Preparation of MCrAlY–Al2O3 Composite Coatings with Enhanced Oxidation Resistance through a Novel Powder Manufacturing Process

MCrAlY–Al2O3 composite coatings were prepared by high-velocity oxygen fuel thermal spraying with bespoke composite powder feedstock for high-temperature applications. Powder processing via a suspension route was employed to achieve a fine dispersion of α-Al2O3 submicron particles on the MCrAlY powder surface. This was, however, compromised by ~ 50% less flowability of the feedstock during spraying. Nevertheless, the novel powder manufacturing process introduced in this study has shown potential as an alternative route to prepare tailored composite powder feedstock for the production of metal matrix composites. In addition, the newly developed MCrAlY–Al2O3 composite coatings exhibited superior oxidation resistance, compared to conventional MCrAlY coatings, with the formation of nearly exclusively Al2O3 scale after isothermal oxidation at 900 °C for 10 h. The addition of α-Al2O3 particles in the MCrAlY coatings as a second phase was found to have promoted the formation of YAG oxides (YxAlyOz) during spraying and also accelerated the outwards diffusion of Al, which resulted in enhanced oxidation resistance.


Introduction
Alumina-forming coatings, also known as bond coats, have been widely used on hot section of gas turbine engines since early 1960s (Ref [1][2][3][4][5]. Among them, a bond coat system of MCrAlY (M = Ni/Co) alloy has been developed based on a balance among several aspects that include oxidation, corrosion and ductility (Ref [6][7][8]. High-velocity oxygen fuel (HVOF) thermal spraying is one of the most widely used coating techniques for the deposition of MCrAlY coatings for high-temperature corrosion/oxidation protection ( . HVOF thermal-sprayed MCrAlY coatings are normally used within the temperature range of 950-1100 C with the formation of a 1-10-lm-thick protective a-Al 2 O 3 scale. This, however, creates a temperature gap for the use of MCrAlY bond coats in other high-temperature applications, e.g., ultra-supercritical (USC) steam boilers and turbines, and other turbomachinery for higher effciencies and lower emissions, which are in operation at temperatures below 950 C but in excess of 650 C (Ref [12][13][14]. No protective a-Al 2 O 3 scale would form on the coating surfaces at these low temperatures as restricted by both kinetic and thermodynamic factors. Metastable hand c-Al 2 O 3 that are non-protective (voids) would form instead in the temperature range of 700-950 C ( Ref 15), which are also fast growing with oxidation rate constants (k p ) being 2-3 orders of magnitude higher than that of a-Al 2 O 3 .
Efforts have been made to develop MCrAlY-Al 2 O 3 composite coatings with dispersion of Al 2 O 3 particles on the surface as nucleation sites to promote the early formation of a coherent a-Al 2 O 3 scale (Ref [16][17][18][19][20]. The challenge is, however, there is no established process to prepare composite powder feedstock with a homogenous distribution of the second phase particles with submicron or nano-sizes that are prone to agglomeration (Ref 6). A recent study by Bolelli et al. (Ref 21) produced NiCrAlY-Al 2 O 3 composite coatings by a ''hybrid'' spraying technique with two injection nozzles feeding both dry NiCrAlY powder and a suspension of Al 2 O 3 particles simultaneously into the gas stream. More recently, satelliting, as a manufacturing technique (Ref [22][23][24][25], was introduced to mix small quantities of hard reinforcing submicron or nanoparticles into larger volumes of soft metallic matrix powders. The ''satellited'' powders were obtained by mixing the two powders in a water solution containing 2.7 wt.% of polyvinyl alcohol (PVA) as a binder, followed by a drying process. This satelliting process has been patented ( Ref 22), and it is now used for a wide range of composite manufacturing applications that involves powder feedstock, e.g., additive manufacturing and cold spraying (Ref [23][24][25][26]. In this study, a similar powder mixing process via a suspension route was introduced with the aim to achieve a uniform distribution of Al 2 O 3 submicron particles with MCrAlY powders, albeit without any binders. The composite feedstock powders were then sprayed onto stainless steels by conventional HVOF thermal spraying. The oxidation behavior of the composite coatings during isothermal heat treatment at 900 C for 10 h was studied compared to the conventional MCrAlY coatings. The effect of the a-Al 2 O 3 dispersion on the composite coatings' oxidation was also discussed.

Preparation of Composite Feedstock
A suspension route was employed to prepare the composite feedstock without any additives. Ethanol denatured (industrial methylated spirit) with a concentration of ethanol over 99% in volume was used as the suspension medium. The mixing process was completed in less than 10 min on a hot plate at 150 C with a magnetic stirring (500 rpm) until fully dried. An AS-300 Hall Flow Meter was used to test the fowability of the composite powder feedstock before spraying. The fowability of the composite powder feedstock was found to be poor, so equal weight of original MCrAlY powders was added to improve the fow rate of the composite feedstock and mixed in a rotary milling system (CAPCO Test Equipment 2VS, Suffolk, UK) at a speed of 250 rpm for less than 1 h-the conditions chosen did not change the particle size and shapes. The homogeneity of the fnal feedstock was checked under optical microscope. The fnal composite powder feedstock containing 5 wt.% of Al 2 O 3 was then used for spraying.

HVOF Thermal Spray Process
Substrates were frst grit blasted with brown alumina (F100, 0.125-0.149 mm) under 6 bar pressure and then cleaned by an ultrasonic acetone bath to remove any embedded alumina particles. The substrates were mounted onto a carousel rotating at 73 rpm with a vertical axis of rotation. Metjet IV, a liquid fuel-based HVOF thermal spray system (Metallisation Ltd., UK), was used for the deposition of the MCrAlY-Al 2 O 3 composite and the conventional MCrAlY coatings with kerosene as the fuel. Both coatings were deposited with six passes under the same condition, and the samples were air-cooled during spraying. The length of the nozzle of the HVOF thermal spray gun was 100 mm, and a standoff distance of 356 mm was used during the spray runs. The fow rates of kerosene and oxygen were 476 mL/min and 900 L/min, respectively, giving a total fow rate of 28.2 g/s and 100% stoichiometry, while nitrogen was used a carrier gas for the powder. The as-sprayed samples were then cut into a dimension of 10 9 10 9 2 mm and placed in alumina crucibles. Isothermal oxidation treatment was carried out at 900 C for 10 h in a box furnace with a heating/cooling rate of 10 C/min.

Characterization Methods
Scanning electron microscope with JEOL in-lens Schottky feld emissions source (FEG-SEM, JEOL 7100F, USA) equipped with an energy-dispersive x-ray (EDX) detector (silicon drift detector size: 150 mm 2 , Oxford Instruments, UK) was used to examine the feedstock powder, the coating surface and cross sections in secondary electron (SE) and back-scattered electron (BSE) modes. An x-ray diffractometer (XRD, D500 Siemens Germany) with Cu Ka radiation (1.5406 A °) was used to identify the phases presented in the powders and coatings in the 20 B 2h B 90 range with a step size of 0.05 and dwell time of 2 s. For the oxidation analysis, the coating samples were mounted in epoxy resin flled with Ballotini glass beads for low shrinkage, and the cross sections were cut and polished down to a 1 lm fnish.

Thermo-Calc Calculation
Thermodynamic calculation of oxidation products as a function of effective partial pressures of O 2 at 900 C was carried out by Thermo-Calc software (Version 2017b) with TCS Ni-based superalloys (TCNi8.1) database following CALPHAD (Computer Coupling of Phase Diagrams and Thermochemistry) ( Ref 27,28). The use of Thermo-Calc on alumina-forming and chromia-forming alloys systems has been validated by our previous work (Ref 29,30) and others (Ref 31 -38).

Results
Composite Feedstock Powders Figure 1 shows the representative morphologies of the composite feedstock powder with 10 wt.% of alumina as examined after drying. In comparison with the original MCrAlY feedstock (left), the composite feedstock exhibits a good dispersion of alumina submicron particles mainly attached on the surface of the MCrAlY surface with a radius ratio in the range of 40 * 80:1. Figure 2 shows the combined and normalized XRD patterns of the composite feedstock powder with 10, 5 and 0 wt.% of alumina (from top to bottom). The MCrAlY powder is composed of a mixture of b-Ni(Al,Cr) phase and c-Ni(Cr) phase with a minor impurity of Ni 2 Al 3 and Ni 5 Y phase. With the increase in the alumina content, the peak intensity of the a-alumina is observed to increase proportionally, while it became diffcult to identify the a-alumina peaks when the a-alumina content was less than 5 wt.% (approximately equal to 2.7 mol.%), which might reach the detection limit of XRD analysis. This could be used as an effective indicator for the examination of the alumina content in the composite coatings using XRD analysis.    with a minor amount of Ni 5 Y phase. The presence of alumina is, however, not identifed on the coatings surface, possibly because the content of alumina near the surface is less than 5 wt. % within the x-ray penetration depth range, which is approximately 3-5 lm for MCrAlY. The assprayed coatings were further examined by SEM and EDX for microstructure and composition. Figure 4 shows the surface morphology and EDX elemental maps of the as-sprayed coatings. The surface features of both coatings are nearly identical consisting of molten splats and partially molten/unmolten MCrAlY particles. The compositions of the coatings surfaces are also very similar as indicated by the EDX semiquantitative analysis showing a majority of Ni, Cr and Al with a minor amount of O (6-7 wt.%) and Y (1-2 wt.%). An Y-rich area is also found on the composite coatings surface (marked by white arrow). The formation of these oxides is believed to be mainly caused by the inevitable oxidation during thermal spraying due to the interaction with atmospheric oxygen (Ref 18). This indeed makes the detection of the presence of original alumina particles even more diffcult in contrast to the newly formed alumina. Figure 5 shows the SEM images and EDX mappings with representative inter-lamella oxides on the cross sections of the two coatings. Both coatings are observed to be dense and uniform with the presence of a certain amount of inter-lamella oxides (dark phase), which are identifed as alumina-rich oxide. The thickness, however, differed with * 50% reduction in the composite coatings although both coatings were prepared under the same conditions. An enrichment of Cr is observed in the alumina-rich oxides in the MCrAlY coating, i.e., (Al,Cr) 2 O 3 , while the aluminarich oxides in the composite coatings contain nearly no Cr but a higher level of Y. The formation mechanism for the different oxide compositions will be discussed later. Figure 6 shows the XRD analysis on the coatings surface after oxidation at 900 C for 10 h. The peaks of both b and c(c 0 ) phase are observed to be narrower compared to these of the as-sprayed coatings (Fig. 3), possibly due to the strain relaxation during oxidation. The presence of alumina is again diffcult to identify as the composite coatings was unnoticeably oxidized with only a minor amount of spinel; while on the MCrAlY coatings surfaces, the peaks of NiO are more prominent together with the peaks of spinel oxide (Ni(Al,Cr) 2 O 4 ). Figure 7 shows the surface morphology and EDX spectra of the oxidation products on the coatings top surface in BSE mode, together with the EDX elemental maps in Fig. 8. On the MCrAlY coatings surface, the area that appears to be brighter in contrast is identifed as Ni-rich oxide (NiO), while dark area corresponds to the Al-rich oxide with the enrichment of Cr and Y. The bright oxides on the MCrAlY coatings (inset image) exhibited typical angular shape of NiO (Ref 39). On the other hand, the surface of the composite coatings appears to be uniformly dark contrast with wrinkled morphology. The EDX maps (Fig. 8) show that the oxides on the composite coatings are composed of mainly Al-rich oxide with nearly no Ni-rich oxide. This agrees with the XRD analysis as less NiO was identifed on the composite coatings. The Al-rich oxide is believed to be alumina although the XRD failed to detect the presence of alumina probably because the oxide scale is too thin and falls under the detection limit of the XRD.  Figure 9 shows the cross-sectional microstructure and compositions of the oxide scale on the two coatings. A coherent Al-rich oxide layer is observed on the composite coatings surface (right) with the thickness of less than 1 lm, while the oxides on the MCrAlY coatings (left) are composed of mixed compositions with Ni-rich, Cr-rich and Al-rich oxide throughout the coating. It is also worth noting that the oxidation analysis presented here was only based on short-term oxidation tests to evaluate the oxidation behavior of the newly developed composite coatings. However, for longer-term heat treatment, the oxidation resistance may differ as the Al element is depleted earlier in the thin coatings than the thick ones due to the continuous growth of Al 2 O 3 and interdiffusion with the substrates. In addition, previous researches have suggested that the thickness of bond coat has an effect on the residual stress in the coatings and may lead to a different failure behavior (Ref 40). Thus, more complicated performance tests (e.g., thermal cycling) are still needed to fully examine the oxidation resistance of these composite coatings.

Discussions
The MCrAlY-based composite coatings reinforced by alumina or other refractory oxides (e.g., Y 2 O 3 , ZrO 2 ) have already been developed previously via different routes (Ref [41][42][43][44][45], and most of these studies were only focused on the wear performance of the composite coatings. This is  (Fig. 4 and 5) based on the different segregations of Y on the top surface and cross section of these two coatings. The EDX semiquantitative analysis gave the concentration of Y in the composite coatings (cross section: 1.6 wt.% and top surface: 1.2 wt.% based on three measurements) relatively higher than that of the MCrAlY coatings (cross section: 1.2 wt.% and top surface: 1.0 wt.%). Since the starting concentration of Y in the MCrAlY powder was only 0.8-1.2 wt.%, it means that the relative concentration of Y in the selected area for EDX Fig. 6 XRD patterns of the oxidized MCrAlY and MCrAlY-Al 2 O 3 composite coatings after short-term exposure to high-temperature treatment at 900 C for 10 h analysis of the composite coatings was increased 1/3 to 1/2. This is possible because the high specifc surface area of the submicron alumina particles would signifcantly increase the probability of reaction with Y and O during spraying, forming stable YAG compounds (Y x Al y O z ) as a mixture of Y 2 O 3 and Al 2 O 3 leading to a locally higher Y content near the alumina particles. The presence of YAG was also observed by Toma et al. (Ref 48) in the HVOF thermally sprayed MCrAlY coatings doped with Al and Y metallic powders in the feedstock. Furthermore, Cr is another important indicator that could reveal the presence of the original a-alumina. The coexistence of Al and Cr is commonly seen in oxides (e.g., corundum) after oxidation of MCrAlY powders ( Ref 18). The absence of Cr in the Al 2 O 3 -rich oxides in the composite coatings (Fig. 5) indicates that these lamella oxides in the composite coatings were retained from the original alumina particles that contain no Cr at all.
The selective oxidation of MCrAlY alloys mainly involves with the thermodynamic that can be explained in Fig. 10. It shows the calculation results of the oxide formation (mole fraction) on the MCrAlY coatings with the composition of Ni-23Cr-11Al-1.2Y as a dependency of O 2 partial pressures (atm) at 900 C at equilibrium. Under low   The dispersion of a-Al 2 O 3 particles, as a second phase, as well as other reactive element (RE) oxides, on the oxidation resistance of ODS (oxide dispersion strengthening) alloys have been discussed at length ( Ref 49,50). The mechanism involves with the dynamic-segregation theory (DST), in which these oxide dispersions mainly segregate at the oxide grain boundaries and oxide-metal interface (Ref 51). The segregations of a-Al 2 O 3 and YAG oxide would hinder the outwards diffusion of undesirable outward cation transport (Ref 52). In particular, the diffusion coeffcient of Al along the grain boundary of Al 2 O 3 is several orders of magnitude higher than that of Ni and Cr [Al: 2.8 9 10 -4 m 2 /s; Cr: 6.9 9 10 -11 m 2 /s; Ni: 2.53 9 10 -10 m 2 /s (Ref 53)]. As a result, the growth of these non-protective spinel and NiO oxides would be greatly inhibited by the dispersed Al 2 O 3 particles. The EDX semiquantitative results in Fig. 7 already showed that the Al content on the composite coatings top surface after oxidation (26.1 wt.%) is nearly twice that on the MCrAlY coatings (13.6 wt.%), while the initial compositions of both coatings prior to oxidation are nearly the same with the Al content being 10-11 wt.% and the Ni content being 62-63 wt.%. The increase in the Al content after oxidation indicates that the outwards diffusion of Al in the composite coatings during oxidation outperformed the competitive outwards diffusion of Ni. The element of Cr, on the other hand, remains a similar level (16-17 wt.%) in both coatings after oxidation (Fig. 7), and also, appears to be granular precipitates with an isolated and dispersed distribution, and a size of 1-2 lm inside the two coatings as shown earlier in Fig. 9. It implies that the Cr was less involved in the competitive outwards diffusion with Ni and Al during oxidation. In addition, the dispersion of the a-Al 2 O 3 and YAG would act as the nucleation sites for the newly formed a-Al 2 O 3 phase (Ref 15) and also suppress the formation of other transient metastable Al 2 O 3 (e.g., h, c-Al 2 O 3 ) or shorten its phase transformation to stable a-Al 2 O 3 (Ref 54,55). It would also inhibit interfacial void growth, thus improving scale adhesion. To summarize, the improved oxidation resistance of the composite coatings with accelerated outwards diffusion of Al is believed to be caused by the dispersion of the a-alumina particles, in combination with the YAG formation during spraying of the composite powder.

Conclusions
A detailed process has been introduced in this study for the preparation of MCrAlY-Al 2 O 3 composite coatings via a suspension route followed by HVOF thermal spraying. A good distribution of alumina submicron particles was observed with the MCrAlY powders, and this, however, led to poor fowability for thermal spraying. Moreover, the presence of a-Al 2 O 3 particles promoted the oxidation of Y during spraying forming YAG inter-lamella oxide. The composite coatings exhibited also superior oxidation resistance with the formation of nearly exclusively Al 2 O 3 scale after short-term oxidation tests at 900 C for 10 h in comparison with the MCrAlY coatings. The presence of a-Al 2 O 3 particles in the composite coatings is believed to have accelerated the outwards diffusion of Al cations and inhibited Ni and Cr along the grain boundaries during oxidation, which therefore promoted the formation of Al 2 O 3 scale. distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.