Introduction

Owing to their superior oxidation resistance at high temperatures, MCrAlY (M: Ni, Co, or Ni + Co) coatings have been widely used to protect Ni-based superalloy components from oxidation and corrosion in high-temperature environments [1,2,3]. The high-temperature oxidation resistance properties of MCrAlY coatings are derived from the outward diffusion of aluminum elements from coating to surface, resulting in the formation of a dense α-Al2O3 scale on the coating surface [4]. Consequently, it is important to pay close attention to manufacturing defects and coating failures during the production process. Moreover, to extend the service life of substrates, repairing the coating without degrading the substrate by removing the original failed coating and applying a new coating [5] is a cost-effective and practical approach.

Traditional coating removal techniques include sandblasting [6], polishing [7, 8], and high-pressure water injection [9]. However, the powerful impact forces generated by these techniques may alter the shape profile of the Ni-based superalloy components. Currently, the majority of high-temperature oxidation resistance coatings are removed using chemical methods [10]. Bouchaud et al., utilized cathodic polarization to promote oxide flaking via hydrogen bubbling [11] in electrolytes containing inorganic acids and organic solvents, including ethanol. In contrast to β-NiAl, the solubility and outward-diffusion capacity of Cr in γ′-Ni3Al increased, resulting in a large difference between the constituent phases of the top oxide scale of the oxidized aluminide-coatings from pure Ni and Ni20Cr substrates [12]. Consequently, the removal rate of the aluminide coatings was significantly slower for the latter substrates when using cyclical potential stripping [12]. The chemically prepared PtAl coatings on Ni-based superalloys were easily removed by a solution containing HCl, H3PO4, and H2O [5]. Clearly, the removal of high-temperature protective coatings is contingent on their composition and the methods of preparation. MCrAlY coatings are an important thermal barrier for superalloy-based components, and the remanufacturing process must account for the removal efficiency of unqualified coatings. However, there are few reports concerning the removal efficacy of MCrAlY coatings deposited by arc ion plating (AIP), which can produce a denser and more uniform coating than the thermal spray method.

The mechanical properties of the substrate and the oxidation resistance of re-coated MCrAlY coatings must also be considered, as they are inevitably affected by the coating-removal procedure [8,9,10,11]. When re-coating the surface of IC6 alloy with NiCrAlYSi, the room-temperature mechanical properties of the substrates could be maintained while the stress-rupture life was decreased at 1100 °C/90 MPa [13]. Compared to conventional coatings, the oxide scale on re-coated NiCoCrAlY coating on Ni-based alloys was more susceptible to peeling [14]. The high-temperature oxidation resistance of the re-coated coatings was responsible for the previous removal depth of the interdiffusion layer of the oxidized CoCrAlY coatings [15], which was removed using a physical stripping technique. This suggests that, during remanufacturing, the effect of the removal step on the performance of the repair coatings, such as mechanical properties and corrosion resistance, must be carefully evaluated.

In this study, NiCoCrAlYTa coatings were deposited using AIP onto DZ22B Ni-based superalloy substrates. Additionally, the effect of a removal method combining electrochemical stripping and sandblasting on the mechanical and oxidation-resistant performance of repaired coatings was studied. Accordingly, the findings of this study were found to be applicable to the remanufacturing of the thermal barrier of Ni-based alloys.

Experimental materials and methods

Substrate and coating preparation

Ingots of the Ni-based superalloy were cut into rectangle specimens measuring 30 mm × 10 mm × 1.2 mm for oxidation and microstructure tests and rod-shaped specimens measuring 33 mm in gauge length and 8 mm in diameter for mechanical property tests. Table 1 displays the nominal composition of the DZ22B substrate and the target for AIP coatings. Before deposition, the surface of the substrate was wet-sandblasted to improve adhesion, then cleaned with acetone and alcohol, and finally dried in a hot air oven for 30 min. Similar to our previous studies [16, 17], a NiCoCrAlYTa coating with a thickness of approximately 30–40 μm was deposited on the specimen surface by AIP, and the processing parameters for the deposition of the ordinary and secondary repair coatings are shown in Table 2. To homogenize the chemical components, the specimens were annealed for 4 h at 1080 °C in a vacuum furnace. For convenience, the standard NiCoCrAlYTa coating will henceforth be referred to as O–C# and the re-coated NiCoCrAlYTa coating as R–C#.

Table 1 Nominal compositions of the cast alloy DZ22B and the NiCoCrAlYTa cathodes (wt.%)
Full size table
Table 2 Platting parameters for NiCoCrAlYTa coatings
Full size table

Coating removal

To avoid maximum surface damage during the coating removal process, the NiCoCrAlYTa-coated specimens were etched for less than 10 min in a 10% hydrochloric acid solution at 70 °C in a Ni-based alloy substrate. Thereafter, they were wet-blasted with a 200-mesh alumina powder suspension in water for 20–60 min under a pressure of 2 atm.

Performance and microstructure characterization

The coated specimens were tested for tensile strength and elongation at room temperature using a CMT4204 material testing machine with a strain rate of 1 × 10–3 s−1. Life and fracture-elongation of stress rupture were measured at 982 °C under a constant tensile stress of 220 MPa. The isothermal oxidation test was conducted in air at 1050 °C for 200 h, and the weight gain was measured every 25 h using a 0.1 mg-resolution precision optical balance. To protect the oxide scale during the preparation of the metallographic coupons, a Ni-layer was applied to the oxidized specimens.

X-ray diffraction (XRD, Smartlab 9 kW, Cu Kα) was used to determine the phases of the NiCoCrAlYTa coatings. Scanning electron microscopy (SEM, Nova NanoSEM 430) and energy dispersive spectroscopy (EDS) were utilized to examine the chemical composition, surface morphologies, and fractography of the specimens.

Results and discussion

Microstructural evolution of the coatings during coating-removing-recoating

Figure 1a depicts the cross-sectional micromorphology of O–C#. It had a thickness of approximately 30 µm and was primarily composed of the γ-Ni phase, γ′-Ni3Al phase, and a small amount of the β-NiAl phase. Typical NiCoCrAlYTa coatings undergo elemental diffusion between coating and substrate following vacuum heat treatment, resulting in the elimination of voids as depicted in Fig. 1b. Figure 1c depicts the X-ray diffraction spectrum of the O–C# coating before and after vacuum heat treatment. After vacuum heat treatment, the β-NiAl phase with (220) and (211) characteristic diffraction peaks gradually disappear, and the γ-Ni/ γ′-Ni3Al phases with (111), (200), and (220) diffraction peaks become sharp [16].

Figure 1
figure 1

SEM cross-sectional image of O–C# (a/b) and its XRD patterns (c) before and after vacuum heat treatment

Full size image

Figure 2a depicts the cross-sectional morphologies of the original NiCoCrAlYTa coating following a 5-min etching. NiCoCrAlYTa coating thickness did not decrease significantly; however, the coating’s interior had corroded and become loose. The top surface SEM image of the 5 min-etched specimens is depicted in Fig. 2b, which reveals that the surface became uneven and contained numerous voids. After 10 min of etching, as depicted in Fig. 2c and d, the coating became thinner but was not completely removed; some coating residue still protected the substrate from severe etching. Wet sandblasting was performed to completely remove the remaining coating and minimize the impact of hydrochloric acid corrosion on the substrate. Figure 2e and f displays the cross section and top surface images of the specimens following 10 min of etching and 30 min of wet sandblasting.

Figure 2
figure 2

Cross-sectional and surface images after etching for 5 min (a, b), after etching for 10 min (c, d), and after wet blasting (e, f)

Full size image

During coating removal, Al, Cr, Ni, and Co have standard potentials of −1.66, −0.74, −0.25, and −0.28 V, respectively [18,19,20,21]. Aluminum and chromium react violently with hydrochloric acid, whereas nickel and cobalt react slowly. These corresponding hydrogen-releasing reactions are depicted by Eqs. (1–4). These reactions promote the formation of voids or pinholes in the coatings, which is advantageous for the ingress of hydrochloric acid. Thus, the process of surface thinning and partial stripping can be accelerated.

$${\text{2Al}} +{\text{6H}}^+ \to {\text{2Al}}^{3+} + {\text{3H}}_{2} \uparrow$$
(1)
$${\text{Cr}} +{\text{2H}}^+{\to } {\text{Cr}}^{{{2} + }} + {\text{H}}_{{2}} \uparrow$$
(2)
$${\text{Co}} + {\text{2H}}^+{ \to } {\text{Co}}^{{{2} + }} + {\text{H}}_{{2}} \uparrow$$
(3)
$${\text{Ni}} + {\text{2H}}^+{ \to } {\text{Ni}}^{{{2} + }} + {\text{H}}_{{2}} \uparrow$$
(4)

Figure 3a and b depicts the cross-sectional images of R–C# coatings before and after vacuum heat treatment, respectively. Similar to the O–C# coatings, the elimination of voids in the coatings was caused by elemental diffusion during vacuum heat treatment. In addition to the blurring of the coating-substrate boundary, no defects are visible in Fig. 3b. Figure 3c displays the X-ray diffraction spectrum of the R–C# coatings. There is no distinction between O–C# and R–C# specimens in terms of microstructure and phase composition. Overall, the effect of the complex removal treatment in this study on the R–C# microstructure can be disregarded.

Figure 3
figure 3

SEM cross-sectional image of R–C# (a/b) and its XRD patterns (c) before and after vacuum heat treatment

Full size image

Mechanical properties

Tensile testing

It is necessary to investigate the effect of substrate removal and reapplication on its mechanical properties. Neither O–C#- nor R–C#-coated specimens exhibited necking during room-temperature tensile tests in air, exhibiting the typical brittle fracture. Figure 4 depicts the tensile fractographic images of O–C# and R–C# specimens at room temperature. The macroscopic fracture morphology appeared comparable to that depicted in Fig. 4a, c. In the high-resolution images (Fig. 4b and d), the fracture surfaces exhibit a laminar structure with numerous secondary cracks. Some smooth terraces were formed in the W-rich precipitated phase depicted in Fig. 4f (Spot 2), and numerous tear edges were observed in the γ/γ′ phase zone of the substrate. The EDS chemical compositions of Spot 1 and Spot 2 are also attached. There are no internal fractures and voids at the coating-substrate interface after fracture, indicating a strong adhesion.

Figure 4
figure 4

Room temperature tensile Fractographies of O–C#-coated (a, b) and R–C#-coated (c, d, e, f) specimens

Full size image

The tensile results at room temperature are depicted in Fig. 5, where σRm represents the ultimate tensile strength, σRp0.2 represents the yield strength, and A represents the tensile fracture-elongation. The ultimate strength and yield strength of the coated specimens were almost identical to those of the DZ22B alloy substrate, while the fracture-elongation was slightly reduced. Either R–C# or O–C# specimens were homogenized once at 1080 °C for 4 h under vacuum. The reduction of the average elongation of R–C# specimens was likely the cause of performance scattering due to the non-uniformity of microstructure in the substrates [22,23,24], because the homogenization produced a small difference in the elemental composition of γ/γ' and precipitated phases. σRp0.2 of the R–C# specimens was greater than 985 MPa, which is within the application requirement's tolerance range. Thus, the current study demonstrated that the effect of the remanufacturing process on tensile performance can be acceptable.

Figure 5
figure 5

Room temperature tensile properties of DZ22B alloy and the coated specimens

Full size image

Stress rupture testing

The rupture life and rupture elongation of DZ22B alloy and NiCoCrAlYTa-coated specimens at 982 °C/220 MPa are depicted in Fig. 6. The NiCoCrAlYTa coating increases the rupture life by more than 20% in comparison to the DZ22B alloy. The average rupture elongation of R–C#-coated specimens was identical to that of O–C#-coated specimens, whereas its stress-rupture life was slightly diminished. However, the average strain rate remained approximately 0.6% h−1 for all specimens, exhibiting slight fluctuations.

Figure 6
figure 6

Tensile rupture performance of DZ22B alloy and the coated specimens under 982 °C/220 MPa

Full size image

The fracture surface morphologies of DZ22B alloy and O–C#- and R–C#-coated specimens at 982 °C/220 MPa are illustrated in Fig. 7. In contrast to the fracture mode of the room-temperature tensile test depicted in Fig. 4, the fracture mode of the stress rupture specimens appears to be a typical mode of micro-void coalescence, as depicted in Fig. 7a2, b2, and c2. Figure 7a3 depicts a thick oxide layer on the fracture surface of a DZ22B alloy specimen. As depicted in Fig. 7a1, b1, and c1, the fractured surface of the DZ22B specimen exhibited significantly more cracks perpendicular to the force axis and larger voids than the fractured surface of the coated specimen. On the R–C# specimens, the parallel deformation lines to the force axis can still be seen. As shown in Fig. 7b3 and c3, the oxidation of the coating-substrate interface is significantly attenuated, despite the presence of some cracks in the coatings. During this high-temperature, the short-duration rupture test, fracture behavior is still governed by the substrate via crack initiation and propagation via voids. The adhesion of the NiCoCrAlYTa coating to the substrate impeded the formation and growth of voids in the substrate, thereby enhancing the rupture life and rupture elongation.

Figure 7
figure 7

Fractography of the DZ22B (a1, a2, a3), O–C# (b1, b2, b3), and R–C#-coated (c1, c2, c3) specimens after rupture tests

Full size image

Because the yield strength of the substrate at 982 °C is greater than 200 MPa [22], the creep is caused by rupture elongation. The embrittled oxide layer on the DZ22B alloy prefers to become the crack initiation site. The stress concentration at the crack tip facilitates the formation of voids. Above 650 and 700 °C, the ductile–brittle transition transformation of β–γ–γ' in MCrAlY coating occurs [23, 24], and the yield and tensile ultimate strengths of the NiCoCrAlYTa coatings at 850 °C are only 58 and 65 MPa, respectively [25], so the formation of through-thickness cracks in the coatings is inevitable under 982 °C/220 MPa; however, their propagation stops near the interface (Fig. 7b3, c3). These coating fragments adhere strongly to the substrate and continue to protect it from oxidation.

The coating-removal step causes the accumulation of internal stresses and surface defects on the substrate and weakens the adhesion between the coating and substrate interface [26, 27]. Consequently, the effective load transferred to the coating of the R–C# specimen decreased, and the deformation lines in Fig. 7a1 can be maintained. Owing to the difference in elastic modulus between the coating and substrate, shear stress can be induced in the interface zone, leading to the formation of a de-adhesive crack parallel to the loading direction (Fig. 7b3, c3) and an increase in the thermal exposure area of the substrate. This fracture is larger in the R–C#-coated specimen than in the O–C#-coated specimen. Similar to other reports [28,29,30,31], these exposed interface regions can exacerbate the surface deterioration and coarsening of γ′-Ni3Al precipitates on the substrates. Overall, the process of repairing the NiCoCrAlYTa coating has no negative effect on the high-temperature durability of the substrate, and the re-coated NiCoCrAlYTa coating can still improve the ductility of the substrate at 982 °C.

Oxidation resistance

Figure 8 depicts the oxidation kinetic curves of DZ22B alloy, O–C#-coated specimens, and R–C#-coated specimens at 1050 °C. The oxidation kinetic curve of DZ22B alloy (substrate) approximates a linear law, indicating that at 1050 °C, DZ22B alloy has almost no oxidation resistance. Both O–C#-coated and R–C#-coated specimens initially undergo a rapid oxidation stage within the first 25 h, accompanied by a significant increase in weight; this is followed by a stable oxidation stage, which is typically dominated by the growth of the alumina scale [32, 33]. The oxidation kinetics curves of the two coated specimens roughly correspond to the parabolic oxidation kinetics law. O–C#- and R–C#-coated specimens have oxidation rate constants (Kp) of 2.59 × 10–12 and 3.16 × 10–12 g2cm−4 s−1, respectively. In previous studies, the oxidation constant for α-Al2O3 layer growth at 1000 °C was determined to be 10–14 g2cm−4 s−1 [14, 34], indicating the presence of elements other than aluminum diffusing to the coating surface to participate in the reaction during isothermal oxidation of R–C#.

Figure 8
figure 8

Isothermal oxidation kinetics of DZ22B alloy and the coated specimens at 1050 °C in air

Full size image

Figure 9 depicts the XRD patterns of a re-coated NiCoCrAlYTa coating that has been oxidized in air at 1050 °C for varying times. During the isothermal oxidation of R–C#, a series of oxides, including α-Al2O3, NiO, Cr2O3, and NiCr2O4, were produced as the reaction products of solute atoms with oxygen [35, 36]. The possible reaction processes of these oxides are listed in Eqs. (5–9) [37, 38]. Increasing the isothermal oxidation time weakens the oxide peaks of NiO, Cr2O3, and NiCr2O4, particularly the (111), (200), and (220) peaks of NiO, which gradually broaden and vanish. However, the characteristic (012), (104), and (116) peaks of α-Al2O3 intensify as oxidation progresses, indicating that the alumina-rich scale provides the greatest oxidation resistance [16, 39].

$${4}\gamma ^{\prime} - {\text{Ni}}_{{3}} {\text{Al}} + {\text{3O}}_{{{2} }} \to {2}\alpha {\text{ - }} {\text{Al}}_{{2}} {\text{O}}_{{3}} { + 12}\left[ {{\text{Ni}}} \right]$$
(5)
$$2\left[ {{\text{Al}}} \right] + 3\left[ O \right] \to \alpha {\text{ }} - {\text{ Al}}_{2} O_{3}$$
(6)
$$2\left[ {{\text{Cr}}} \right] + 3\left[ O \right] \to \alpha {\text{ }} - {\text{ Cr}}_{2} O_{3}$$
(7)
$$\left[ {{\text{Ni}}} \right] + \left[ {\text{O}} \right] \to {\text{NiO}}$$
(8)
$${\text{NiO}} + {\text{Cr}}_{{2}} {\text{O}}_{{{3} }} \to {\text{NiCr}}_{{2}} {\text{O}}_{{4}}$$
(9)
Figure 9
figure 9

XRD patterns of the NiCoCrAlYTa re-coated specimens after oxidization at 1050 °C

Full size image

The cross-sectional morphologies of the re-coated NiCoCrAlYTa coatings after oxidation at 1050 °C for various times are depicted in Fig. 10. In Fig. 10a, after 25 h of oxidation, a thin but continuous almost pure α-Al2O3 scale (dark contrast arrow) is formed on the coating surface; and the mixed oxides (light contrast arrow) are sporadically distributed on top of this α-Al2O3 scale, which are responsible for many of the characteristic peaks shown in Fig. 9. Comparing the thickness of the pure Al2O3 scale of the 25 h oxidized specimen to that of the 100 h oxidized specimen, the pure Al2O3 scale of the 100 h oxidized specimen became thinner; the mixed-oxide layer became twice as thick, and the Al2O3 fraction in it increased; some voids formed locally in the mixed-oxide cluster zone, as indicated by the arrow in Fig. 10b. These voids within the oxide layer are likely the result of the Kirkendall effect caused by the diffusion of elements between the NiCoCrAlYTa coating and the oxide layer [40, 41], as well as the volume shrinkage resulting from the transformation of θ-Al2O3 to α-Al2O3 [42,43,44]. After 200 h of oxidation (Fig. 10c), the oxide layer clearly separated into two sub-layers, one with a high fraction of Al2O3 (darker contrast) and one with a low fraction of Al2O3 (lighter contrast). EDS analysis reveals that Ni, Co, and Cr elements diffuse outward from the top surface, and large voids and cracks form in the sub-layer, resulting in the peeling off of the coating.

Figure 10
figure 10

Cross-sectional images of the R–C# coating after oxidization for a 25 h, b 100 h, and c 200 h

Full size image

A continuous and dense scale of nearly pure α-Al2O3 contributes significantly to the oxidation resistance of MCrAlY-coated components by inhibiting the out-diffusion of various alloying elements. As oxidation progresses, voids and microcracks form in this pure α-Al2O3 sub-layer, leading to its fracture; subsequently, the oxidation of the NiCoCrAlYTa coating is exacerbated [45]. The solid-state reaction between the mixed oxides produces NiCr2O4 spinel, which influences the density of the mixed oxide [46,47,48]. In addition, voids and microcracks can be produced in the oxide layer by both oxide growth stresses and thermal expansion stresses [49, 50]. Because the affinity of Al with O is the strongest of all alloying elements, inward-diffusion O prefers to form α-Al2O3 to produce a dense growing sub-layer, while other elements tend to diffuse through this sub-layer to form a mixed-oxide top sub-layer. As shown in Fig. 1b, after the α-Al2O3 sub-layer is fractured, Al diffusion to the top surface accelerates, resulting in the formation of a mixed-oxide layer containing a high proportion of Al2O3 compounds. Because the diffusion rate of Cr is slower than that of Al and Ni [12], particularly the high affinity with O fixes the Al in the bottom sub-layer of the oxide layer, as shown in Fig. 10c, a new top sub-layer with a low fraction of Al2O3 forms. The difference in Kp between R–C#- and O–C#-coated specimens is likely attributable to the defect density in the coating and interface during the removal step. This merits further study.

Conclusions

  1. (1)

    The deposited NiCoCrAlYTa coating can be removed completely and efficiently by a combining hydrochloric acid solution etching and wet sandblasting. The re-coated coating had the same constituent phases as the original coating after undergoing this complex removal procedure. Additionally, the re-coated DZ22B alloy specimens had a tensile yield strength of > 985 MPa at room temperature.

  2. (2)

    The conventional- and repair-NiCoCrAlYTa-coating-coated specimens exhibited greater stress-rupture life and fracture-elongation than the substrate alloy at 982 °C/220 MPa. During rupture tests, the re-coated specimens demonstrated a similar average creep strain rate of ~ 0.6% h−1, which was similar to the DZ22B alloy-coated specimens and the conventional-coated specimens. The elimination step decreased the rupture life and elongation by a small amount.

  3. (3)

    When oxidized in air at 1050 °C, the oxidation kinetic curve of the specimens that were re-coated with NiCoCrAlYTa coating conformed to the parabolic oxidation law, similar to the conventional-coated specimens. The oxidation rate constant (Kp) for the standard coating- and repair coating-coated specimens was 2.59 × 10–12 and 3.16 × 10–12 g2cm−4 s−1, respectively.