Ni₃Al-base superalloy Pt-modified aluminide coating; Re-rich diffusion barrier; Cyclic oxidation; Interdiffusion; Second reaction zone

A high-performance Ni₃Al-base single crystal superalloy has been developed to serve as a turbine vane in advanced aero-engines because of its excellent high-temperature mechanical properties [1]. An effective strategy to increase the power-out efficiency and reduce the emission of greenhouse gases (GHG, such as CO₂ and NO_x), is to enhance the in-let combustion temperature in such aero-engines. The mechanical property of this superalloy that permits its afore-mentioned use is mainly attributed to second-phase strengthening by γ-Ni and solid-solution hardening by molybdenum in both the γ and γ′ phases [2,3]. Similar to other superalloys, prior to serving in high-temperature environments, the hot-section components comprising Ni₃Al-based alloys necessarily require reliable protections against oxidation and hot corrosion, both of which can be achieved by utilizing protective metal coatings. The metallic bond coats can be either MCrAlY (M = Ni and/or Co) or aluminide diffusion coatings [[4], [5], [6], [7]], wherein both types of coating series possess the capability to form low-growth-rate oxide scale (predominantly α-Al₂O₃) to resist further attack by oxygen. Generally, aluminide coatings have been used extensively to protect high-temperature structural materials against oxidation and hot corrosion owing to their excellent oxidation resistances at low capital costs. The commercial aluminide diffusion-coatings can be obtained by pack cementation, or via slurry or gaseous deposition methods [[8], [9], [10]].

However, simple aluminide coatings cannot retain their protection ability in harsher environments that require them to possess a longer lifetime [11,12]. Voids formed at the interface between the oxide scale and the coating, which further deteriorated the adhesion of the thermally grown oxide (TGO) at the surface [13,14]. Moreover, the creep resistance of the oxide scale formed on aluminide coatings is usually not high enough to hinder the interface rumpling, where considerable amounts of scale spallation can occur [15,16]. Subsequently, more aggressive oxidation attacks occurred at the spalled areas, shortening the lifetime of the coatings, especially during the cyclic exposure experienced by the coatings while in service. To overcome this inadequacy, auxiliary elements, including Hf [17], Cr [18], Si [19,20], Pt [21], and Y [22], which had been verified as positive, were doped into simple aluminide coatings, in order to increase resistances against both oxidation and hot corrosion. Among these elements, platinum was the most effective in enhancing the performance of aluminide coatings because of its featured positive effects, such as promoting uphill diffusion of Al and enlarging the β phase region [6,23,24]. In the industry, state-of-the-art platinum-modified aluminide coatings are commonly manufactured by initially electroplating the Pt layer (5-7 μm), following which vacuum-annealing is performed to dilute the Pt concentration at the surface. As a final step, aluminization using gaseous methods is carried out, including above-pack or chemical vapor deposition (CVD) [25].

In the case of a Pt-modified aluminide coating, much attention has been paid to the fact that Pt enhances oxidation resistance by reducing oxide growth and coating degradation rates, mitigating scale spallation tendencies, and promoting α-Al₂O₃ formation, etc. It is confirmed that Pt-modified aluminide coatings can reliably protect the underlying Ni-base superalloys [6,7]. However, a drawback of this coating is that in the presence of high Al content, it possesses an elevated interdiffusion potential into the substrate, especially at greater temperatures. At the same time, the outward diffusion of refractory elements (such as Mo and W) from the substrate also destroys the adhesion of the oxide scale owing to the formation of volatile oxide species at the surface [1]. It is reported that high Mo content (up to 14 wt.%) in a Ni₃Al-based intermetallic alloy ruined the integrity of the Al₂O₃ scale because of the volatilization of MoO₃ above 1100 °C [1,2]. This damage to the oxide scale limited the lifetime of the protective bond coat, because spallation and the cracking of the alumina scale significantly accelerated the oxidation mode from being stable to being rapidly aggressive [26,27].

As the Ni₃Al-based intermetallic alloys are designated to serve at elevated temperatures, the harmful formation of volatile oxide induced by the outward diffusion of refractory metals should be inhibited or retarded to facilitate successful protection by the Pt-modified aluminide coatings. To introduce a diffusion barrier (DB) is an effective and feasible way to suppress the interdiffusion between the coating and the substrate. Many studies regarding the DB have been reported, including attempts to create metallic and ceramic barriers (such as those composed of NiCrO [28], and AlN [29]). It is well understood that the stability of a DB layer is extremely essential to prohibit interdiffusion, because during exposure at high temperatures, the DB may experience microstructure changes, e.g., phase transformations, grain growth, interface reaction, and decomposition [30]. Unlike brittle DBs made of ceramic, rhenium (Re) with a melting point of 3186 °C is a suitable candidate that is capable of suppressing elemental interdiffusions because of its stable nature and good interfacial cohesion [30]. The solubility of Re in the β-NiAl phase at 1200 °C is 0.085 at.% [31]. Furthermore, M.S.A. Karunaratne et al. [32] established that Re displayed the slowest diffusion rate in nickel. All these factors are considered merits that benefit Re in serving as a DB for β phase Pt-modified nickel aluminide coatings. However, the utilization of Re as a DB for this promising coating has not been reported, despite its evident promise.

In the current work, a Re-rich diffusion barrier between the outer β-(Ni,Pt)Al and the Ni₃Al-based single crystal superalloy is specifically tailored to acquire a low-diffusion NiRePtAl coating. To assess the effect of the Re-rich DB on oxidation behavior, cyclic oxidation tests were conducted on simple nickel aluminide (NiAl), normal NiPtAl, and NiRePtAl coatings at 1100 and 1150 °C. Microstructural changes and elemental mappings of the various coatings after the oxidation test were characterized. The mechanisms responsible for the improved oxidation resistance and the mitigated extent of interdiffusion in the NiRePtAl coating are discussed with a particular emphasis on inhibiting diffusion of the refractory metal, Mo.

A Ni3Al-based single crystal superalloy (nominal composition: 6-7 wt.% Al, 12-14 wt.% Mo, 1-3 wt.% Re, balanced Ni) was used as the substrate alloy. After being cut, rectangular specimens of size 15 × 10 × 2 mm were ground with 400-grit SiC paper and humidity blasted with high using 300 mesh alumina grit under 0.3 MPa. Prior to coating deposition, the samples were degreased in a boiling NaOH aqueous solution (50 g/L) for 10 min, followed by ultrasonic cleaning in acetone and ethanol for 15 min, individually. After this step, these fresh specimens were ready to be electroplated with Ni-Re composite plating.

The NiRePtAl coating system containing a Re-rich layer was prepared by sequential electroplating and low-activity vapor phase aluminizing treatments. The deposition of Ni-Re plating was conducted on the sandblasted samples in a mixed solution that was composed of NiSO₄·6H₂O, NaCl, H2BO₃, Na2SO4, C₁₂H₂₅NaSO₄, and KReO₄. Prior to electroplating, the pH value was adjusted to be 5 using diluted H₂SO₄. The Ni-Re composite layer of 5 μm thickness was deposited on the substrate alloy.

Next, the electroplating of Pt was conducted on the Ni-Re coated samples in an alkaline solution comprising Pt(NH₃)₂(NO₂)₂ at 80 °C. A Pt film of 5 μm in thickness was deposited on the samples, followed by a regular vacuum-annealing (< 6 × 10^-3 Pa) treatment at 1050 °C for 1 h to reduce the residual stress and achieve a sufficient volume of solution of Ni, Re, and Pt.

Finally, a low-activity gaseous phase aluminization treatment was carried out in a vertical furnace filled with argon at 1050 °C for 5 h. For a detailed account of the aluminizing parameters used, please refer to our recent published works [7]. For comparison, coatings of NiAl and NiPtAl were prepared on the same substrate alloy using the same aluminization parameters. A simple NiAl coating was obtained without the electroplating of Ni-Re and Pt, while normal NiPtAl coating skipped over the deposition of Ni-Re.

Cyclic oxidation tests were conducted at 1100 °C and 1150 °C in static air, using an automated thermal cycling furnace to lift the coating specimens up and down at the appropriate times according to a timetable. In each cycle, the samples were initially exposed inside the furnace for 50 min and were next moved out to cool for 10 min. After certain cycles of oxidation, the samples were weighed to measure the net change in mass, during which procedure, neither the number of peels nor the amount of spalled crumbs of oxide scale were recorded. The change in the mass of the coatings was measured by utilizing three parallel specimens via the use of an electronic balance (BP211D, Sartorius, Germany) with a sensitivity of 10^-5 g.

The phase constitution of the coatings and the oxides were identified by X-ray diffraction (XRD, X’ Pert PRO, Cu Kα radiation at 40 kV, PANalytical, Almlo, Holland). The surface and cross-sectional morphologies and the qualitative composition of the samples were characterized by a field emission-scanning electron microscope (SEM, Inspect F50, FEI Co., Hillsboro, OR) equipped with an energy dispersive X-ray spectroscope (EDS, X-Max, Oxford Instruments Co., U.K.). A second electron (SE) mode was utilized to observe the surface morphology, while the backscattered electron (BSE) mode was chosen to characterize the cross-sectional morphology. The precise composition and elemental mapping profiles were characterized using an electron probe micro-analyzer (EPMA-1610, Shimadzu, J.P.). The roughness of the coating surface was characterized by an optical interferometer (Alpha-Step IQ, KLA-Tencor, USA). To protect the oxide scale from possible spallation, a layer of electroless nickel was deposited on the coating samples and mounted in resin before the cross-sectional observation was carried out.

Fig. 1 shows the surface and cross-sectional morphologies of the as-received NiAl, NiPtAl, and NiRePtAl coatings. Similar surface morphologies of visible grains divided by grain boundary ridges could be observed on all the three coating samples. Such morphologies were typically common in Pt-modified nickel aluminide coating prepared through gaseous aluminization techniques. The average grain sizes of the NiPtAl and NiRePtAl were relatively larger than those of the NiAl coating. As shown in Table 1, the measured roughness (R_q) values for the NiPtAl and NiRePtAl coatings before oxidation were almost identical (0.41 μm and 0.4 μm). The corresponding cross-sectional morphologies of the three as-received coatings are shown in Fig. 1(b, d, and f).

The coatings of NiAl and NiPtAl uniformly comprised two layers: the outer layer comprising the β phase, and the interdiffusion zone (IDZ) containing topologically close-packed (TCP) particles. The dark particles at the interface between the NiAl outer layer and IDZ were mainly alumina grits or kirkendall pores. However, there were very few kirkendall pores at the interfaces for NiPtAl and NiRePtAl coatings. This is because Pt can increase the outward diffusion rate of Ni from the substrate to the coating during the aluminizing process, so the consumption of Ni at the interface between the coating and the substrate can be supplemented to reduce the formation of pores [33]. In contrast, the NiRePtAl coating was composed of three layers: OZ, DB, and IDZ. The outer layer was the functional zone holding the β phase B2 structure that resists oxidation, while the intermediate DB layer was composed of Re-rich precipitates having thicknesses of about 6 μm. White TCP rods could be observed within the IDZ (∼17 μm) that were formed during vapor phase aluminization.

Fig. 2 shows the X-ray diffraction results for the three coating specimens. Although all three coatings exclusively exhibited the β-NiAl crystal structure, the angles of the characteristic β peaks for the three coatings differed from each other. A leftward-shifting behavior of the peaks (between 60-80 degrees) could be observed for the NiPtAl and NiRePtAl coatings when compared to the peaks obtained with the NiAl coating. This phenomenon indicated that the crystal lattice had been enlarged, leading to some distortion, by including exotic atoms of Pt.

3.2.1. Kinetic curves

The mass change curves of the three coatings during the cyclic oxidation test at 1100 °C are shown in Fig. 3. The NiAl coating showed a rapid mass gain in the early stage of oxidation because of the formation of metastable oxide at the surface, followed by drastic weight losses after 80 cycles. Because the lifetime of a high-temperature protective coating is often defined as the point at which the mass gain reached a negative value [7], we conclude that the lifetime of the simple NiAl coating in the current study lies between 240-250 cycles. The rapid weight losses are attributed to the spallation of the oxide scales or even to that of the coated areas.

Compared to the mass gain of the simple NiAl, that of the normal NiPtAl coating was similar in the early stage of oxidation (before 60 cycles). However, the scale adhesion was much improved in the case of the Pt-modified aluminide coating, as this coating steadily increased until 180 cycles. For the samples coated with the NiRePtAl coating, the mass change curves showed a much lower increase during the entire cyclic oxidation test, including during the initial and stable stages. The mass change curve of NiRePtAl was smooth, in which almost no spallation of the oxide scale occurred. According to the mass change results, the NiRePtAl coating notably improved the oxidation resistance for the Ni₃Al-base substrate alloy, indicating that it was competent enough to offer reliable protection.

3.2.2. Surface and cross-sectional morphologies

Fig. 4 shows the XRD patterns of the three coating specimens after cyclic oxidation at 1100 °C for various cycles. It can be seen that the oxide scale formed on simple NiAl was composed of mainly α-Al₂O₃ and NiAl₂O₄ after 300 cycles, wherein the formation of the spinel oxide was normally regarded as a failure of selective oxidation of Al. For the NiPtAl and NiRePtAl coatings, after 500 cycles the oxides were exclusively composed of α-Al₂O₃ and the coatings mainly consisted of β and γ/γ′ phases. The exclusive formation of α-Al₂O₃ on NiPtAl and NiRePtAl verified the enhancement of oxidation resistance to Pt-modification.

Fig. 5 shows the surface and cross-sectional morphologies of the coating specimens after undergoing different cycles of cyclic oxidation at 1100 °C. As shown in Fig. 5(a), a coarse surface morphology could be observed in the NiAl coating specimen, where the visible spallation of oxide scale over large areas could be observed after 300 cycles. With the addition of Pt, a similar structure comprising ridges and valleys of oxide scale was formed. At higher scales of magnification, we observed that the exfoliation of oxide scale on the NiPtAl coating after 500 cycles at 1100 °C has taken place. For the NiRePtAl coating (Fig. 5(e)), the surface was uniformly covered by integral cluster-like α-Al₂O₃ scale after 500 cycles, and furthermore, the undulation level of the oxide scale was alleviated. The absolute roughnesses (R_q) were 1.21 μm for NiPtAl and 0.77 μm for NiRePtAl after 500 cycles (see Table 1).

The cross-sectional morphologies of the three coatings after cyclic oxidation for the different cycles are shown in Fig. 5(b, d, and f). Fig. 5(b) confirms a poor oxidation resistance for simple NiAl because many internal oxide particles became connected and some of them even extended into the substrate after 300 cycles. The grey internal oxide particles could possibly be NiAl₂O₄ spinel, which was displayed in the XRD result. As observed from Fig. 5(d), the exfoliation of the oxide scale occurred at some areas, with the specific locations being in accordance with undulation peaks. The white particles depicted in the higher magnification image in Fig. 5(d) was verified as Mo-rich by EDS, indicating that the Mo atoms successfully diffused from the substrate into the coating. As shown in Fig. 5(f), the oxide scale formed on the NiRePtAl coating after 500 cycles was continuous, and the morphology at higher magnification confirmed the development of adherently thin alumina scale. To qualitatively compare the change in the compositions of the NiPtAl and NiRePtAl coatings, EDS measurements were carried out on the marked rectangles labeled 1 and 2. As listed in Table 2, the residual Al content of the NiRePtAl coating was comparatively higher than that of the NiPtAl coating. A higher amount of residual Al certainly benefits the coating, permitting it to achieve a longer service life. Another difference is that the NiPtAl coating contained a little amount of Mo after cyclic oxidation for 500 cycles, but Mo was absent in the NiRePtAl coating after being subjected to identical oxidation.

The EPMA mappings of the main elements in the three coating samples after experiencing cyclic oxidation for 300 cycles are shown in Fig. 6, Fig. 7, and 8. We observe from Fig. 6 that Al was consumed in large amounts, leaving behind only a few isolated β phase grains in the NiAl coating. From the distributions of Al, Ni, and O, we confirmed that the NiAl₂O₄ spinel particles formed not only in the oxide scale at the surface, but also inside internal oxides. The elements of Re and Mo were mainly segregated in regions of IDZ and SRZ. The slight distribution of Mo in OZ indicated that it had diffused from the substrate into the outer coating, especially in the γ′-Ni₃Al, because of the higher solubility in the γ′ phase than in the β phase.

It can be observed from Fig. 7 that the distributions of Al and Pt matched well in the NiPtAl coating after oxidation for 300 cycles. Unlike simple NiAl, the NiPtAl coating was still dominant with the β phase, which ensures a higher capacity to form the protective alumina scale and contributes to better oxidation resistance. Rather than only existing in the OZ domain, a slight amount of Pt diffused into a deeper region of the NiPtAl coating. Refractory elements of Mo and Re were mainly segregated in the IDZ and SRZ, and the emergence of Mo could be rarely detected in the outer coating.

As shown in Fig. 8, a continuously compact and thin scale of Al₂O₃ scale formed on the surface of the NiRePtAl coating after the identical oxidation test for 300 cycles. The OZ of the coating was completely composed of β phase, while a comparable concentration of Al was identified within the IDZ. Beneath the IDZ, the content of Al was much lower in the substrate. A continuous and adherent DB layer was formed at the substrate/coating interface, which was predominantly composed of Re and Cr. Meanwhile, it is interesting to notice that the Mo was enriched at the OZ/IDZ interface and was coincident with the Re-rich zone, indicating a high capability of the Re-rich to stabilize Mo.

3.3.1. Kinetic curves

Fig. 9 shows the mass change curves of NiPtAl and NiRePtAl coatings during the cyclic oxidation test at 1150 °C. In the initial oxidation stage (up to 20 cycles), both of the coatings exhibited significant mass gains, followed by steadier mass gains thereafter. After the number of cycles reached 80, an initial weight loss was apparent in the NiPtAl coating that became drastic after 140 cycles. Compared with the NiPtAl coating, the NiRePtAl coating showed relatively lower mass gains during the entire cyclic oxidation test. The change in mass increased until 180 cycles, at which point the loss in weight of the NiRePtAl coated specimens was visibly lower. According to the cyclic oxidation kinetic curves at 1150 °C, the NiRePtAl coating showed better oxidation resistance.

3.3.2. Surface and cross-sectional morphologies

Fig. 10 presents the surface and cross-sectional morphologies of the NiPtAl and NiRePtAl coatings after oxidation at 1150 °C for 200 cycles. As shown in Fig. 10(a), a spalled area (∼50 μm in width) could be observed on the NiPtAl coating, implying a weak bonding of the oxide scale. A more adherent oxide scale remained on the surface of the NiRePtAl coating, as shown in Fig. 10(b), where cracks and pores were observed instead of spallation in the amplified image. It can be seen from Table 1 that the absolute roughnesses were 3.52 μm for the NiPtAl and 3.13 μm for the NiRePtAl after cyclic oxidation was carried out for 200 cycles at 1150 °C. From these results, we observe that the differences of roughness were not as large as those observed at 1100 °C.

Fig. 10(c and d) demonstrate the cross-sectional BSE images of the two coatings after cyclic oxidation at 1150 °C for 200 cycles. Severe spallation of the alumina scale occurred in the case of the NiPtAl coating (see Fig. 10(c)), and a wide interdiffusion zone, about 40 μm, formed between the coating and the superalloy substrate. Beneath the IDZ, a thick SRZ of about 108 μm formed with the presence of large amounts of needle-like TCP precipitates in the substrate, wherein the needle-like TCP phases were obviously detrimental to the mechanical properties of the single crystal superalloy. According to Table 2, the mean compositions of Al, Ni, Pt, and Mo in the outer layer of the NiPtAl coating were 27.19 at.%, 65.26 at.%, 4.3 at.%, and 1.86 at.%, respectively. In contrast, an intact oxide scale was observed on the surface of the NiRePtAl coating after 200 cycles, and the Re-base DB was very stable and well-bonded between the coating and the substrate. There was almost no Mo detectable in the OZ of the NiRePtAl coating (area 5 in Table 2) and the concentration of Mo inside the DB layer was 36.22 at.% (spot 6) as obtained by EDS. It was satisfying to note that the SRZ thickness beneath the NiRePtAl coating was 81 μm, indicating a 25 % reduction from that of the NiPtAl coating.

As shown in Fig. 11, Fig. 12, the distributions of the main elements after cyclic oxidation tests had been conducted for 200 cycles at 1150 °C were profiled by EPMA. It can be observed from Fig. 11 that the NiPtAl coating consisted of two phases: one had a strong Al signal belonging to the β phase, and another had a lower Al signal corresponding to γ′-Ni₃Al. The relative volume fractions of β and γ′ phases in the NiPtAl coating were about 30 % and 70 %, individually. Mo successfully diffused from the substrate to the coating, especially within the γ′-Ni₃Al zone. In the case of NiRePtAl, after oxidation for 200 cycles at 1150 °C, it can be observed that the phase transformation from β to γ′ still occurred in the OZ, but the volume fractions of β and γ′ were 80 % and 20 %, which indicates that the degradation was effectively mitigated by the Re-rich DB layer. Similar distributions of Re, Cr, and Mo that were adjacent to the substrate/coating interface were observed, resembling the result shown in Fig. 8.

Generally, the service life of a coating is highly related to the growth rate and the spallation resistance of the oxide scale during thermal cycling exposure. With low oxidation rate and good bonding, α-Al₂O₃ is regarded as the best oxide scale offering reliable protection. However, the integrity of the alumina scale could be destroyed by the formation of gaseous Mo oxide. As shown below, the standard Gibbs free energy changes ($\Delta {{G}^{\text{ }\!\!\theta\!\!\text{ }}}$) while forming MoO₃ can be calculated [34]:

From 25 to 2000 °C, the standard Gibbs energy change because of MoO₃ production is negative, which indicates that the reaction occurs naturally during the cyclic oxidation tests. As 1100 and 1150 °C are located in the temperature range conducive to the formation of gaseous MO oxide, during cyclic oxidation, Mo is prone to combining with oxygen to form gaseous MoO₃, which decreases the scale adhesion and deteriorates the high-temperature oxidation resistance.

We infer from the experimental results that the NiRePtAl showed the best performance during the cyclic oxidation tests carried out at both 1100 °C and 1150 °C. Plain aluminide coating (NiAl) failed after oxidation for 300 cycles at 1100 °C. NiPtAl coating could not develop continuously adherent alumina scale on the surface for long, although it demonstrated an excellent capacity to grow intact α-Al₂O₃ on nickel-based single crystal superalloys [[5], [6], [7]]. In fact, one of the difficulties in utilizing the current Ni₃Al-based superalloy for high-temperature applications is the high concentration of Mo, because this refractory metal forms a volatile oxide species at high temperatures. To overcome such disadvantages, the most effective solution is to inhibit or suppress the outward diffusion of Mo.

4.1.1. Influence of Pt

As reported, the incorporation of Pt into nickel aluminide coating could induce uphill diffusion of Al and enlarge the β phase zone in the Ni-Al-Pt ternary system [35,36]. After 300 cycles of oxidation at 1100 °C, most of the NiAl coating changed to the γ′ phase (70 %), wherein Mo was enriched because of its relatively high solubility in γ′. This observation confirms that the outward diffused Mo tends to gather in the γ′-Ni₃Al phase of the coating. As Ni₃Al contained lower Al, the enrichment of Mo in this phase would be very detrimental to the oxidation resistance. By contrast, the coatings with Pt preserved β as the dominant phase after 300 cycles of oxidation at 1100 °C. Because of the extremely low solubility of Mo in the β phase [1,2], Mo was hardly identifiable in the β phase dominant OZ of the NiPtAl and NiRePtAl coatings (see Fig. 7, Fig. 8). This result shows that Pt retarded the degradation of the coating, following which it restrained the outward diffusion. Based on the cyclic oxidation behavior at 1100 °C, it appears that Pt did have a positive effect on suppressing the undesirable outward diffusion of Mo.

However, the result shown in Fig. 11 indicates that Mo successfully diffused out to the γ' phase, although Pt was uniformly distributed in the entire OZ. Therefore, the effect whereby Pt inhibits Mo diffusion is limited. The absence of Mo in OZ of NiPtAl at 1100 °C can primarily be attributed to the low solubility of Mo in the β phase. Because Pt possesses capabilities in enlarging the β phase zone and mitigating the phase degradation from β to γ', it could retard the outward diffusion of Mo to some extent. The different enriching behavior of Mo in plain and Pt-modified aluminide coatings is reasonable because the composition range of Al for pure NiAl is very narrow. Once the temperature is higher, the consumption rates of Al to form alumina scale or diffuse inwardly will be greater. Thus, the β phase will inevitably transform to the γ' phase, wherein the latter could accommodate more Mo. As a result, Mo was observed in the OZ of the NiPtAl coating after 200 cycles of oxidation at 1150 °C (Fig. 11).

4.1.2. Influence of the Re-rich layer

As shown in Fig. 11, Fig. 12, the degradation rate from β to γ′ phase in the NiRePtAl coating became slower because the Re-rich layer reduced coating degradation by inhibiting the inward-diffusion rate of Al from the coating to the substrate. The results in Fig. 6, Fig. 11 demonstrate that the outward diffusion of Mo occurs significantly from the substrate to the γ′-Ni₃Al phase. The Re-rich DB delayed the degradation from β to γ′ by reserving the Al content in the OZ, wherein the β phase could hardly accommodate Mo. This is the first contribution of the Re-based DB in the inhibition of Mo.

It is also worth noting that the concentrations of Cr and Mo in the NiRePtAl coating are much lower than those in the NiPtAl coating (see Table 2). This is because, the Re-rich layer inhibits the migration of refractory elements in the superalloy towards the coating. For the NiRePtAl coating, EPMA and EDS analyses confirmed that Mo was enriched in the Re-rich layer rather than in the outer coating, which means that Mo and Cr were trapped within the Re-rich layer. This Re-rich DB layer formed at the coating-substrate interface was confirmed as the σ-Re (Cr) phase, which effectively restrained the outward diffusion of refractory elements [30]. R.L. Mannheim et al. [37] have pointed out that the solid solution of molybdenum in Re could be up to 47.5 at.% at low temperatures, wherein the solubility increases with temperature. It is confirmed that with low formation energy, Mo atoms were inclined to combine with Re, resulting in the higher diffusion activation energy of Mo in Re [36]. Therefore, the outward diffusion of Mo was effectively inhibited in the case of the NiRePtAl coating.

4.2.1. Improving oxidation performance

Compared to those of NiAl and NiPtAl coatings, the mass gain curves for the NiRePtAl coating showed a reduced oxidation rate with the exposures at 1100 °C and 1150 °C (see in Fig. 3, Fig. 9). Furthermore, the thickness of the oxide scale formed on the coating with Re-rich DB was thinner than that of other coatings. Additionally, Fig. 5, Fig. 10 confirmed the lowest rumpling extent for the NiRePtAl coating, which matched with the roughness change results shown in Table 1. In other words, the coating of NiRePtAl exhibited a reduced oxidation rate and a lower rumpling extent simultaneously.

As is well-established, the participation of Mo by virtue of its reaction with oxygen at the surface deteriorates the integrity of the alumina scale by forming gaseous MO₃ [[1], [2], [3]]. Compared with those of the NiAl and NiPtAl coatings, the enriching levels of Mo were always lesser in the OZ of the NiRePtAl coating after the cyclic oxidation tests. On the other hand, this refractory metal was found enriched within the Re-rich layer (see Fig. 8, Fig. 12), which meant it was efficiently fixed within the DB. As less Mo was able to pass through the Re-rich layer and take part in oxidation, a more adherent and intact alumina scale could form on the surface of the NiRePtAl coating. This is the primary contribution of the Re-rich DB layer to the enhanced oxidation resistance.

It has been reported that the diffusion coefficient of Al in NiAl coating was two orders of magnitude greater than that of Al in Re-base DB [38]. The Re-rich DB layer could effectively inhibit inward diffusion of Al from passing through. Thus, the other advantage was that the Re-based diffusion barrier could reserve more Al in the outer coating, which mitigated the degradation from the high-Al β phase to the γ' [39,40]. The EPMA results (see Figs. 6,7,8,11, and 12) clearly demonstrated that the higher volume fractions of β were achieved in the NiRePtAl coating after identical cyclic exposures at both 1100 and 1150 °C. Judging by the high concentration of Al in OZ, we can confirm that the NiRePtAl coating can grow exclusive alumina scale and repair it immediately in the following oxidation test. As a result, better oxidation resistance was observed in the NiRePtAl coating.

At the same time, one of the spontaneous benefits of remaining β is to decrease the degree of surface undulation because volume shrinkage by β→γ' degradation was deemed a major cause of surface rumpling [41]. In NiRePtAl coating, the Re-rich layer successfully retarded this phase transformation by inhibiting the inward-diffusion of Al. As a result, the volume shrinkage induced by the phase degradation from β to γ' was limited, resulting in a relatively smoother TGO on the NiRePtAl coating.

In brief, the effect of the Re-rich layer on improving the oxidation performance and reducing the rumpling extent of NiRePtAl coating should be mainly attributed to the Re-rich DB that reserves more Al in forming and repairing alumina scale, and restrains the outward diffusion of Mo.

4.2.2. Inhibiting SRZ formation

During high-temperature exposure, the inward-diffusion of Al from the coating causes a cellular transformation and results in the formation of a secondary reaction zone in the substrate of the single crystal superalloy [42,43]. The formation of the SRZ, especially the emergence of the needle-like TCP phase, decreases the ductility and creep resistance of the Ni-base single crystal superalloy because of the initiation of microcracks at the tip of these brittle intermetallic-precipitates [44]. This is regarded as the most serious problem in the adoption of a high-temperature protective coating on the Ni-based single crystal superalloy. As the Ni₃Al-base single crystal superalloy possesses a γ/γ' coherent structure similar to that of the Ni-based SC alloy (their percentages of the γ' phase differ), precipitation of the needle-like TCP phase is harmful to the current single crystal superalloy as well.

According to the experimental results, a wide SRZ (108 μm in thickness) was formed beneath the NiPtAl coating at 1150 °C for 200 cycles. In contrast, the thickness of SRZ formed under the NiRePtAl coating was 81 μm, which indicates a 25 % reduction. Besides, the density of white TCP precipitates for the NiRePtAl coating was lower than that of the NiPtAl coating. This should be definitely ascribed to the presence of a Re-rich diffusion barrier, which restrained the inward-diffusion of Al. As the interdiffusion of elements between NiRePtAl coating and SC substrate had been greatly relieved, the total thickness of SRZ after long time thermal exposure could be significantly decreased as well. Two reasons are responsible for the NiRePtAl coating having less SRZ formation. Firstly, as mentioned above, Re-rich DB layer can mitigate the inward diffusion of Al in the coating. Less interdiffusions of Al and Ni will undoubtedly be helpful in stabilizing the original γ/γ' coherent structure. Refractory elements like W, Mo and Re in the substrate are concentrated in γ phase. Thus, the SRZ formation is inhibited to a certain extent. Secondly, there was thinner thickness of IDZ beneath the NiRePtAl coating during the aluminization process due to the Re-rich layer, as shown in Fig. 1(f). This reflects lesser interdiffusion between coating and substrate, which also retarts formation of SRZ. As long as the Re-rich DB layer remains adjacent to the substrate/coating interface, and maintains its thickness, it will retard the interdiffusion of elements, including those of Al, Ni, and other refractory metals. Thus, the SRZ formation is inhibited to a certain extent.

Actually, the present Re-rich layer acted as a mesh-like diffusion barrier, although it could not prevent the elemental interdiffusions by 100 %. Elements could still pass through this layer at a gap/channel between two Re-rich particles. In industrial applications, it is not necessary to completely inhibit the elemental interdiffusions, because to insulate diffusion between the coating and the substrate would raise another worrying concern: the bonding between them. Considering the factors listed above, we are satisfied that the present metallic Re-rich layer is suitable to serve as a diffusion barrier for the Pt-modified aluminide coating on Ni- or Ni₃Al- based single crystal superalloys.

A low-diffusion β-(Ni,Pt)Al coating with a Re-rich DB layer was prepared on a Ni₃Al-base single crystal superalloy via electroplating and gaseous aluminizing treatments. In comparison with simple aluminide (NiAl) and NiPtAl coatings, the cyclic oxidation behavior of the NiRePtAl coating was evaluated at 1100 and 1150 °C for 500 and 200 cycles. According to the experimental results, the following conclusions can be drawn:

(1) Because of the presence of the Re-based DB, the outward diffusion of Mo was effectively inhibited for the NiRePtAl coating. Pt and Re both contributed to the prevention of outward diffusion, which was achieved by stabilizing the β phase and via the fixation of Mo.

(2) Better oxidation performance was observed in the NiRePtAl coating, which showed lower rates of oxide growth and phase degradation.

(3) The NiRePtAl coating developed continuous and adherent TGO scales on the surface during the cyclic oxidation tests, during which the extent of rumpling of the alumina scale was much lower than that of the NiAl and NiPtAl coatings.

(4) Elemental interdiffusions between the NiRePtAl coating and the substrate were greatly reduced by the Re-rich layer, wherein the thickness of the SRZ formation and the amount of TCP precipitates in the alloy matrix were noticeably decreased.

This project was sponsored by the Key-Area Research and Development Program of Guangdong Province (2019B010936001). This work was also financially supported by the National Natural Science Foundation of China (Grant Nos. 51671202 and 51301184).