E ﬀ ects of Al Sputtering Film on the Oxidation Behavior of NiCrAlY Bondcoat

: In this study, the oxidation behavior of Al coated NiCrAlY bondcoat is investigated. It is known that many methods are applied to improve the lifetime of bondcoat in thermal barrier coatings. Herein, the Al sputtering method is selected to increase the Al content, which does not change the structure of bondcoat. Thin Al ﬁlm of ~2 µ m was sputtered on the surface of bondcoat, which improved the oxidation resistance of NiCrAlY bondcoat. Experimental results showed that, after oxidation for 200 h at 1200 ◦ C, the formation of a dense and continuous α -Al 2 O 3 / Cr 2 O 3 multilayer was observed on the Al coated bondcoat surface. In contrast, a mixed oxides (NiO, Cr 2 O 3 and spinel oxides) layer formed on the surface of the as-sprayed bondcoat samples. Results of the cyclic oxidation at 1050 ◦ C within 204 h indicated that the Al sputtering method can improve the oxidation resistance of bondcoat. This study o ﬀ ers a potential way to prolong the lifetime of thermal barrier coatings and provides analysis of the oxidation mechanism.


Introduction
Thermal barrier coatings (TBCs) are widely employed to thermally protect gas turbine engines. The detailed functions of TBCs include the improvement of the efficiency, durability, and properties in high temperature operation environments [1][2][3][4]. TBCs generally comprise three layers, including a ceramic topcoat, an interlayer bondcoat, and a superalloy substrate. It has been proven that TBCs are used in the hot sections as an important component in protecting the gas turbines from oxidation, thermal fatigue, corrosion, wear, and erosion [5][6][7][8].
The load-carrying substrate is typically the Ni-based superalloy. The state of art topcoat material is 6-8 wt.% yttria-stabilized zirconia (YSZ) ceramic coating, which provides the advantages of low thermal conductivity, high thermal-expansion coefficient, and high fracture toughness [9][10][11]. YSZ topcoat can be deposited by mainly two methods: electron beam physical vapor deposition (EB-PVD) and air plasma spray method (APS) [12][13][14]. Interlayer bondcoat is between bond topcoat and substrate, which protects the superalloy substrate from being oxidized in severe environments and serves the critical role of providing adhesion between the substrate and the topcoat [15,16]. Two kinds of bondcoat are widely used in the TBCs system [1,2,6]: one is overlay bondcoat such as MCrAlY (M = Ni and/or

Al Sputtering Process
Direct current magnetron sputtering technique is used to produce the Al film, which was coated on the surface of NiCrAlY bondcoat. The details of the sputtering conditions are shown in Table 3. The thickness of Al film coated on the surface of NiCrAlY bondcoat was~2 µm, which was estimated from processing parameters.

Oxidation
Isothermal oxidation test was conducted at 1200 • C in static air at atmospheric pressure, the as-sprayed and Al coated NiCrAlY samples were heated for 1 h, 5 h, and 200 h. Thermal cycling oxidation test was performed at 1050 • C for 204 h. By measuring weight gains of samples, the oxidation behavior of as-sprayed and Al coated NiCrAlY samples was evaluated. One thermal cycle of 12 h consisted of 15 min ramp-up, 11 h isothermal soak at 1050 • C, and 45 min cool-down to ambient temperature (~25 • C). The weight measurements were taken after each cooling cycle. The precision of the measuring balance was ±0.1 mg.
The oxidation kinetics of bondcoats can be quantified by the value of weight change per unit area at a certain high temperature [18,22,24,32,33,35]. The pattern of kinetics curve generally shows a parabolic weight gain behavior in the main process of oxidation, followed by a weight loss ending which corresponds to a failure period in TBCs system.

Sample Characterization
Scanning electron microscopy (SEM, JSM-7000F, Tokyo, Japan) in second electron mode and back scattered electron mode was used to observe the surface and cross-sectional morphologies of as-sprayed and Al coated NiCrAlY samples. X-ray spectroscopy (EDX) was used to examine Coatings 2020, 10, 376 4 of 14 the chemical composition of the samples. The mean area and length of TGO scale were measured using ImageJ software. A series of SEM images were obtained for each sample with a magnification of 1000×. The resolution of these images was 600 dpi. Moreover, twenty images were randomly chosen from the cross-section of each sample. The X-ray diffraction (XRD, Philips X'pert, Amsterdam, Netherlands, Cu Kα radiation, 45 kV, 40 mA) method was used to identify the phases of oxidized samples. All XRD patterns were recorded by running X-ray diffractometer the condition of scan step = 0.02 • and 2θ = 20-80 • .

Microstructure of as-Sprayed and Al Coated NiCrAlY Bondcoat Samples
In Figure 1, the cross-section of as-sprayed NiCrAlY bondcoat on the Inconel 738 superalloy substrate is shown. It can be observed that the total thickness of NiCrAlY bondcoat is~150 µm, which contact the Ni-based substrate tightly. In addition, microcracks and voids are produced during the spray process, showing the non-fully dense morphology. Moreover, the air plasma sprayed process also introduces the formation of Al 2 O 3 oxides. These segmented Al 2 O 3 veins are unevenly dispersed in the NiCrAlY bondcoat [18,36].
The oxidation kinetics of bondcoats can be quantified by the value of weight change per unit area at a certain high temperature [18,22,24,32,33,35]. The pattern of kinetics curve generally shows a parabolic weight gain behavior in the main process of oxidation, followed by a weight loss ending which corresponds to a failure period in TBCs system.

Sample Characterization
Scanning electron microscopy (SEM, JSM-7000F, Tokyo, Japan) in second electron mode and back scattered electron mode was used to observe the surface and cross-sectional morphologies of as-sprayed and Al coated NiCrAlY samples. X-ray spectroscopy (EDX) was used to examine the chemical composition of the samples. The mean area and length of TGO scale were measured using ImageJ software. A series of SEM images were obtained for each sample with a magnification of 1000×. The resolution of these images was 600 dpi. Moreover, twenty images were randomly chosen from the cross-section of each sample. The X-ray diffraction (XRD, Philips X'pert, Amsterdam, Netherlands, Cu Kα radiation, 45 kV, 40 mA) method was used to identify the phases of oxidized samples. All XRD patterns were recorded by running X-ray diffractometer the condition of scan step = 0.02° and 2θ = 20-80°.

Microstructure of as-Sprayed and Al Coated NiCrAlY Bondcoat Samples
In Figure 1, the cross-section of as-sprayed NiCrAlY bondcoat on the Inconel 738 superalloy substrate is shown. It can be observed that the total thickness of NiCrAlY bondcoat is ~150 µm, which contact the Ni-based substrate tightly. In addition, microcracks and voids are produced during the spray process, showing the non-fully dense morphology. Moreover, the air plasma sprayed process also introduces the formation of Al2O3 oxides. These segmented Al2O3 veins are unevenly dispersed in the NiCrAlY bondcoat [18,36].   Figure 2c,d, respectively. The Al peak shown in Figure 2d indicates corresponded Al layer between bondcoat and Ni-plating in Figure 2b.
From Figure 3, it can be seen that Al film was successfully deposited on the surface of NiCrAlY bondcoat samples. Figure 3a shows the correctional morphology at higher magnification of Al/NiCrAlY interface as compared to Figure 2b  of Ni and Al element are presented in Figure 2c,d, respectively. The Al peak shown in Figure 2d indicates corresponded Al layer between bondcoat and Ni-plating in Figure 2b. From Figure 3, it can be seen that Al film was successfully deposited on the surface of NiCrAlY bondcoat samples. Figure 3a shows the correctional morphology at higher magnification of Al/NiCrAlY interface as compared to Figure 2b Figure 4 presents the XRD spectra of as-sprayed and Al coated NiCrAlY bondcoat samples after the isothermal oxidation for 1 h at 1200 °C. The XRD results show that the Cr2O3 phase is found on both the surface of as-sprayed and Al coated samples (Figure 4a,b). In addition, both θ-Al2O3 and α-Al2O3 phase is observed in as-sprayed bondcoat samples, while only α-Al2O3 phase is investigated   Figure 4 presents the XRD spectra of as-sprayed and Al coated NiCrAlY bondcoat samples after the isothermal oxidation for 1 h at 1200 • C. The XRD results show that the Cr 2 O 3 phase is found on both the surface of as-sprayed and Al coated samples (Figure 4a,b). In addition, both θ-Al 2 O 3 and α-Al 2 O 3 phase is observed in as-sprayed bondcoat samples, while only α-Al 2 O 3 phase is investigated in Al coated bondcoat samples. It is noted that the intensity of the α-Al 2 O 3 peaks of as-sprayed sample is smaller than that of Al coated sample. From Figure 4, it can be concluded that θ-Al 2 O 3 phase is only detected in the as-sprayed sample and the intensity of α-Al 2 O 3 phase is stronger in the Al coated sample.  The surface morphology of the as-sprayed and Al coated NiCrAlY bondcoat samples after isothermal oxidation for 1 h at 1200 °C is shown in Figure 5a,b, respectively. Some protrudes, pores, and pits are observed in the fine oxide particles on the surface of samples. EDX spectra were obtained from the whole surface of samples. Results in Figure 5c,d show that strong Al and Cr peaks can be found, indicating that the oxides are mainly Al and Cr oxides. Ni peaks correspond to the Ni oxides. From Figure 4, it is confirmed that both θ-Al2O3 and α-Al2O3 phase are observed in as-sprayed sample, while only α-Al2O3 phase is found in Al coated sample. From Figure 5, the oxides are identified/speculated from their typical morphology reported in earlier works and EDX data. In Figure 5a, it can be seen that whisker like/bladelike oxides θ-Al2O3 form dominantly on the surface of as-sprayed bondcoat samples [25,26,33] with minor web like or dense equiaxed structure oxides α-Al2O3 [25,27], while in Figure 5b, only web like or dense equiaxed structure oxides α-Al2O3 can be observed. EDX spectra in Figure 5c,d indicate that the Al content may increase in the as-sprayed bondcoat compared with that of Al coated bondcoat. The surface morphology of the as-sprayed and Al coated NiCrAlY bondcoat samples after isothermal oxidation for 1 h at 1200 • C is shown in Figure 5a,b, respectively. Some protrudes, pores, and pits are observed in the fine oxide particles on the surface of samples. EDX spectra were obtained from the whole surface of samples. Results in Figure 5c,d show that strong Al and Cr peaks can be found, indicating that the oxides are mainly Al and Cr oxides. Ni peaks correspond to the Ni oxides. From Figure 4, it is confirmed that both θ-Al 2 O 3 and α-Al 2 O 3 phase are observed in as-sprayed sample, while only α-Al 2 O 3 phase is found in Al coated sample. From Figure 5, the oxides are identified/speculated from their typical morphology reported in earlier works and EDX data. In Figure 5a, it can be seen that whisker like/bladelike oxides θ-Al 2 O 3 form dominantly on the surface of as-sprayed bondcoat samples [25,26,33] with minor web like or dense equiaxed structure oxides α-Al 2 O 3 [25,27], while in Figure 5b, only web like or dense equiaxed structure oxides α-Al 2 O 3 can be observed. EDX spectra in Figure 5c,d indicate that the Al content may increase in the as-sprayed bondcoat compared with that of Al coated bondcoat.

NiCrAlY Bondcoat Samples after Isothermal Oxidation for 1 h at 1200 • C
Coatings 2020, 10, 376 7 of 14 oxides are identified/speculated from their typical morphology reported in earlier works and EDX data. In Figure 5a, it can be seen that whisker like/bladelike oxides θ-Al2O3 form dominantly on the surface of as-sprayed bondcoat samples [25,26,33] with minor web like or dense equiaxed structure oxides α-Al2O3 [25,27], while in Figure 5b, only web like or dense equiaxed structure oxides α-Al2O3 can be observed. EDX spectra in Figure 5c,d indicate that the Al content may increase in the as-sprayed bondcoat compared with that of Al coated bondcoat.   The cross-sectional morphology of as-sprayed and Al sputtered NiCrAlY bondcoat samples are shown in Figure 7a,b, respectively. After isothermal oxidized at 1200 °C for 200 h, TGO forms on the surface of bondcoat in both samples. In Figure 7a,b, four sites A, B, C, and D are marked on the TGOs in the SEM images. These sites are the areas used to measure the compositions of Al, Cr, and Ni by EDX showing in Table 4. In addition, Figure 7c-f presents the EDX spectra of corresponding areas. The cross-sectional morphology of as-sprayed and Al sputtered NiCrAlY bondcoat samples are shown in Figure 7a,b, respectively. After isothermal oxidized at 1200 • C for 200 h, TGO forms on the surface of bondcoat in both samples. In Figure 7a,b, four sites A, B, C, and D are marked on the TGOs in the SEM images. These sites are the areas used to measure the compositions of Al, Cr, and Ni by EDX showing in Table 4. In addition, Figure 7c-f presents the EDX spectra of corresponding areas. The cross-sectional morphology of as-sprayed and Al sputtered NiCrAlY bondcoat samples are shown in Figure 7a,b, respectively. After isothermal oxidized at 1200 °C for 200 h, TGO forms on the surface of bondcoat in both samples. In Figure 7a,b, four sites A, B, C, and D are marked on the TGOs in the SEM images. These sites are the areas used to measure the compositions of Al, Cr, and Ni by EDX showing in Table 4. In addition, Figure 7c-f presents the EDX spectra of corresponding areas.  In both samples, the TGO scales consist of a dark inner layer and a bright outer layer. Besides, some pores and voids are observed in the outer layer in both bondcoat samples. EDX data show that Al content is 80 and 85 wt.% at sites A and C, respectively. This indicates that the inner layers of oxides in both samples are α-Al 2 O 3 layers. On the other hand, Cr content at site B and D is 60 and 72 wt.%, respectively, which means that the outer layers of oxides consist of Cr 2 O 3 + Ni(Al,Cr) 2 O 4 mixed oxides [18,22,24,27].
Since the TGOs are non-uniformly distributed on the surface of both bondcoat samples, the thickness of the TGO of as-sprayed bondcoat sample is in the range of~6.3 to~20.2 µm, while the thickness of the TGO of Al coated bondcoat sample is in the range of~3.6 to~15.3 µm. Moreover, the average thickness of the TGO of as-sprayed sample is~17.2 µm, in which the average thickness of α-Al 2 O 3 layer is <2.0 µm. For Al coated bondcoat sample, the average thickness is~7.4 µm, while the average thickness of α-Al 2 O 3 layer is~3.3 µm. Figure 8 shows the average thickness of oxides as a function of oxidation time for as sprayed sample (Figure 8a) and Al coated sample (Figure 8b). For both samples, the thickness of oxides increases with prolonged oxidation time. At both 5 h and 200 h, the thickness of oxides in the as-sprayed sample is larger than that of oxides in Al coated sample. α-Al2O3 layer is <2.0 µm. For Al coated bondcoat sample, the average thickness is ~7.4 µm, while the average thickness of α-Al2O3 layer is ~3.3 µm. Figure 8 shows the average thickness of oxides as a function of oxidation time for as sprayed sample (Figure 8a) and Al coated sample (Figure 8b). For both samples, the thickness of oxides increases with prolonged oxidation time. At both 5 h and 200 h, the thickness of oxides in the as-sprayed sample is larger than that of oxides in Al coated sample.

Thermal Cycling Oxidation Behavior of NiCrAlY Bondcoat Samples
Cyclic oxidation was conducted at 1050 °C. NiCrAlY bondcoat samples before and after Al deposition treatment were studied as a function of oxidation time. In Figure 9, the curve 9a and 9b

Thermal Cycling Oxidation Behavior of NiCrAlY Bondcoat Samples
Cyclic oxidation was conducted at 1050 • C. NiCrAlY bondcoat samples before and after Al deposition treatment were studied as a function of oxidation time. In Figure 9, the curve 9a and 9b show the weight gain per unit area as a function of oxidation time (∆M/S vs. time) for as-sprayed and Al coated NiCrAlY bondcoat samples, respectively. It can be found that the ∆M/S of Al coated bondcoat samples is lower than that of as-sprayed bondcoat samples. In addition, the oxidation rate of the samples was calculated by ratioing the weight gain per unit area at 204 h to the longest oxidation time during the experiment, i.e., 204 h. The oxidation rate of as-sprayed and Al coated samples are 3.4 × 10 −4 and 1.7 × 10 −4 mg·cm −2 ·h −1 respectively. This result is in agreement with the results obtained in isothermal oxidation.   Figure 10a,b shows the schematics of oxidation process of as-sprayed and Al coated NiCrAlY bondcoat samples, respectively. Figure 10 illustrates the effect of Al sputtering treatment on the oxidation behavior of NiCrAlY bondcoat.  Figure 10a,b shows the schematics of oxidation process of as-sprayed and Al coated NiCrAlY bondcoat samples, respectively. Figure 10 illustrates the effect of Al sputtering treatment on the oxidation behavior of NiCrAlY bondcoat.

Discussion
(b) Al coated sample. The cyclic oxidation was carried out at 1050 °C for 204 h. Figure 10a,b shows the schematics of oxidation process of as-sprayed and Al coated NiCrAlY bondcoat samples, respectively. Figure 10 illustrates the effect of Al sputtering treatment on the oxidation behavior of NiCrAlY bondcoat. In 0 h state, the morphologies of as-sprayed and Al coated NiCrAlY bondcoat samples are similar. EDX line scan of Figure 2 and EDX mapping of Figure 3 shows that the Al film is successfully deposited on the surface of NiCrAlY bondcoat. It was reported that α-Al2O3 is more In 0 h state, the morphologies of as-sprayed and Al coated NiCrAlY bondcoat samples are similar. EDX line scan of Figure 2 and EDX mapping of Figure 3 shows that the Al film is successfully deposited on the surface of NiCrAlY bondcoat. It was reported that α-Al 2 O 3 is more stable than θ-Al 2 O 3 [25]. Therefore, with higher temperature, larger Al content, or longer oxidation time, either α-Al 2 O 3 forms without θ-Al 2 O 3 or the θ-Al 2 O 3 phrase transform to α-Al 2 O 3 phase [25,26]. In addition, after the formation of continuous α-Al 2 O 3 layer during oxidation at high temperature, the oxidation process of bondcoat is dramatically slowing down [24][25][26][27]. This is due to the fact that α-Al 2 O 3 has slower growth rate compared to other oxides such as Cr 2 O 3 and NiO [32,33]. Thus, higher Al content in NiCrAlY bondcoat are reported to prevent the formation of metastable aluminas and mixed oxides [24,27,32,33]. This leads to the longer lifetime of bondcoat samples in severe environments [37][38][39].

Discussion
The Cr 2 O 3 phase is observed and detected on both as-sprayed and Al coated NiCrAlY bondcoat samples, after isothermal oxidation at 1200 • C for 1 h (Figure 10a,b). The formation of Cr 2 O 3 phase is due to the high inter-diffusion coefficient and the high concentration of Cr. In addition, the depletion of Al resulted from the oxidation during the APS process also leads to the formation of Cr 2 O 3 phase [40][41][42].
After isothermal oxidation for 1 h at 1200 • C, whisker-like θ-Al 2 O 3 formed on the surface of as-sprayed bondcoat samples, while dense equiaxed structure α-Al 2 O 3 oxides can be found on the surface of Al coated samples. Oxides of θ-Al 2 O 3 are prone to grow on the bondcoat surface, which possesses a relatively low Al content [40][41][42][43]. On the surface of Al coated samples, higher Al content is achieved. Besides, the activity of Al is higher than that of Cr and Ni. Therefore, when the Al coated bondcoat is exposed to the air, Al is in favor of diffusion into the surface area [41][42][43]. The results are consistent with the earlier study from Nijdam et al. [42,43]. In the initial stages (<1 h) of oxidation, the quantity of Al oxides rapidly increases on the NiCrAl alloys surface.
Moreover, the difference between as-sprayed bondcoat sample and Al coated sample is that as-sprayed sample has a free surface, while Al coated sample is lack of the free surface. This may affect the Al in bondcoat diffusing upward towards the surface. A possible mechanism is proposed. During the initial oxidation process, excessive levels of oxygen are available on the surface area of bondcoat samples. For the Al coated sample, the Al in Al layer reacts with the oxygen and forms a thin layer of α-Al 2 O 3 oxides, which can slow down the Al diffusion upward from bondcoat. While for as-sprayed sample, the Al in bondcoat tends to diffuse to the surface reacting with oxygen from the onset due to high activity compared with Ni and Cr in bondcoat. However, the formation time for α-Al 2 O 3 layer in as-sprayed sample is slower than that in Al coated sample.
After isothermal oxidation for 200 h at 1200 • C, both as-sprayed and Al coated NiCrAlY bondcoat samples form TGO, which consisted of two layers, i.e., the inner layer of dark α-Al 2 O 3 and the outer layer of bright mixed oxides (Cr 2 O 3 + Ni(Al,Cr) 2 O 4 ). The oxides layer distribution is in agreement with earlier studies [18,22,33,38,40,41]. From the thermodynamics point of view, the formation of α-Al 2 O 3 layer is due to the fact that Al is a probable element in the bondcoat, which is prone to react with oxygen at the initial stage of oxidation [37]. After the depletion of Al occurs, Cr 2 O 3 phase form in the bondcoat. This is because of the large content of Cr element (20 wt.%) and the high inter-diffusion coefficient characteristic of Cr [38]. Owing to the low diffusivity and solubility of oxygen in the NiCrAlY bondcoats, the diffusion of Cr is from the deep part to the surface, thus internal Cr 2 O 3 oxides are not formed in the samples [40]. Ni diffuses through the microcracks in Al 2 O 3 and Cr 2 O 3 phases, which reacts with Cr and Al on the surface of bondcoats, forming the spinel oxides. Based on this mechanism, after isothermal oxidation for 200 h, the two-layer structure of oxides was observed.
A part of the α-Al 2 O 3 layer formed in the as-sprayed samples is from the θ-Al 2 O 3 to α-Al 2 O 3 phase transformation during the isothermal oxidation process [25,26,33]. The volume expansion (12%) is accompanied with the phase transformation form θ-Al 2 O 3 to α-Al 2 O 3 [2,[25][26][27]. This leads to the formation of some pores and cracks in theα-Al 2 O 3 layer in as-sprayed samples, while a dense α-Al 2 O 3 layer is observed in the Al coated samples. For Al sputtered samples, the formation of an intact α-Al 2 O 3 layer barrier is helpful to prevent the formation of spinel oxides during the isothermal oxidation; while for as-sprayed samples, the formation of pores and cracks in the spinel oxides accelerates the oxygen attack and leads to the non-uniform TGO. Therefore, a thinner and finer TGO is observed in the Al coated samples as compared to the as-sprayed samples [44,45]. The results from isothermal oxidation in Figure 8 and cyclic oxidation in Figure 9 also demonstrate that the as-sprayed samples possess a larger weight gain per unit area than that of the Al coated samples, indicating a better oxidation resistance of Al coated samples.
One major flaw of this study is that the obtained results are not in the context of TBCs, but in the bondcoat samples without topcoat. This is related not only to the oxidation kinetics, but also to the interaction among the bondcoat fractal roughness, biaxial TGO stresses, and the elastic properties of ceramic topcoats. Over the years, studies found that investigating on the oxidation kinetics of thermal sprayed bondcoats exclusively without its interaction with topcoats does not give a clear picture of its effects on TBC lifetime [46][47][48][49][50]. Therefore, the design of experiments could be improved in the future work.

Conclusions
In this study, dense and continuous Al film (thickness =~2.0 µm) was successfully sputtered on the surface of NiCrAlY bondcoat. This Al film improved the oxidation resistance of the NiCrAlY bondcoat. Cyclic oxidation was performed at 1050 • C for 204 h. Results showed that the weight gain per unit area of Al coated bondcoat samples was smaller than that of as-sprayed bondcoat samples. Thus, better oxidation resistance was achieved by Al sputtering.
In the coming study, more complicated structural thermal barrier coatings are planned to be produced using the sputtering method, such as functional graded coatings. However, the balance between composition improvement and mechanical properties of the coatings should be carefully investigated.

Conflicts of Interest:
The authors declare no conflict of interest.