The Effect of Shot Blasting Abrasive Particles on the Microstructure of Thermal Barrier Coatings Containing Ni-Based Superalloy

Abstract

Grit particles remaining on the substrate surface after grit blasting are generally considered to impair the thermal performance of thermal barrier coatings (TBCs). However, the specific mechanisms by which these particles degrade the multilayer structure of TBCs during thermal cycling have not yet been fully elucidated. In this study, the superalloy substrate was grit-blasted using various processing parameters, followed by the deposition of thermal barrier coatings (TBCs) consisting of a metallic bond coat (BC) and a ceramic top coat (TC). After thermal shock tests, local thinning or discontinuities in the thermally grown oxide (TGO) layer were observed in TBCs where large grit particles were embedded at the BC/substrate interface. Moreover, cracks originated at the concave positions of the TGO layer and propagated vertically towards BC; these cracks may be associated with additional stress imposed by the foreign grit particles during thermal cycling. At the BC/substrate interface, crack origins were observed in the vicinity of large grit particles (~50 μm).

1. Introduction

Thermal barrier coatings (TBCs) are applied to protect critical hot components of gas turbine engines, such as combustion chambers and turbine blades, from extreme temperatures and thermal degradation [1,2,3]. The TBCs system is a multi-layer structure that is mainly comprised of a superalloy substrate, a metallic bond coat (BC), and a ceramic top coat (TC) [4,5]. The primary function of the top coat (TC) is to provide thermal insulation for components operating in high-temperature environments. Simultaneously, the bond coat (BC) serves to mitigate thermal mismatch between the substrate and the TC, thereby improving the adhesion strength of the ceramic top coat [6]. Prior to the deposition of the TBCs, grit-blasting treatment was often employed to enhance the adhesion and wetting property of the BC by roughening the substrate surface and promoting the mechanical interlocking of the deposited coating [7,8,9]. Some of the grit particles were inevitably trapped in the substrate due to plastic deformation of the substrate overlaying with the particles [9]. These foreign grit particles have important consequences for the thermal fatigue lifetime of TBCs due to possible stress concentration at the grit/substrate interface [10,11].

Studies have demonstrated a detrimental effect of the foreign grit particles on the thermal fatigue properties [9,12,13]. During thermal cycling consisting of 2 h heating at 1120 °C and 15 min rapid cooling to room temperature, additional stress is generated around the grit particles due to the mismatch in thermal expansion coefficients between the grit particles, the metal substrate, and the ceramic coating [13]. As a result, cracks would preferentially originate in the vicinity of the embedded grit particles when the stress is accumulated sufficiently beyond a critical level. However, the cracking of substrate or deposited coat around the retained grit particles does not always occur during thermal cycling [14,15], which may be associated with the size and shape of the particles [16]. Previous research on the effects of retained grit particles has been limited and primarily focused on a qualitative description of the conflicting roles of these foreign particles. The underlying mechanism by which retained grit particles influence thermal fatigue behavior has yet to be fully explored. At what size of grit particles is cracking most likely to occur in the BC or TC? By what mechanism do the retained grit particles influence thermally produced oxide (TGO) development and expansion?

Owing to the excellent high-temperature fatigue resistance, Mar-M247 nickel-based superalloy is widely used in the blade of heavy gas turbine [17,18]. This work presents a set of grit-blasting experiments conducted on the Mar-M247 Ni-based superalloy of different processing parameters. The objective is to acquire substrate surfaces with diverse grit characteristics. Next, thermal shock experiments were conducted to investigate how grit distribution affects the thermal fatigue characteristics of the TBCs. Interfacial microstructure deterioration during temperature cycling was analyzed using scanning electron microscopy (SEM) equipped with an energy dispersive spectrometer (EDS) and nanoindentation experiments.

2. Experimental

2.1. Materials and Coating Fabrication

The substrate material employed was Mar-M247, a nickel-based superalloy known for its exceptional resistance to creep and oxidation at elevated temperatures [17]. The chemical composition of Mar-M247 is provided in

Table 1. The blasting parameters for the sample A and sample B.

Sample	Blasting Angle	Blasting Pressure	Nozzle Moving Velocity	Blasting Distance	Grit Mass Feeding Rate
A	30°	0.3 MPa	5 mm/s	200 mm	3 kg/min
B	90°	0.5 MPa	5 mm/s	200 mm	3 kg/min

. Prior to TBC deposition, the samples with dimensions of 20 mm × 60 mm × 5 mm were grit blasted with alumina particles of size 24# mesh. The grit blasting was performed under a pressure of 0.3 MPa, 0.4 MPa, and 0.5 MPa and a blasting angle of 30°, 45°, 60°, 75°, and 90° to achieve different surface grit characteristics. Two samples exhibiting prominent variations in the distribution of grit particles were chosen from a set of grit-blasted samples and designated as samples A and B correspondingly. It should be mentioned that the blasting pressure was 0.3 MPa for sample A and 0.5 MPa for sample B, respectively, while the blasting angle was 30° for sample A and 90° for sample B. The detailed parameters including other blasting distance, grit mass feeding rate, etc. are listed in Table 1. Before surface characterization, the sample surface was cleaned sufficiently by compressed air. After grit blasting, an intermediate metallic coat was deposited by high velocity oxygen fuel (HVOF, JP-8000, Praxair technology, Indianapolis, IN USA.) system attached with a TAFA-5200 gun. The composition of the precursor power utilized for the intermediate coat is also listed in

Table 2. Chemical composition (wt.%) of Mar-M247 superalloy substrate samples and bond coat powder.

Sample	Co	Cr	W	Al	Ta	Mo	Ti	Hf	C	Ni	Y
A	10.2	8.2	9.8	5.7	2.9	0.81	0.96	1.5	0.15	Bal.	-
B	9.9	8.5	10.2	5.5	2.8	0.76	0.91	1.2	0.13
Bond coat	Bal.	21	-	8	-	-	-	-	-	32	0.5

. It was shown that the BC was mainly composed of nickel (Ni), cobalt (Co), tungsten (W), chromium (Cr), and aluminum (Al) elements, with a small amount of tantalum (Ta), Hafnium (Hf), titanium (Ti), and molybdenum (Mo), etc. The top YSZ had a nominal composition comprising 92 wt.% ZrO₂ and 8 wt.% Y₂O₃. The metallic BC and ceramic TC precursor powders were purchased from Oerlikon Metco Inc., New York, NY, USA. The YSZ ceramic coating was sprayed on the BC using an atmosphere plasma spraying (APS, Sulzer Metco Unicoat, Westbury, NY, USA.) system with an F4MB-XL plasma gun (Oerlikon Metco Inc., Westbury, NY, USA).

2.2. Microstructure Characterization

The cross-sections of coatings with different grit characteristics were polished using standard metallographic techniques and examined by scanning electron microscopy (SEM, Zeiss Ultra 55, Oberkochen, Germany). The changes in mechanical properties of the substrate in the vicinity of alumina particles were examined using nanoindentation tests. During nanoindentation tests, a diamond Berkovich indenter (Micro Star Technology, Huntsville, TX, USA) with a radius of ≈20 nm was used, and the area function of the indenter was calibrated using a standard fused silica sample before tests to ensure accuracy of experimental data. The nanoindentation tests were performed under constant load mode, in which the peak load was set to 200 mN, and the loading rate was controlled at 2 mN/s.

2.3. Thermal Fatigue Test

Thermal shock cycling tests were carried out to reveal the thermal fatigue behavior of YSZ-based TBCs. The thermal cycling testing method is referred to the international standard of ISO 14188:2012 [19]. During thermal shock cycling tests, the samples were first heated to 1000 °C in an electric furnace, kept at the temperature for 30 min, and then quenched into the water. The water temperature during the thermal shocking tests was controlled at 25 ± 2 °C. The thermal shock cycling test to evaluate the thermal fatigue property of TBCs was used elsewhere [20,21]. The samples for cross-sectional microstructure analysis were cut from the TBCs after 300 thermal shocking cycles.

3. Results

3.1. Surface Characterization

As shown in Figure 1, SEM examination of the grit-blasted sample surfaces revealed the distribution of abrasive alumina particles retained on substrate surfaces after grit-blasting treatment. The SEM imaging was carried out under backscattered electron (BSE) mode with a low magnification to reveal the global distribution of alumina particles. The alumina particles were readily identified as dark zones in the BSE images, as confirmed by the EDS mapping of Al and O elements. The dark zones were distinguished from the substrate and the distribution of alumina particles was quantified by processing and computerizing the scanning electron microscope (SEM) images using Image J 1.45 software. The size of the particles was measured to be ~138 ± 5 μm for sample A and ~198 ± 5 μm for sample B, respectively, while the volume fraction of the particles was ~12% for sample A and ~20% for sample B.

SEM imaging was performed under secondary electron and backscattered mode to observe more details on the alumina particles and substrate deformation, respectively, with a higher magnification across the surface. The alumina particles maintained on the substrate surface were found to be dispersed randomly and exhibited irregularity. The size of the retained grit particles in sample A was greater than that of sample B. More precisely, big particles in sample B exhibited visible microcracks, as seen by the circles in Figure 2c. Several fine particles were observed around the large particles, which possibly originated from the fragmentation of pristine alumina particles upon grit-blasting. The large grit particles (size > 50 μm) exhibited full cracking in the interior of embedded particles upon the impact with the substrate surface, resulting in a number of dispersed fragments around the particles, while the small particles (size < 30 μm) appeared to be less fragmented, possibly derived from the small blasting pressure and low blasting angle. The kinetic energy of the particles is mainly exhausted by deforming the substrate through micro-cutting action. Furthermore, the substrate surface feature after grit-blasting differed for the two samples. The surface of sample A displayed micro-cutting characteristics, as seen by the white arrows in Figure 2b, whereas the surface of sample B exposed crater-like structures. Differences in deformation characteristics can be attributed to variations in the angle of grit blasting. For sample A, the angular alumina particles were blasted at a low angle of 30°, which led to substrate deformation through cutting mode. In sample B, however, the high blasting angle always led to substrate deformation through impacting, forming crater-like features.

3.2. Spallation Resistance

Thermal shocking tests were conducted to examine the impact of abrasive alumina particle dispersion on the thermal fatigue characteristics of TBCs. In Figure 3, the samples’ spallation degree and related surface characteristics after 300 heat cycles are displayed. The spallation degree of sample B was determined to be 28%, which is approximately double the spallation degree of sample A (15%). This indicates that the presence of large alumina particles in the substrate strongly affects the thermal fatigue behavior of TBCs.

3.3. Interface Microstructure

To reveal how the retained alumina particles facilitate the spallation of the TBCs, we further examined the interfacial microstructure of the as-deposited TBCs and the TBCs subjected to thermal shocking tests, as shown in Figure 4. For the as-deposited TBCs, the SEM images identified the presence of alumina particles embedded at the substrate/BC interface in both samples. It can be seen that the particle size of sample B is much larger than that of sample A, which is consistent with the observation in Figure 2. The large particle with a size of ~40 μm in the sample B was able to penetrate into the subsurface of the substrate to the depth of ~20 μm, while the particles in the sample A appeared to be adhered on the substrate surface. It should be noted that no visible microcracks were developed around the embedded particles. After thermal shocking tests, TGO formed along BC/TC interface with undulated morphology. A clear difference in TGO morphology was observed between the two samples. Sample A exhibited a continuous TGO layer at the TC/BC contact. Still, sample B displayed local thinning or discontinuity in its TGO layer, as seen by the white arrows in Figure 4b. More importantly, cracks grew vertically from the concave positions at the TC/BC interface towards BC for the samples. In sample A, the front of the TGO has grown into a depth of ~10 μm inside the BC. Nevertheless, in sample B, a fracture began at the interface between the TC and BC and moved vertically into the contact between the BC and substrate by linking the empty spaces within the BC. Furthermore, it was shown that the size of alumina particles substantially impacted the formation of fractures in BC. In sample B, the development of microcracks inside BC was observed in the vicinity of the large alumina particle, the size of which was measured to be ~30 μm. The microcracks originated around the large particles and further propagated along the BC/substrate interface direction (Figure 4b). However, the alumina particles of smaller size (2–5 μm) that were spread at the interface between the BC and substrate in sample A did not result in noticeable delamination of the interface or the onset of cracks, as visualized in Figure 4a.

Figure 5 and Figure 6 show the cross-section microstructure of samples A and B obtained by SEM and corresponding EDS mapping. It was demonstrated that the TBCs had a multi-layer structure, as evidenced by the well-defined compositional boundaries between TC, BC, and substrate. At the TC/BC interface, the oxide scale was continuously formed, the composition of which was identified to be mainly Al₂O₃ from the EDS mapping. Outside of the continuous TGO layer at the TC/BC interface, an oxide scale was also observed inside BC, which may be due to oxygen penetration from TC into BC through open voids. The voids were preferentially formed at the intersection of splats during the deposition of BC [22], which offered internal oxygen diffusion pathways at high temperatures and accelerated the growth of the oxide scale inside BC [23], raising the possibility of coating spallation.

By contrast, a clear disparity in oxide size was seen within the BC section of the samples. Compared to sample A, sample B exhibited a higher percentage of oxide scale within the BC. Furthermore, the oxide scale expanded vertically up to the interface between the BC and substrate, indicating a more pronounced level of internal oxidation. Figure 7 demonstrates that the local oxide scale in BC had a composite structure, with an exterior layer of Al₂O₃ and an inner layer heavily containing phases of Co and Ni. In addition, some alumina particles with edged morphology were observed at the BC/substrate interface, identified as Al₂O₃ from EDS elements mapping. As shown in Figure 8, the origin and propagation of microcracks were observed in the vicinity of large alumina particles but not in the vicinity of small particles, suggesting the effect of grit particle size on the degradation microstructure of TBCs during thermal cycling. It can be concluded that the larger the alumina particle size, the greater extent of internal oxidation is developed inside BC.

3.4. Subsurface Hardening Induced by Alumina Particles

Figure 9 shows the optical images and corresponding hardness cartography obtained from a 7 × 3 indentations array around the alumina particles with different sizes. The step size for the nanoindentation test was ~40 μm. A hardness gradient was distinguished in the vicinity of the particles, where the hardness decreased progressively away from the interface to the interior. With an increase in particle size, the hardness gradient showed a larger magnitude, and the impacted zone expanded proportionally. The hardness gradient was within the range of 5.8 GPa to 4.0 GPa for the large particle measuring 130 μm in length, 5.1 GPa to 4.0 GPa for the medium particle measuring 60 μm, and 4.6 GPa to 4.0 GPa for the small particle measuring 30 μm. The hardness gradient would have important consequences for the thermal fatigue performance of TBCs. First, the deformation of the substrate induced by the impacting particles would increase the dislocation density and promote atomic diffusion during thermal cycling [24]. Furthermore, considerable compressive stress was introduced in the substrate subsurface. Thermal cycling relaxes compressive stress by inducing interface rumpling, a crucial process for interface oxidation [25,26].

4. Discussion

4.1. Effect of Alumina Particles on the TGO Growth

The grit-blasting treatment is a highly efficient method for augmenting the surface roughness of the substrate and improving the adherence of the applied coating [6,8,10]. However, the process unavoidably leaves contaminants from the blasting abrasive particles on the substrate surface. With varying blasting processing parameters, the size and mass of particles retained on the substrate surface differ [27,28,29]. The integrity of the TBCs is closed related to the grit-blasting parameters. For the sample A grit-blasted at low angles, the abrasive particles graze the substrate surface via micro-cutting action, forming shallow surface grooves features (Figure 2). Resultantly, the particles seem to be adhered on the surface in the cross-sectional SEM images of as-deposited TBCs for the sample A. When the substrate grit-blasted at high angles, the particles that impact the surface tend to form crater-like features (Figure 4). The large blasting pressure contributes to the large depth at which the particles embedded in the sample B, resulting a higher surface roughness than the sample A. As the cooling process during thermal cycling causes thermal expansion mismatch with the coatings, it is expected that the presence of foreign retained blasting particles would have a negative impact on the thermal fatigue behavior at the coating/substrate interface [9]. However, the detailed process of how the grit particles degrade the interfacial microstructure of TBCs during thermal cycling has not yet been fully explored. It is well known that the evolution of TGO during thermal cycling is the critical factor leading to the failure of the TBCs [30,31], so the effect of particle size on the growth of TGO should be considered.

As shown in Figure 1, the alumina particles in sample B are larger and more than those in sample A. Rapid cooling from high temperatures would impose compressive stress on the TGO layer and possibly tensile stress on the concave positions of the undulated TGO layer [32,33]. Due to the thermal mismatch, alumina particles at the BC/substrate interface would induce additional stress on the BC layer, which is further passed on to the TGO layer and changes the stress intensity and distribution [34]. The larger the particle size, the higher elastic strain energy is undertaken at the interface, leading to a larger stress imposed on the TGO layer. Owing to the low toughness [35], microcracks possibly originate at the concave position of TGO layer where the stress is locally concentrated [33]. As a result, the oxygen can go through the microcracks to react with the metal elements in BC, leading to internal oxidation inside BC. The presence of interconnected voids also facilitates the microcracks propagation in BC, which is evidenced by the observation that the nucleation and growth of oxide scale are favored at sites of profuse voids (Figure 7). Consequently, the microcracks in sample B grow vertically up to BC/substrate interface, causing the degradation of the interfacial microstructure of TBCs.

4.2. Effect of Alumina Particles on the Interfacial Microstructure

The SEM observations reveal that the BC/substrate interface prefers to detach in the vicinity of embedded large alumina particles. It is widely known that during thermal cycling, the embedded alumina particles can act as stress raisers favoring crack initiation, especially at the angular positions [36]. From the results, it is reasonable to deduce that there is a critical particle size beyond which the interface cannot tolerate strain incompatibility and stress mismatch, leading to the initiation of cracks near particles and subsequent crack propagation along BC/substrate interface. When the grit particles are small (~20 μm), the strain incompatibility can be accommodated through interface modulation. Resultantly, microcracks in the vicinity of alumina particles are not observed. As shown in Figure 8, the large compressive stress is constrained in the substrate around large particles, as indicated by the greater extent of substrate hardening. The relaxation of the stored stress during thermal cycling can induce interface rumpling and accelerate the interface oxidation or delamination.

It should be noted that microcracks in the vicinity of alumina particles not only propagate along the BC/substrate interface but also propagate into the surrounding BC or substrate deeply, which is associated with the vertical distribution of tensile stress in TBCs during thermal cycling [33,37]. As thermal cycling proceeds, the oxygen can diffuse into the cracks easily, forming oxides within the cracks. The formation of the brittle oxides in BC and along the BC/substrate interface would significantly increase the propensity of coating spallation, which is consistent with the result of Figure 2. In sample A with small alumina particles, the microcracks nucleation around alumina particles is not observed, suggesting that the stress concentration level reduces with decreasing the particle size during thermal cycling. As a result, the bonding interface between BC and substrate in sample A is virtually intact with little oxide scale or microcracks. After grit-blasting, another contribution to interfacial bonding strength may arise from the surface deformation features. The scratching characteristics that arise from the substrate distortion caused by the cutting of abrasive particles are considered more advantageous for the coating adhesion than the crater-like characteristics. This is because the scratched grooves provide a superior mechanical interlocking effect between the coating and the substrate [38].

5. Conclusions

The Mar-M247 Ni-based superalloy substrate samples with dimensions of 20 mm × 60 mm × 5 mm was grit-blasted with a series of processing parameters and then deposited with a metallic bond coat (BC) and top ceramic coat. The grit particles embedded in the BC/substrate interface play a detrimental role by acting as stress raisers favoring microcracks nucleation. In this work, the thermal fatigue behavior of the thermal barrier coatings (TBCs) having different grit particles at the BC/substrate interface was investigated by scanning electron microscopy. We further compare the cross-sectional microstructure of TBCs after thermal shocking tests and explore the possible mechanism of the grit particles degrading the TBCs. The detailed conclusions are summarized below:

• As the abrasive particle size increases from ~20 μm to ~50 μm, the likelihood of cracks starting from the concave regions of the thermally grown oxide layer towards BC also increases. In the sample containing small grit particles (<20 μm), the fractures developed to a depth of approximately 10 μm within the BC. These cracks then extended vertically up to the interface between the BC and substrate by linking the empty spaces in the BC with larger grit particles with size of ~50 μm.
• When the particle size exceeds a critical threshold ranging from 30 μm to 50 μm, the interface between the alumina particles and bonding coat cannot withstand the strain incompatibility caused by thermal mismatch. This results in the initiation of cracks near the large particles of size > 50 μm. This is supported by the SEM observation that cracks arise and propagate clearly near large grit particles (~50 μm).

This work identifies the effect of grit-blasting abrasive particle size on the microstructure of thermal barrier coatings, which has not been explicitly elaborated before. These findings can provide valuable guidelines in controlling the grit-blasting particles characteristics, which is of great significance to the thermal properties of thermal barrier coatings.