Benefits of Zr addition to oxidation resistance of a single-phase (Ni,Pt)Al coating at 1373 K

1. Introduction

Driven by the continuous demands for improving output efficiency of advanced gas turbine engines, thermal barrier coatings (TBCs) have been extensively used in the past decades and achieved significant progress [[1], [2], [3]]. Combined with advanced film cooling systems, the current state-of-the-art TBC systems can decrease surface temperature of turbine blades by more than 150 K, ensuring that gas-turbine engines operate above the melting-point of Ni-based single crystalline superalloys [4]. Due to their remarkable integrated resistance against high temperature oxidation and hot corrosion, Pt-modified NiAl coatings are widely used in TBCs as bond coat [[5], [6], [7]]. Numerous studies have reported the benefits of Pt addition to NiAl coatings, such as inhibiting the detrimental effect of S, promoting the selective oxidation and adhesion of α-Al₂O₃ scale, expanding the β-NiAl phase area, and postponing the degeneration of β to γʹ phase, etc. [[8], [9], [10], [11]].

However, the oxide scales formed on Pt-modified NiAl are prone to spallation due to the unwanted surface undulation or “rumpling” of the coating during cyclic oxidation [12,13], which causes delamination of the ceramic top coats of TBC systems. To solve this problem and further enhance the adhesion of thermally grown oxide (TGO), the addition of reactive elements (REs) such as Zr, Hf, or Y to NiAl alloys or aluminide coatings was attempted [[14], [15], [16]]. Indeed, the judicious addition of REs can present exceptional advantages, such as improved scale adhesion and reduced rate of oxide scale growth during extended thermal cycles. While several hypotheses were proposed to explain the benefits of addition of REs to NiAl alloys or aluminide coatings on their oxidation performance, mechanisms of the process remain in debate [[17], [18], [19], [20], [21]].

Zr is a typical RE, and has been recently employed to improve the oxidation and hot corrosion resistance of alloys. The research of Hong et al. [22] demonstrated that the addition of Zr into a two-phase Pt-modified aluminide coating would form oxide pegs in substrate, improving spallation resistance during cyclic oxidation. An additional effect is that Zr possesses high chemical reactivity, therefore it can easily segregate and gather S beneath the oxide scales, thus preventing S segregation [23]. The positive effect of Zr on scale adhesion was mainly attributed to its role in changing the microstructure of the oxide scale [24].

However, few papers have reported that introducing Zr along with Pt modified aluminide coatings. In our recent work [23], Zr was introduced into a single-phase (Ni,Pt)Al by co-depositing Pt-Zr using a simple electroplating approach, followed by vacuum annealing and gaseous phase aluminization treatments. Compared to electron beam physical vapor deposition [22], the above method is cost-effective and more adaptable to depositing coatings on components featuring complex shapes. In addition, the results of our previous study [23] indicated that a juditious addition of Zr to (Ni,Pt)Al coating increased Type-I hot corrosion resistance evidently.

To further clarify doping effect of Zr in single phase Pt-modified nickel aluminide coating and add knowledge in this area, the isothermal oxidation and cyclic oxidation behavior of Zr-doped β-(Ni,Pt)Al coatings was investigated and compared with those of plain NiAl and conventional (Ni,Pt)Al coatings in the present study. The influence of Zr on the early and stable stages of oxidation was analyzed in terms of oxidation kinetics and microstructure changes. Based on the results, the effects of Zr on oxidation of single-phase (Ni,Pt)Al coating were extensively discussed.

2. Experimental

2.1. Materials and coating preparation

A second-generation Ni-based single crystal superalloy (composition: 7-Cr, 7.5-Co, 5-W, 1.5-Mo, 6.5-Ta, 6.2-Al, 3-Re, traces C, and balance Ni, all in wt.%) was used as substrate. Cylindrical specimens (15 mm in diameter and 2 mm in thickness) were machined from the [001]-orientated crystal rods using a spark discharge machine. The specimens were then ground down to a 400-grit surface finish and humidly blasted using 300 mesh alumina grits. Samples were degreased in boiling 50 g/L NaOH aqueous solution for 10 min followed by ultrasonic cleaning using acetone and ethanol for 30 min each.

As a chemically inert element, Zr is stable in all acids except for hydrofluoric and nitrohydrochloric acids. Thus, a suspension consisting of pure metallic Zr particles (0.1-10 μm) and acidic Pt-coating solution was used to deposit Pt-Zr composite coating.

A magnetic stirrer was used to ensure the levitation of Zr particles at 10-20 rpm. Afterwards, a homogenization treatment was conducted on the specimens coated with Pt-Zr composite layer at 1343 K in vacuum (<6 × 10^-3 Pa) for 1 h to remove gaseous H₂ and chemically dilute the content of Pt on the surface. The low-activity gaseous phase aluminization treatment was conducted in a vertical furnace filled with Ar at 1348 K for 4.5 h. Details of the aluminization treatment can be found in literature [25]. Lastly, a single-phase Zr-doped (Ni,Pt)Al coating approximately 50 μm thick (including the underlying interdiffusion zone) was obtained. For comparison, single batch coatings of plain NiAl and conventional (Ni,Pt)Al were prepared to ensure that the deposition parameters and coating composition were the same.

2.2. Isothermal and cyclic oxidation tests

Isothermal oxidation tests were performed in a muffle furnace in open air at 1373 K for 300 h. Alumina crucibles were pre-heated at 1373 K until no weight change was observed, and the combined weights of the samples and crucibles were recorded, ensuring engagement of spalled oxide crumbs. An electronic balance (BP211D, Sartorius, Germany) with an accuracy in 0.01 mg was utilized to measure the average mass gain from three parallel samples at various oxidation intervals. A thermal gravity analyzer (TGA, Thermax 700 thermal microbalance, Thermo Cahn, USA) was adopted to evaluate early oxidation behavior of the Pt-modified aluminide coatings.

Cyclic oxidation tests were carried out in static air using an automated vertical furnace rig. Each thermal cycle consisted of maintaining the sample at 1373 K for 50 min followed by cooling it in the air for 10 min. After the given number of cycles, the specimens were removed and weighed using BP211D electronic balance, followed by the next round of cyclic oxidation tests.

2.3. Characterization

A nano-indenter (TI 950 Tribo-Indenter, Hysitron, US) was used to measure Young’s modulus and hardness values of the prepared coating samples. Prior to indentation, cross-sectional specimens underwent strict grinding and polishing treatments that ensured that the surface was flat and negligible residual stress remained. The compression rate was maintained at 0.05 nm/s and the maximum indenting depth was controlled as 350 nm. The input Poisson’s ratio during the testing was 0.30. To avoid interference from neighboring impression, the distance between two indenting spots was set for no longer than 20 μm. For each specimen, sixteen spots were selected to acquire the average hardness and Young’s modulus values.

Surface topography of the coating after different numbers of oxidation cycles was analyzed using a surface profiler (Micro XAM-100, US) at room temperature. Samples were immediately observed under a surface mapping microscope to detect their surface roughness. A light spot, 3 × 3 mm in size, was emitted onto the center of each tested samples, and their surface roughness was then obtained after analyzing the reflected light using a microscope. The surfaces of the samples were not contaminated during testing since it was a nondestructive test method.

The roughness of the coatings was characterized using three statistical parameters from the digital output of the interferometer. The Rq parameter, which can be calculated using the equation below, is proportional to the rumpling amplitude, although the exact relationship depends on the specific geometry of the undulations:

$R_{q}=\sqrt{1/n\sum_{i=1}^{n}(Z_{i}-\bar{Z})^{2}}$

where n is the number of data points, Z_i is the height of each point, and $\bar{Z}$ is the average height of the entire measured array.

X-ray diffraction (XRD, X’ Pert PRO, Cu Kα radiation at 40 kV, PANalytical, Almelo, Holland) was utilized to examine the phase constitution of the coating specimens. Samples for cross-sectional observation were deposited on a thin layer of chemically plated Ni and mounted in resin. The cross-sectional microstructures of specimens were observed using a field-emission scanning electron microscope (FE-SEM, Inspect F50, FEI Co., Hillsboro, OR, US) equipped with an energy dispersive X-ray spectrometer (EDS, X-Max, Oxford Instruments Co., Oxford, UK). A transmission electron microscope (TEM, JEOL 2100F, Japan) equipped with a tracer energy dispersive X-ray spectrometer (EDS) was used to identify the microstructure and chemical composition (in scanning transmission electron microscopy (STEM) mode) of the oxide scales.

3. Results

3.1. Isothermal oxidation behavior

As reported previously, all the three coatings of Zr-doped (Ni,Pt)Al, conventional (Ni,Pt)Al, and plain NiAl consisted of an outer zone (OZ) of β- NiAl/(Ni,Pt)Al and an underlying inter-diffusion zone (IDZ). The detailed microstructure of the as-received coatings was reported previously [23].

Fig. 1 illustrates mass gain curves and square of mass gain for different coatings during the isothermal oxidation tests. It can be observed from Fig. 1a that the weight uptakes of the coatings increased in an approximately parabolic manner with time. During the initial oxidation stage (up to 70 h), mass gain of the Zr-doped (Ni,Pt)Al coating sample was higher than that of the conventional (Ni,Pt)Al coating sample, as illustrated in the TGA curves in the inset of Fig. 1a. However, after 70 h, mass gain of the conventional (Ni,Pt)Al sample exceeded that of the Zr-doped (Ni,Pt)Al sample, and at 300 h, the final total mass gain of Zr-doped (Ni,Pt)Al was 0.82 mg cm^-2, which was approximately 41% lower than that of the Zr-free (Ni,Pt)Al coating (1.40 mg cm^-2). The greatest mass gain during the entire oxidation process was observed for the NiAl coating.

As can be seen in Fig. 1b, square of the mass gain of (Ni,Pt)Al and Zr-doped (Ni,Pt)Al matched the fitted line perfectly. The oxidation rate constants (k_p) were obtained by regressing the data illustrated in Fig. 1b. During the stable oxidation stage (after 70 h), k_p values for Zr-doped (Ni,Pt)Al and (Ni,Pt)Al were 1.57 × 10^-3 and 7.10 × 10^-3 mg² cm^-4 h^-1, respectively. The results confirmed the beneficial effect of incorporating Zr into (Ni,Pt)Al, which reduced the parabolic rate of oxide growth by a factor of 5.

As mass gain of Zr-doped (Ni,Pt)Al was higher than that of Zr-free (Ni,Pt)Al in the early oxidation stage, it was necessary to investigate the oxidation behavior of both coating samples during this stage. Fig. 2 illustrates the surface morphology of (Ni,Pt)Al and Zr-doped (Ni,Pt)Al coating samples after isothermal oxidation tests at 1373 K for 1 and 5 h, respectively. As shown in Fig. 2a, after 1 h of oxidation, surface of the (Ni,Pt)Al coating did not change compared to that of the original sample, except for some ridges [25]. Some voids and cracks could be observed. Those might be ascribed to the local detachment between the scale and coating, which could have been induced by the thermal stress during the cooling process. The scale mainly consisted of mixed oxide according to the EDS results (area 1 in Table 1). For the Zr-doped (Ni,Pt)Al sample, a needle-like morphology emerged for the oxide scale for most of the surface. This was the typical structure of transient alumina (θ-Al₂O₃) (area 3 in Table 1). Apart from the needle-like oxide structure, the rest of the surface was smooth. The EDS results revealed the presence of mixed oxide scale (area 2 in Table 1), as shown in Fig. 2b, where the magnified image at the top-right of the panel provided a better view of the specific oxide. After 5 h of oxidation, surface of the (Ni,Pt)Al coating evolved and presented a larger height difference in grain boundaries and interiors (Fig. 2c). A certain number of voids and cracks could still be observed on the surface. However, surface of the Zr-doped (Ni,Pt)Al exhibited a mixture of needle-like and clustered oxide scale, which can be more clearly observed in the magnified inset in Fig. 2d. Although both types of oxide scale on the (Ni,Pt)Al and Zr-doped (Ni,Pt)Al coatings were identified to be mainly Al₂O₃, as shown in Table 1, the disappearance of voids and cracks in Fig. 2b and d indicated that superior oxide adhesion was achieved for the Zr-doped (Ni,Pt)Al coating compared to the Zr-free one. This confirms that the addition of Zr to (Ni,Pt)Al improved the adhesion between the scale and coating.

Correspondingly, to determine the phase composition of the oxide, both coating samples were subjected to luminescence spectroscopy measurements after respective oxidation for 1 and 5 h, as shown in Fig. 3a and b. The characteristic peaks of θ-Al₂O₃ and α-Al₂O₃ can be observed at 14,402 and 14,432, and 14,575 and 14,645 cm^-1, respectively [26,27], as marked by dashed lines. The different frequencies for the luminescence of θ-Al₂O₃ and α-Al₂O₃ were presumably caused by the different Cr³⁺-O^2- distances in the crystal structures of θ-Al₂O₃ and α-Al₂O₃. It can be observed from Fig. 3a that for the (Ni,Pt)Al coating sample, besides predominant α-Al₂O₃ a small amount of θ-Al₂O₃ phase was still detected in the oxide scale after 1 h of oxidation, while after 5 h of oxidation, only α-Al₂O₃ was present. By contrast, both θ-Al₂O₃ and α-Al₂O₃ were present in the oxide scale of the Zr-doped coating after 1 or 5 h of oxidation, which indicated that Zr prevented the θ-Al₂O₃ to α-Al₂O₃ phase transformation. This phenomenon was consistent with observation by Burtin et al. [28].

Fig. 4 depicts the surface and cross-sectional morphologies of plain NiAl, conventional (Ni,Pt)Al, and Zr-doped (Ni,Pt)Al coating specimens after 300 h of oxidation. As can be clearly observed from the surface morphologies, substantial spallation of the oxide scale occurred on the surface of the NiAl coating sample (Fig. 4a), which implied a poor bonding strength between the oxide scale and NiAl coating. When Pt was incorporated, the state of the surface of the (Ni,Pt)Al coating was superior to that of NiAl, and only presented cracks on it (Fig. 4b). This demonstrates the advantages of incorporating Pt to enhancing the adherence and crack resistance of the oxide scale. Furthermore, no cracks or spallation were observed on the surface of Zr-doped (Ni,Pt)Al (Fig. 4c). These results evidently demonstrated that the synergistic combination of Pt and Zr exhibited a much more pronounced effect on improving the scale adhesion than using Pt alone.

Comparing the cross-sectional morphology of the three different coating samples, three main differences were observed. First, the oxide scales formed on the NiAl coating sample were porous and loose, while continuous and adherent Al₂O₃ scales were formed on the surfaces of the two Pt-modified coating samples, although some cracks appeared at the oxide scale/coating interface. These findings were in agreement with the above-mentioned surface morphology results. Meanwhile, the oxide scale for the Zr-doped (Ni,Pt)Al sample was the thinnest and most uniform of all three coating samples, thus demonstrating that trace amounts of Zr could noticeably reduce the growth rate of the oxide scale. Additionally, the undulation/rumpling extent of the oxide scale was the smallest for the Zr-doped coating sample. Second, the amount and size of the degraded γʹ phase of the Zr-doped (Ni,Pt)Al sample were smaller than those of the (Ni,Pt)Al and NiAl samples, indicating that the phase degradation from β to γʹ was prevented by the addition of Zr. The Al contents of the outer layer of the Zr-doped (Ni,Pt)Al and (Ni,Pt)Al samples measured by EDS were 33.26 and 19.67 at.%, respectively. Due to the loss of Al, phase transformations are inevitable during oxidation, and it would be important to preserve more β phase in the protective bond coat of Pt-modified aluminide. Third, the amount of topologically closed packed (TCP) phase precipitated in the Zr-doped (Ni,Pt)Al sample was the largest of all three coating samples, since the counterparts in the (Ni,Pt)Al and NiAl solid samples dissolved again in the γ phase.

The morphology of the oxide scales formed on the Zr-doped (Ni,Pt)Al coating sample after 300 h of oxidation and elemental mapping of its main elements (Al, O, Ta, and Zr) using STEM-EDS are illustrated in Fig. 5. It is intriguing to notice that considerable amounts of Ta and Zr segregated in the heart-shaped area (marked by the red dashed line in Fig. 5) between the grain boundaries in the oxide scale. This is in agreement with the dynamic-segregation theory introduced by Pint [29]. However, the effects of Ta and Zr on the growth of oxide and scale adhesion were unknown.

3.2. Cyclic oxidation

Fig. 6 illustrates the mass change curves of different coating specimens during the cyclic oxidation test at 1373 K. The sample coated with plain NiAl exhibited a rapid mass gain for the first 300 oxidation cycles followed by a drastic weight loss thereafter. Normally, the mass changes during the cyclic oxidation test reflect a combination between the weight increase of the oxide scale and weight loss due to the spallation of scale. The notable weight loss of NiAl after 300 oxidation cycles indicated that the oxide scale or even part of the NiAl coating began to suffer from severe spallation, which implied that the coating nearly lost its protective ability.

For the samples coated with (Ni,Pt)Al and Zr-doped (Ni,Pt)Al, the mass change curves presented a steady and slow increase during the entire cyclic oxidation test, which implied that the Pt-modified NiAl coating exhibited superior spallation resistance. Albeit, the mass changes of the two coating samples were extremely similar during the entire cyclic oxidation test, the mass of the (Ni,Pt)Al coating sample decreased twice at approximately 560 and 750 oxidation cycles, as indicated by the arrows in Fig. 6. On the other hand, the mass change of the Zr-doped (Ni,Pt)Al coating sample increased continuously, which implied that the judicious addition of Zr could help to maintain exceptional scale adhesion and superior spallation resistance during extended cyclic oxidation tests.

Fig. 7 illustrates the cross-sectional morphologies of the three coating specimens after 100 and 200 oxidation cycles. After 100 cycles, a relatively thick and porous oxide scale formed on the NiAl coating sample, as shown in Fig. 7a. The light presence of oxide on the upmost surface was confirmed as NiAl₂O₄ spinel by EDS and XRD analysis, where the underlying scale with dark appearance was α-Al₂O₃. The coating significantly deteriorated as β transformed to γʹ phase due to the outward diffusion of Al to form oxide scale, especially in the vicinity of the TGO/coating interface. In the meantime, a large amount of needle-shaped TCP phase, which was detrimental to the mechanical integrity of the coated system, precipitated beneath the IDZ because of the interdiffusion of the coating and substrate. Although similar extensive TCP phases formed in the second reaction zone after 100 oxidation cycles for the samples coated with (Ni,Pt)Al and Zr-doped (Ni,Pt)Al (Fig. 7b and c, respectively), the oxide scale on the surfaces of both samples was thinner, denser, and exclusive Al₂O₃. The extent of the deterioration from β to γʹ phase was less significant. The Zr-doped (Ni,Pt)Al coating exhibited the strongest ability to protect the substrate from high temperature oxidation, as demonstrated by the thickness of the oxide scale and deterioration extent of the coating. More importantly, the Zr-doped (Ni,Pt)Al coating specimen was less susceptible to wrinkling compared to the conventional (Ni,Pt)Al coating specimen. This will be discussed in further details below. The same trends observed in Fig. 7b and c after 100 cycles were observed in Fig. 7e, and f, after 200 oxidation cycles.

Fig. 8 shows the surface and cross-sectional morphologies of the three coating specimens after 800 cycles. As shown in Fig. 8a, the crack and spallation behavior of the oxide scale can be observed on the plain NiAl coating sample, which implies a poor adhesion of the oxide scale. A similar nodular structure consisting of ridges and valleys was formed on the surface of the samples that incorporated Pt (Fig. 8b and c). The undulation of the surface morphology and the presence of ridges and valleys on the surface was regarded as “rumpling” [13,[30], [31], [32]]. Compared with the conventional (Ni,Pt)Al coating specimen, the Zr-doped coating exhibited improved scale bonding, which was conferred by the addition of Zr since cracks only emerged on the surface of the (Ni,Pt)Al coating sample and not on the surface of the Zr-doped one.

The cross-sectional morphology illustrated in Fig. 8d also confirms the poor oxidation resistance of the plain NiAl coating since a porous and discontinuous oxide scale featuring voids, cracks, and even spallation formed on the NiAl coating sample after 800 cycles. As shown in Fig. 8e, the oxide scale that formed on the conventional (Ni,Pt)Al coating sample was continuous and intact, and only exhibited a small number of cracks. By comparison, Fig. 8f presents a superior oxide scale microstructure, where no cracks were observed and the amount of degradation from β to γʹ phase in the oxide scale was smaller. Furthermore, the undulation level of the oxide scale developed on the Zr-doped (Ni,Pt)Al coating sample was significantly smaller than that of the conventional (Ni,Pt)Al coating sample.

From the surface and cross-sectional morphologies of the oxidized coating specimens, we could determine that Pt incorporation could effectively improve the oxidation resistance of NiAl, while adding traces of Zr could reduce alumina growth rate and alleviate the undulation level of the surface during cyclic oxidation. To quantitatively and visually compare the surface rumpling behavior of the (Ni,Pt)Al and Zr-doped (Ni,Pt)Al coating specimens, a 3D topological surface profiler was used to measure the surface roughness after different numbers of cycles during the cyclic oxidation test.

Fig. 9 illustrates the 3D topological morphologies and evolution of the absolute surface roughness (R_q) of the (Ni,Pt)Al and Zr-doped (Ni,Pt)Al coating specimens after 0 (original state), 200, and 800 cycles. It can be seen from Fig. 9a and b that before oxidation, the surface of the (Ni,Pt)Al coating sample was uniformly flat, while the surface of the Zr-doped (Ni,Pt)Al one was generally flat except for some obvious sharp bugles. As a result, the R_q value of the Zr-doped (Ni,Pt)Al coating sample was slightly higher than that of normal (Ni,Pt)Al. As the number of oxidation cycles increased, the roughness of all samples exhibited a continuously increasing trend but different increasing rates were observed for each sample. The roughness of the (Ni,Pt)Al coating increased much faster than that of the Zr-doped (Ni,Pt)Al sample, and the surface of the (Ni,Pt)Al sample was coarser than that of the Zr-doped (Ni,Pt)Al one after only 200 cycles (the R_q values of the (Ni,Pt)Al and Zr-doped (Ni,Pt)Al coating samples were 1824 and 1783, respectively). The difference in roughness between the samples further increased as the oxidation time increased. Fig. 10 shows the relative roughness versus oxidation cycle to maximally avoid the impact of original roughness for the various coating samples, where R_q and R_q0 represent the surface root mean square roughness of the samples after oxidation and in the original state, respectively. As shown in Fig. 10, albeit the original surface of the (Ni,Pt)Al coating sample was finer than that of the Zr-doped (Ni,Pt)Al one, after oxidation, the roughness of the surface of the (Ni,Pt)Al coating sample was higher than that of the Zr-doped (Ni,Pt)Al one, whether the samples were subjected to 200 or 800 oxidation cycles. The results of the relative roughness evolution confirmed, once again, that incorporating Zr into (Ni,Pt)Al coatings can reduce the undulation of the coating surface during cyclic oxidation. These findings were consistent with the cross-sectional microstructures shown in Fig. 7, Fig. 8.

4. Discussion

4.1. Effects of Zr on the oxidation performance of β-(Ni,Pt)Al coating

4.1.1. Effect of Zr on the early stage of oxidation

According to Ref. [28], large ions, such as RE ions in relatively open cubic lattices could inhibit the martensitic transformation from cubic to hexagonal α-Al₂O₃. Since the structure of θ-Al₂O₃ is not as dense and compact as that of α-Al₂O₃, θ-Al₂O₃ grows faster than α-Al₂O₃, which explained why the mass gain of the Zr-doped (Ni,Pt)Al was larger than that of the conventional (Ni,Pt)Al in the early stage of oxidation.

As a matter of fact, before growing stable α-Al₂O₃ phase, aluminum oxide goes through an incubation period, where some metastable alumina phases, such as θ-Al₂O₃, form and then transform into the thermodynamically stable α-Al₂O_3. The precondition for this transformation process is that α-Al₂O₃ nuclei reach critical grain size [33]. It has been reported that impurities present in aluminum oxide would greatly affect the incubation time [34], for instance, Zr could effectively slow down the phase transformation from metastable θ-Al₂O₃ to α-Al₂O₃. The positive strain energy brought by the larger size of Zr⁴⁺ ions (r = 84.0 pm for a coordination number of six) compared to Al³⁺ ions (r = 67.5 pm for a coordination number of six) contributed to effectively increasing the stability of the parent, metastable θ-Al₂O₃ phase [35]. Therefore, as shown in Fig. 3, when comparing the Zr-free and Zr-doped (Ni,Pt)Al coatings, the metastable θ-Al₂O₃ phase was detected on the surface of Zr-doped (Ni,Pt)Al coating for longer oxidation times, which translated into the Zr-doped (Ni,Pt)Al coating being able to withstand extended periods of faster oxidation. These findings were consistent with the TGA curves in Fig. 1a. Indeed, this effect exerted by Zr could significantly affect the evolution of scale during the initial transient stage of oxidation and following steady growth behavior of the stable α-Al₂O₃.

4.1.2. Effect of Zr on the stable stage of oxidation

Obviously, the total mass gain of the Zr-doped (Ni,Pt)Al coating after isothermal exposure at 1373 K for 300 h was smaller than that of the conventional (Ni,Pt)Al. After the incorporation of Zr, the mass gain as well as the k_p values notably decreased, indicating that Zr exhibited an evidently positive effect on reducing the oxide growth rate for the single phase β-(Ni,Pt)Al coating. The k_p value of the Zr-doped (Ni,Pt)Al coating (1.57 × 10^-3 mg² cm^-4 h^-1) decreased dramatically compared to that of conventional β-(Ni,Pt)Al (7.10 × 10^-3 mg² cm^-4 h^-1), however, it was still larger than that of the Hf-doped (Ni,Pt)Al coating (3.69 × 10^-4 mg² cm^-4 h^-1) [36].

Zr atoms, unlike Pt, present a higher affinity for the Al sub-lattice sites, and only a limited amount of Zr was allowed to dissolve in the NiAl lattice (less than 0.5 at.%) [[37], [38], [39]]. However, even small amounts of Zr can significantly affect the diffusion of Al atoms at high temperature. This occurred because Zr follows a similar, although not identical, diffusion path as Al. Moreover, the diffusion activation energy, Q, of Zr is higher and its diffusion pre-factor, D₀, is lower than that of Al. According to the equation used to calculate the diffusion coefficient (D = D₀e^{$-\frac{Q}{RT}$}), Zr atoms diffuse much slower than Al atoms under the same conditions. Thus, Zr atoms act as partial path blockers for the diffusion of Al, and this blocking effect does not involve the formation of precipitates at grain boundary [40].

The slow movement of Zr compared to Al inhibits the normal outward short-circuit transport of Al along the oxide scale grain boundaries and contributes to slowing the growth of oxide. According to the classical oxidation theory, the growth rate of Al₂O₃ on an alumina forming alloy is determined by the simultaneous outward-diffusion of Al³⁺ cations and inward-diffusion of O^2- anions. Kinetically, when one reaction mechanism is constrained, it becomes the new rate-limiting step. The outward-diffusion of Al³⁺ cations along scale grain boundaries was the new rate-limiting step for the growing Al₂O₃ scale. Macroscopically, the inhibition of the outward transport of cations results in a reduction in the parabolic oxidation rate constant. Therefore, although Zr inhibits the phase transformation from θ-Al₂O₃ to α-Al₂O₃ in the early oxidation stage and leads the faster mass gain, its blocking effect results in the overall slower oxidation growth rate.

After 300 h of isothermal oxidation, the large number of precipitates and TCP phases in the NiAl and conventional (Ni,Pt)Al coatings disappeared, and this was mainly related to the phase transformation process. It is well known that the Al-rich β phase would degenerate into γʹ phase and gradually even into γ phase due to the loss of Al. Refractory elements such as W, Mo, Ta, and Re are more prone to dissolve in the γ rather than γʹ phase due to the higher solubility in the γ phase compared to γʹ. For the NiAl and conventional (Ni,Pt)Al coating, most of the β phase degraded following the β → γʹ → γ path, and a large amount of γ phase formed. As a result, white precipitates and needle-like TCP, which mainly consisted of refractory elements, re-dissolved in the γ phase. By comparison, large numbers of precipitates and needle-like TCP phases were still observed in the Zr-doped (Ni,Pt)Al coating after 300 h of oxidation. This was attributed to the phase transformation that occurred in the Zr-doped (Ni,Pt)Al coating being mainly β → γʹ instead of γʹ → γ, and can be demonstrated by the presence of a significant amount of β phase in the OZ of Zr-doped (Ni,Pt)Al. The above results also demonstrated that Zr addition could significantly decrease the rate of the β → γʹ transition.

4.2. Effects of Zr on the cyclic oxidation performance of β-(Ni,Pt)Al coating

4.2.1. Effects of Zr on the oxide scale/coating adhesion

The cyclic oxidation mass-change curves (Fig. 6) along with surface and cross-sectional morphology images (Fig. 7, Fig. 8) indicate that Zr incorporation into β-(Ni,Pt)Al coating could enhance the adhesion and spallation resistance of the oxide scale. The mass change values of both Pt-modified coatings after isothermal oxidation and cyclic oxidation for 300 h are listed in Table 2. Assuming that the oxide scale that formed on the Zr-doped (Ni,Pt)Al and conventional (Ni,Pt)Al coating sample did not spall after 300 h of isothermal oxidation, the mass of spallation can be defined as the difference between the weight of the scale after 300 h of isothermal oxidation and that after 360 cycles of oxidation (since 360 cycles of oxidation require 300 h). Obviously, the mass of spallation of the Zr-doped (Ni,Pt)Al coating (0.48 mg cm^-2) was much smaller than that of (Ni,Pt)Al (1.03 mg cm^-2), which evidently demonstrated the advantages of Zr incorporation on improving oxide scale adherence and spallation resistance.

These satisfactory results were consistent with Pint’s previously published study [29]. Pint concluded that reactive-element-induced improvements in scale adhesion were a result of the segregation of Zr at the metal-scale interface before diffusing into the oxide scale (Fig. 5). As it is well established, some voids form at the coating/oxide scale interface when oxide scale grows on the coating, and the amount of the interfacial voids depends on the surface energy of the coating/oxide scale interface. If the small separated voids grow or combine to reach critical size, they will become a cracking source. When the coating sample is subjected to repeated heating and cooling, the stress induced by the coefficient of thermal expansion (CTE) between the oxide scale and coating may lead to the cracking and spallation of the oxide scale through the cracking source.

Indigenous S in alloys or S from electroplating can lower the free energy of alloys by segregating at the coating/oxide scale interface, which would further lower the interfacial energy and reduce the activation energy for void growth [29]. Obviously, S-free alloys would not be subjected to this type of failure, but manufacturing S-free alloys would be much more expensive compared to manufacturing Ni-base single crystalline superalloys using traditional methods. In the current study, Zr was doped into the coating, and could move from the coating to the coating/oxide scale interface and then toward the oxide scale grain boundaries. As it segregated at the coating/oxide scale interface, Zr could increase the energy of the interface, thereby diminishing the detrimental effect of S segregation. Zr segregation did not eliminate void nucleation at the coating/oxide scale interface, but could prevent it from growing rapidly, therefore, good scale adhesion was maintained for a longer period. This might be the reason that a judicious addition of Zr could improve scale adhesion during extended thermal cycling.

4.2.2. Effects of Zr on surface rumpling

As shown in Fig. 9, although the original surface of the Zr-doped (Ni,Pt)Al coating sample was coarser than that of (Ni,Pt)Al, after 200 or 800 oxidation cycles, the undulation level of Zr-doped (Ni,Pt)Al was smaller than that of (Ni,Pt)Al. From the graph in Fig. 9, it can be seen that the Zr-doped (Ni,Pt)Al coating sample was less susceptible to rumpling due to the beneficial effect of the added Zr. These results raised the intriguing question of how zirconium affected rumpling. Two rumpling mechanisms, which have been previously proposed by researchers [[41], [42], [43]], will be discussed, taking into consideration the experimental results described above.

One mechanism relies on the CTE mismatch between the coating and superalloy. Boone et al. [41] revealed that the rumpling of the surface might be caused by the CTE mismatch between the substrate and coating leading to plastic deformation of aluminide coatings. Recently, a collection of CTE results for different aluminide alloys and Ni-based single crystalline superalloy became available [44]. Indeed, there is a significant CTE mismatch between β-(Ni,Pt)Al and the Ni-based single crystalline superalloy, and that might cause the phenomenon of rumpling. However, the presence of REs (Y, Hf, Zr, etc.) exhibited no detectable effect on thermal expansion [44], implying that both coatings: (Ni,Pt)Al and Zr-doped (Ni,Pt)Al should have exhibited the same rumpling amplitude. This was clearly not the case since the observed undulation level decreased obviously after Zr incorporation into (Ni,Pt)Al. Therefore, the CTE mismatch did not play a key role in causing the different rumpling behavior of the (Ni,Pt)Al and Zr-doped (Ni,Pt)Al coatings.

Comparing the cross-sectional morphologies of the (Ni,Pt)Al and Zr-doped (Ni,Pt)Al coating specimens after 100, 200, and 800 oxidation cycles (Fig. 7, Fig. 8), the β → γʹ phase degradation of the Zr-doped (Ni,Pt)Al coating was less significant than that of (Ni,Pt)Al. Therefore, the β → γʹ phase transformation, which involves substantial volumetric changes, was proposed to be the cause of the rumpling of the coatings [43]. Theoretically, there are two types of β ↔ γʹ phase transformation. One is the reversible β to γʹ phase transformation based on the Ni-Al binary phase diagram [45], which might be expected to cause plastic deformation of the coating once β degrades into β + γʹ on cooling and β + γʹ converts to β on heating, according to the β/(β + γ′) phase boundary equilibrium. Another β↔ γʹ phase transformation is attributed to the Al depletion due to the outward Al diffusion to form Al₂O₃ and interdiffusion between coating and substrate. Both these phase transformations would produce volumetric changes and generate thermal stress. However, the volumetric change and plasticity induced by the former phase transformation would be larger than that of the latter one because the former transformation is reversible. The β phase first degrades to γʹ generating an $\widetilde{2}$.5% volume decrease on heating, and then transforms back into β again, potentially generating an $\widetilde{3}$% volume increase on cooling in one cycle. Obviously, it should be noted that the area and amount of γʹ phase in the Zr-doped (Ni,Pt)Al coating after 100, 200, and 800 cycles, be it the equilibrium “primary” γʹ phase due to the Al depletion or secondary γʹ phase, are smaller than those of the conventional (Ni,Pt)Al coating (Fig. 7, Fig. 8). This implied that the volumetric change and thermal stress induced by the β ↔ γʹ phase transformation in the Zr-doped (Ni,Pt)Al coating were larger than those of the (Ni,Pt)Al coating. Thus, delaying the β to γʹ phase transformation by adding Zr was one of the reasons for the rumpling differences between the Zr-doped (Ni,Pt)Al and conventional (Ni,Pt)Al coatings.

Furthermore, Fig. 11 shows the Young’s modulus and hardness of the (Ni,Pt)Al, Zr-, and Hf-doped (Ni,Pt)Al coating samples. It can be observed that both the Zr- and Hf-doped (Ni,Pt)Al exhibit higher Young’s modulus values than (Ni,Pt)Al. The Young’s modulus reflects the deformation ability and resistance to strain of coatings. Therefore, when Zr- and Hf-doped (Ni,Pt)Al suffered from the same magnitude of stress as (Ni,Pt)Al, they were less prone to surface roughening because of their higher Young’s modulus values. These results were also consistent with our previous investigation [36]. Meanwhile, a number of experimental studies indicated that using REs, such as Hf or Zr for doping can substantially increase the creep resistance of NiAl alloys [46]. We can speculate that another role of Zr would be to decrease the creep rate of the (Ni,Pt)Al coating, thereby reducing the magnitude of rumpling. Furthermore, as shown in Fig. 8, after oxidation, Zr prefers to segregate in the grain boundaries of the oxide scale. It is well known that the creep resistance of oxide scale containing small amounts of large ions would be greater than that of undoped scale of similar grain size [[47], [48], [49]]. Thus, Zr also plays an important role in increasing the creep resistance to deformation of the coating surface. This assumption was verified using the recent mechanics model put forward by Balint and Hutchinson [50]. In their model, the oxide acted as “wet blanket’’ and suppressed changes in the surface geometry of the aluminide coating rather than inducing rumpling. Therefore, the more creep resistant the oxide scale, the smaller the rumpling amplitude.

Summarizing the aforementioned results, it can be perorated that the coating-substrate CTE mismatch mechanisms did not provide a convincing explanation for the observed decrease in rumpling. Instead, more significant attention should be paid to the β ↔ γʹ phase transformations and deformability of the coating itself. Meanwhile, it is possible that the improvement in creep resistance of the coating and TGO due to the addition of Zr led to the Zr-doped (Ni,Pt)Al coating being less susceptible to rumpling.

5. Conclusions

A novel Zr-doped (Ni,Pt)Al coating was prepared through co-deposition of a Pt-Zr composite plating and successive treatments of vacuum annealing and gaseous phase aluminization. Based on the experimental results, the following conclusions can be drawn:

(1)Zr-doped (Ni,Pt)Al coating exhibited superior isothermal oxidation and cyclic oxidation resistance compared to normal (Ni,Pt)Al and NiAl at 1373 K.

(2)In the early stage of oxidation, Zr in the coating could retard phase transformation from θ-Al₂O₃ to α-Al₂O₃ in the oxide scale.

(3)During the isothermal oxidation tests at 1373 K for 300 h, mass gain and parabolic rate constant k_p of Zr-doped (Ni,Pt)Al coating were 41% and almost one-fifth of the corresponding values for those of a Zr-free (Ni,Pt)Al coating counterpart. Meanwhile, the degradation rate from β to γ’ in coating interior decreased significantly.

(4)Zr was inclined to segregate to grain boundaries of alumina scales, which matched the enriching position of tantalum.

(5)During cyclic oxidation, the extent of scale rumpling was evidently decreased by the unique addition of Zr into the single phase β-(Ni,Pt)Al coating system because Zr incorporation postpones phase transformation from β to γ’ and increases Young’s modulus of the β-(Ni,Pt)Al coating.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (Grant Nos. 51,671,202 and 51,301,184), and the Defence Industrial Technology Development Program (Grant No. JCKY2016404C001). This Project was also sponsored by “Liaoning BaiQianWan Talents” Program.

The authors have declared that no competing interests exist.

---