The Effect of Bond Coat Roughness on the CMAS Hot Corrosion Resistance of EB-PVD Thermal Barrier Coatings

Abstract

In a high-temperature, high-flame-velocity, and high-pressure gas corrosion environment, the intercolumnar pores and gaps of electron beam–physical vapor deposition (EB-PVD) thermal barrier coatings (TBCs) may serve as infiltration channels for molten calcium–magnesium–alumino–silicate (CMAS), leading to the severe degradation of TBCs. In order to clarify the relationship between the roughness of the bond coat and the CMAS corrosion resistance of the EB-PVD TBCs, 7 wt.% yttria-stabilized zirconia (7YSZ) TBCs were prepared on the surfaces of four different roughness-treated bond coats. The effect of the bond coat roughness on the columnar microstructure of the EB-PVD YSZ was investigated. The effect of the change of the bond coat’s microstructure on the CMAS corrosion resistance of the EB-PVD YSZ was studied in detail. The results showed that the reduction in the roughness of the bond coat contributes to the improved formation of the EB-PVD YSZ columns. The small and dense columns are similar to a lotus leaf-like structure, which could reduce the wettability of CMAS and minimize the spread area between the coating and the CMAS melt. Thus, the CMAS corrosion resistance of the coating can be greatly improved. This preparation process also provides a reference for the preparation of other TBC materials, improving the resistance to CMAS hot corrosion.

1. Introduction

As the key thermal protection materials for advanced gas turbine engines, thermal barrier coatings (TBCs) can significantly improve the service life of hot-end metal components, reduce fuel consumption, and increase the working temperature of the engine combustion chamber [1,2,3]. At present, the most widely used TBC is 7–8 wt.% yttria-stabilized zirconia (YSZ) [4,5]. However, in the complex service environment where molten calcium–magnesium–alumino–silicate (CMAS) infiltrates into TBCs, the dissolution of yttrium in 7–8% YSZ reduces the stability of the non-transformable tetragonal (t’)-phase YSZ and induces the phase transition from t’ phase to monoclinic (m) phase during thermal cycles, with a 3–5% volume expansion, accelerating the degradation of TBCs [6,7,8] and resulting in premature failure. Therefore, the degradation of TBCs caused by molten CMAS has received extensive attention.

TBCs prepared by electron beam–physical vapor deposition (EB-PVD) have a dense columnar structure with good strain tolerance at high temperatures [9]. However, the pores and gaps between the coating’s columnar structure provide a convenient corrosion channel for molten CMAS, which makes the TBC system more susceptible to molten CMAS and reduces the service life of TBCs [10]. To date, research on improving the CMAS corrosion resistance of TBCs has primarily focused on developing the following three aspects [11,12]: (i) anti-penetration coatings to replace YSZ, which can prevent the infiltration of molten salts, such as zirconate, pyrochlore, tantalite, and aluminate; (ii) sacrificial coatings on the top of YSZ, such as Al₂O₃ and MgO, which can react with molten CMAS to form compounds with a higher melting point than that of CMAS; and (iii) non-wetting coatings on the top of YSZ [13,14] to reduce the spread area between molten CMAS and TBCs. However, these approaches are difficult to achieve, and they also reduce the lifetime or increase the thermal conductivity of TBCs [11].

The wetting of CMAS melt plays an important role in terms of the degradation of TBCs by CMAS; the factors affecting wetting include liquid surface tension, solid surface energy, and surface roughness [15]. CMAS undergoes a short-duration diffusion process, which is mainly caused by a sudden decrease in the surface tension of molten CMAS [16]. Wu et al. [17] verified the relationship between surface roughness and the wettability of the top coat. As the surface roughness increases, the spreading rate of its CMAS melt gradually decreases [18]. Some scholars have also studied the relationship between the surface structures of YSZ coatings and the wetting behaviors of CMAS/YSZ systems to improve the CMAS resistance of YSZ coatings [19,20]. Unfortunately, little research has been conducted on the relationship between the surface roughness of the bond coat and the CMAS hot corrosion resistance of 7YSZ EB-PVD TBCs.

In this paper, 7YSZ EB-PVD TBCs were deposited on NiCrAlY bond coats with four roughness treatments. The morphology of the microstructure of the 7YSZ coating and the wetting behavior of CMAS melt on the 7YSZ coating under different bond coat roughness treatments were systematically studied. The purpose of this paper was to investigate the effect of the bond coat’s roughness on the CMAS hot corrosion resistance of 7YSZ coatings prepared by EB-PVD.

2. Experimental Procedure

Superalloy GH536 was selected as the material for substrates. Before depositing the bond coats, 30 mm × 10 mm × 3 mm substrates were ground with 240#, 400#, 800#, and 1200# mesh silicon carbide sandpaper and polished with diamond suspensions with a particle size of 1 μm. An NiCoCrAlY bond coat was then deposited on the nickel-based superalloy substrate by vacuum-arc ion plating (AIP) technology (DH-700) with a thickness of 80 μm. The detailed chemical composition of the bond coat is listed in

Table 1. Chemical composition (wt.%) of the bond coat.

Ni	Co	Cr	Al	Y
Bal.	10~12	18~25	8~12	0.2~0.5

. The specific deposition parameters are shown

Table 2. AIP deposition parameters for the bond coat.

Layers	Bombardment Current (A)	Bombardment Bias Voltage (V)	Furnace Vacuum (Pa)	Coating Thickness (μm)	Time (min)
Bond coat	60–70	600–800	5 × 10−3	80	90

. The surface of the bond coat was polished with 100#, 400#, or 1200# silicon carbide sandpaper (Eagle brand). In addition, the as-prepared bond coat sample without polishing treatment was used for comparison. A commercial YSZ ingot was employed in the EB-PVD process, with a chemical composition of 7 wt.% Y₂O₃ and 93 wt.% ZrO₂. The diameter of the 7YSZ ingot used for evaporation was 50 mm. The 7YSZ columnar coating was deposited on both bond coat groups using the EB-PVD method, with the specific parameters shown in

Table 3. EB-PVD parameters for the 7YSZ coating.

Layers	Electron Beam Current (A)	Rotation Speed (rpm)	Voltage (kV)	Heating Temperature (°C)	Vacuum Chamber Pressure (Torr)	Deposition Rate (μm/min)
7YSZ coating	0.5	10	8.5	950	1.0 × 10−5	2

. The average thickness of the prepared 7YSZ coatings was approximately 170 μm. A surface roughness profiler (MarSurf PS 10, Mahr GmbH, Esslingen, Germany) was utilized to measure the surface of the bond coat after different roughness treatments and the corresponding surface of the deposited 7YSZ coating. The samples were cut into 10 mm × 10 mm × 3 mm sections for the CMAS corrosion test.

The composition of CMAS was 33CaO-9MgO-13Al₂O₃-45SiO₂ (mol%), which was determined by referring to the deposits collected from the actual parts of the engine [21]. The CMAS powder and alcohol were mixed and ball milled for 8 h to obtain a uniform mixture. After drying in an oven at 80 °C, the mixture was calcined in a high-temperature muffle furnace at 1500 °C for 2 h to retain a molten state, followed by water quenching. The final CMAS powder was obtained after grinding and sieving (500# mesh).

The Φ3 × 1 mm CMAS disc bulk was formed with 16 mg CMAS powder using a briquetting machine, with the disc bulk placed at the center of the 10 mm × 10 mm × 3 mm YSZ sample surface. Normally, CMAS compositions do not have a melting point but possess a melt interval. A value of 1250 °C most likely lies within the interval close to the onset of melting [22,23], 1250 °C was determined as the test temperature [24]. The test sample was pushed along a track into a 1250 °C muffle furnace, where the geometric evolution of the CMAS bulk over time was recorded by optical imaging techniques. The contact angles were measured by ImageJ Contact Angle plugin (version 1.52a), and Adobe Photoshop CS6 was employed to measure the spread area of CMAS on the surface of the 7YSZ coating after the wetting test. To further investigate the interaction between coatings under different surface roughness treatments of the bond coat and molten CMAS deposits, 7YSZ coating samples with four different bond coat roughness treatments were subjected to CMAS infiltration experiments. The CMAS powder was uniformly deposited on the surface of the four TBC samples in a paste formation in the CMAS infiltration test and then heated to the same temperature as that used for the wetting test, held for 30 min, and cooled with the furnace. The amount of CMAS loaded on the coating was approximately 6 mg/cm².

A scanning electron microscope (SEM, MIRA3, TESCAN, Brno, The Czech Republic) and an energy-dispersive spectrometer (EDS, Ultim MAX-40, Oxford, UK) were used to observe the microscopic morphology of the 7YSZ coatings with different roughness bond coats, as well as the wetting behavior under CMAS corrosion.

3. Results and Discussion

3.1. Characterization of Microstructure

Figure 1a shows that the average surface roughness of the as-prepared 100#-, 400#-, and 1200#-polished bond coats is 3.54, 1.13, 0.50, and 0.10 μm, respectively. The corresponding roughness of the 7YSZ coating deposited on these bond coats is 4.88, 2.23, 1.56, and 0.62 μm, respectively. Obviously, the 7YSZ coating surface deposited on the smoother bond coat is also smoother, indicating that the roughness of the bond coat has a certain effect on the roughness of the 7YSZ coating prepared by EB-PVD. The lower the bond coat roughness, the lower the overall roughness of the column surface prepared by EB-PVD. The trend of column density in Figure 1b clearly shows that as the roughness of the bond coat decreases, the columns prepared by EB-PVD are denser and therefore finer. From top to bottom of all the coatings, the density of the columns increases, which represents the typical columnar structural growth form.

Figure 2a–d presents the microscopic morphologies of cross sections of the 7YSZ coating samples with four different bond coat roughness treatments. The columnar structures perpendicular to the bond coat are typical structures fabricated by EB-PVD, and this columnar structure has inherent high strain resistance, which is beneficial in terms of improving the long-term thermal exposure lifetime of TBCs [25]. As seen in Figure 2a–d, the columnar growth pattern of the 7YSZ samples becomes more uniform, and the growth direction is more consistent. Meanwhile, the columnar gaps become smaller, and the linear density of the column increases. This shows that the columnar structure is related to the surface roughness of the bond coat. TBCs prepared on a flatter bond coat have a more uniform columnar structure. The surface morphologies in Figure 2e–h exhibit a columnar appearance on the top of the columns, which is typical of most coatings prepared by EB-PVD [26]. With a decrease in the roughness of the bond coat, the surface structure is gradually densified, and the growth orientation of the columns tends to be consistent. Therefore, the pretreatment of the bond coat roughness is more favorable for the nucleation of YSZ grains and the growth of columns. Figure 2i–l shows the cross-sectional morphologies of the coating under higher magnification. It can be clearly seen that the coating columnar crystals of the sample treated with 1200# silicon carbide sandpaper appear to be denser, and the columnar gaps between the columns are narrower in comparison with samples treated with other bond coat roughness. The columnar crystals of the untreated sample exhibit non-uniform and irregular growth, and the columnar gaps appear to be quite large. Figure 2m–t presents higher-magnification images of the coating microstructure from the selected areas of Figure 2a–d. Feather-like microstructural morphologies can be clearly observed in all samples, although this microstructural feature is the most dominant in the untreated sample and diminishes with increasing bond coat surface treatment.

3.2. Wetting Experiments

The wetting process of the 7YSZ coatings with different bond coat roughness treatments at 1250 °C is shown in Figure 3a, and the change of CMAS contact angle over time is presented in Figure 3b. The geometry of the CMAS disc bulk changed from a cylinder to a sphere within 30 s and slowly spread out, and the contact angle stabilized after approximately 100 s. It can be seen from the CMAS wetting process in Figure 3 that the CMAS melt is almost completely spread on the 7YSZ coating surface at 100 s for the sample without bond coat roughness treatment. With a decrease in the roughness of the bond coat, the spreading speed of the CMAS bulk on the coating surface decreases, and the spreading area decreases. This phenomenon is similar to the lotus effect [27,28], which strongly indicates that the EB-PVD coating possesses enhanced resistance to the CMAS wetting behavior that could prevent the reaction between CMAS melt and the 7YSZ coating to a certain extent. The least rough bond coat possessed the finest EB-PVD columns, which can offer greater resistance to CMAS wetting by increasing the contact angle towards higher values and combining more feather arms to split the glass flow [29,30].

Table 4. Contact angle and spread area of CMAS on the surface of 7YSZ coatings with different roughness bond coats under CMAS wetting test.

Bond Coat Treatment	Contact Angle (°)	Spread Area (mm2)
As-prepared (untreated)	12.3	70.6
Polished with 100# silicon carbide sandpaper	14.2	65.1
Polished with 400# silicon carbide sandpaper	15.4	49.4
Polished with 1200# silicon carbide sandpaper	16.3	40.4

shows the contact angle and spread area of CMAS on the surface of the 7YSZ coatings after the CMAS wetting test. Among them, the contact angle of the as-prepared sample is smaller, and the spread area is larger. From this, it can be inferred that the 7YSZ coating without bond coat roughness treatment is more likely to be wetted by the CMAS melt and to react with it, thus accelerating the failure of the coating [15]. Conversely, the contact angle associated with the CMAS melt on the YSZ deposited on the smoother bond coat (1200#) is larger, and the spread area between the 1200# sample and the molten CMAS is smaller. This suggests that the 1200# sample is less likely to be eroded by CMAS. In other words, the microstructure of TBCs plays a key role in determining the wetting tendency of CMAS melts on 7YSZ coatings [10,31]. Therefore, with a decrease in roughness of the bond coat, the contact angle gradually increases, and the spread area with the CMAS melt becomes smaller, which further indicates that a lower roughness of the bond coat can improve the CMAS corrosion resistance of the coating.

3.3. Infiltration Experiments of Molten CMAS on 7YSZ Coatings

The infiltration of molten CMAS onto the 7YSZ coatings with different bond coat roughness treatments is shown in Figure 4. The CMAS melt penetrates from top to bottom along the intercolumn gap in the four samples. Figure 4a–c shows that the samples are fully infiltrated from top to bottom by CMAS melt and that the infiltration areas are large. Figure 4d shows that the CMAS of the 1200# sample with the lowest bond coat roughness has a smaller infiltration area compared to the other three samples. After the coating was infiltrated by CMAS melt for 30 min at 1250 °C, the columns of the 7YSZ coating deposited on the untreated bond coat were significantly degraded due to the excessive roughness, and the 7YSZ coating was more easily penetrated by the molten CMAS. The molten CMAS reacts with the 7YSZ coating, resulting in the failure of TBCs. As the roughness of the bond coat decreases, the columns of the 7YSZ coating can maintain the original structure under corrosion, and the columns have a higher density. It can be observed from the EDS map that the CMAS droplets on the surface of the 7YSZ coating without the bond coat roughness treatment are completely spread out and infiltrate along the YSZ columnar gaps, which is similar to the case of EB-PVD columns subjected to CMAS corrosion [32]. On the other hand, the spread area between CMAS droplets and the 7YSZ coating on which the bond coat had been treated is small, as is the degree of CMAS corrosion. The results confirm that the reduction in bond coat roughness was, to a certain extent, beneficial to the improvement of the CMAS hot corrosion resistance of the coating.

To further confirm the effect of different bond coat roughness treatments on the CMAS corrosion resistance of EB-PVD 7YSZ TBCs, the micro-morphologies at different positions of the 7YSZ coatings prepared on bond coats of four different roughness after CMAS hot corrosion are shown in Figure 5. The columnar gaps of the 7YSZ coating without roughness treatment of the bond coat are larger than those of the other samples. There are evident traces of CMAS infiltration and the CMAS melt aggregates between the columns of the coating, and the structure degraded accordingly (Figure 5a). Near the bottom, it can be observed that columns grew on the uneven bond coat, and the columnar gaps are relatively larger. With reduced surface roughness of the bond coat, the gaps of the TBCs system become smaller. The CMAS melt is distributed in a small amount in the columnar gaps, and the infiltration strength is low, as shown in Figure 5d. Structural degradation was not obvious after corrosion and also preserved the good columnar crystal morphology of the EB-PVD coating. These results confirm that roughness pretreatment of the bond coat can improve the CMAS corrosion resistance of the TBC system.

4. Conclusions

In this study, 7YSZ coatings were successfully deposited on bond coats with four different roughness treatments using an EB-PVD system. CMAS wetting experiments and CMAS infiltration experiments were carried out on 7YSZ TBCs deposited on four different bond coat roughness treatments. The main conclusions are as follows:

• The average surface roughness of the as-prepared 100#-, 400#-, and 1200#-polished bond coats is 3.54, 1.13, 0.50, and 0.10 μm, respectively. The corresponding roughness of the 7YSZ coating deposited on these bond coats is 4.88, 2.23, 1.56, and 0.62 μm, respectively. In the four different roughness treatments, the columns of the EB-PVD 7YSZ became denser with decreasing roughness of the bond coat. Moreover, the feather-like microstructural morphologies became less apparent with decreasing roughness of the bond coat.
• During the CMAS wetting process at 1250 °C, the 7YSZ coating prepared by EB-PVD tended to present with a lotus effect. This effect was particularly pronounced on the samples with the 1200#-polished bond coat. In the four different roughness treatments, the contact angle between the CMAS droplets and the 7YSZ coating increased, and the spread area became smaller as the roughness of the bond coat decreased.
• After the CMAS infiltration test at 1250 °C, the EB-PVD 7YSZ columns grown on the 1200#-treated (smoothest) bond coat could still maintain clearly visible columnar crystals. It is believed that the roughness pretreatment (i.e., polishing) of the bond coat can improve CMAS corrosion resistance by reducing the width of columnar gaps and consequently minimizing the penetration and flow of molten CMAS into the columns. In addition, with increasing bond coat surface smoothness, the feather-like microstructural feature also becomes less dominant, which clearly favors the mitigation of capillary motion-assisted infiltration of molten CMAS into the columns.