs.test time for Y-doped samples

( 1 )
To improve the oxidation resistance of Ni-based superalloys, doping with Y effectively decreased the oxidation rate, and after oxidation at 1100 °C for 100 h, the mass gain of the 70 ppm Y alloy decreased from 12.6 to 5.3 g/m 2 .Furthermore, the 660Y alloy exhibited the lowest mass gain, spallation amount, and oxide scale thickness.Therefore, in this study, theoptimal Y content was 660 ppm.(2) The alloy investigated in this study was Al 2 O 3 formation alloy, which has a typical doublelayer structure of the oxide scale, where one layer contained Al 2 O 3 and Ni(Al,Cr) 2 O 4 spinel primarily, whereas the other layer contained oxide/carbide particles of refractory elements.(3) Y addition into the Ni-based single-crystal superalloy resulted in a Y-containing adhesive oxide scale, which decreased the thickness of both the outermost Ni(Al,Cr) 2 O 4 layer and inner Al 2 O 3 layer.

Figure 7 Figure 8
Figure 7 Spallation amount of oxide scale with different amounts of Y addition

Figure 10
Figure 10 Typical section morphologies of oxide scaleand element area distribution maps of nail-like internal oxide in 660Y sample by the composition of the alloy.It is widely recognized that the addition of RE elements can improve the oxidation resistance of Ni-based superalloys.The Y-doped alloys showed clear differences in the 0Y sample.Compared with the 0Y sample, Y doping reduced the oxidation rate and the total mass gain after 100 h of isothermal oxidation.

Table 1
Chemical composition of points shown in Figure 4 (at.%) rapidly, thereby inhibiting the formation of Ni(Al,Cr) 2 O 4 .It appeared that he formation of Y 2 O 3 affected the Al 3+ / O 2− diffusion behavior at the grain boundaries

Table 2
Chemical composition of points shown in Figure9(at.%)

Table 3
Chemical composition of points shown in Figure10(at.%)

Table 4
Characteristics of oxide scale for different amounts of Y addition Figure 11 Oxidation mechanism of 0Y and Y-doped samples