1 Introduction

Thermal barrier coatings (TBCs) are used to protect aircraft turbine engines, allowing them to operate at higher temperatures and thus improve their efficiency. TBCs are advanced materials composed of ceramic oxides with refractory properties, applied as protective coatings to the hottest part of turbine engines [1, 2]. Typically, TBCs are manufactured using either atmospheric plasma spray (APS) or electron-beam physical vapor deposition (EB-PVD) techniques. Each processing technique produces different types and levels of porosity. EB-PVD coatings have inter-columnar gaps that can facilitate infiltration of molten particles. On the other hand, APS coatings do not display interconnected porosity, which significantly affects the infiltration kinetics [3].

Yttria-stabilized zirconia (YSZ) has been reported as the most widely used ceramic for TBCs in the aircraft industry due to its outstanding features as a thermal barrier material [4]. TBCs protect the metallic components of the turbine against physical degradation from foreign particles such as volcanic ashes (VAs) and serve as thermal and chemical insulators [5]. However, operating temperatures are increasing and will reach the point where YSZ is no longer viable [6, 7]. Also associated with increasing temperatures, VAs have been reported to infiltrate into the porous YSZ coatings, eventually causing the failure of TBCs [8]. This increase in turbine engine temperatures to improve its efficiency demands for alternative materials.

Materials based on rare-earth zirconates (REZ) with A2Zr2O7 stoichiometry, where A represents rare-earth elements, offer a number of exceptional attributes. These include a distinctive lattice structure, exceptional chemical stability, superior ionic conductivity, elevated coefficients of thermal expansion, low thermal conductivity, high photocatalytic activity, and superior melting points [9]. These characteristics give them wide applications potential in high-temperature TBCs for ceramic materials. Among these types of materials, gadolinium zirconate (GZO) emerges as an alternative material due to its outstanding features as a thermal barrier material, such as high melting point, high temperature phase stability, low thermal conductivity, high sintering resistance and, especially, high resistance to infiltration of molten Si-based particles. GZO presents a pyrochlore structure (Gd2Zr2O7), which exhibits a high degree of disorder structure and has been reported to contain several oxygen vacancies that act as phonon dispersion points, generating desirable properties for thermal isolation material. Additionally, several studies have shown that REZ materials exhibit high resistance to infiltration of calcium-magnesium–aluminum-silicates (CMAS) and other siliceous melts such as VAs. Mikulla et al. [10] prepared four GZO microstructures by EB-PVD, manipulating both the substrate temperature and the rotational speed. Among these microstructures, one exhibited a refined architecture characterized by diminished intercolumnar gaps and elongated plume arms, which resulted in a significant reduction of CMAS infiltration within a remarkable range of 56% to 72%. The presence of a garnet phase ((Gd,Zr,Ca)3(Mg,Al,Fe,Ti,Zr)2(Si,Al,Fe)O12), which manifested as a continuous layer interfacing with apatite ((Gd,Ca)4(Gd,Zr)6(SiO4)6O2) and fluorite (Gd,Zr,Ca)O1.x was recognized as a valuable reaction product that improves CMAS resistance, thus offering a promising avenue for enhanced material performance.

Furthermore, Chen et al. [11] studied the impact of nonstoichiometric Gd2-xZr2+xO7+0.5× on the mechanical characteristics and thermal conductivity of GZO. The study employed first-principles calculations and validated the findings through solid-state reaction methods. The research highlights the directional dependency of the material’s elasticity, which intensifies as the levels of Gd3+ and Zr4+ increase due to lattice distortion from the excessive presence of these elements. A moderate excess of Zr4+ increases the hardness of GZO, while an excessive amount of both Zr4+ and Gd3+ decreases it. The disordered occupation of lattice atoms amplifies the nonlinearity of lattice waves and increases phonon scattering with increasing concentrations of Gd3+ and Zr4+, resulting in decrease in the thermal conductivity.

These contributions focused on the search of materials to reduce the degradation of TBCs in aircraft engines by CMAS and VAs. Currently, Mexico hosts two active volcanoes exhibiting pyroclastic activity in one of the most densely populated regions of the country, namely Popocatepetl [12] and Colima [13]. In particular, the most recent pyroclastic activity has been observed in the Popocatepetl volcano [14]. The continued activity of these volcanic formations poses a consistent threat to the air traffic within the region, an issue of particular significance given Mexico’s pivotal role as a transportation corridor connecting North and South America. While previous studies have explored the infiltrative impact of VAs in YSZ TBCs deposited via APS [5], the influence of Colima VAs on GZO ceramics has only been investigated in the context of ceramic coupons [14].

This study aims to assess the rate at which VAs from Colima volcano infiltrate into GZO ceramic coatings deposited by atmospheric plasma spraying. Additionally, the formation of re-precipitated apatite-type compounds that hinder infiltration at the reaction front is examined. The study evaluates the formation of stable phases with a composition close to the nominal value as a function of the infiltration time and the microstructural characteristics of the coatings, including relative density. Furthermore, this work analyzes the microstructure and precipitated phases in the reaction front. In addition, a comprehensive analysis was carried out using DoE methodology to analyze the components influencing the infiltration process and their interrelation. Conditions that minimize the deterioration of the GZO coating while exposed to the high temperatures of the VAs melt were identified.

2 Experimental procedure

This section describes the processing conditions used for the deposition of the coatings. It includes information on the materials, equipment, and parameters used. In addition, the infiltration of the coatings with VAs, the materials and the parameters examined in the tests performed are detailed. The characterization techniques and equipment used to evaluate the physicochemical properties of the infiltrated coatings, microstructure, structure, and elemental analysis, including the methods used for sample preparation are also described. Additionally, a description of the factorial DoE analysis parameters used to assess the infiltration test results is provided.

2.1 Manufacture of free-standing thermal barrier coatings

For the fabrication of the thermal barrier coatings (TBCs), commercial GZO powders (Höganäs, Germany) were used as feedstock material and deposited on alumina (Al2O3) substrates, manufactured by powder pressing and sintering to obtain 2.5 cm diameter round coupons. An APS system (9 MB, Oerlikon Metco) was used to manufacture the free-standing TBC coatings. The powders were sprayed on Al2O3 substrates under the following conditions: current of 520 A, voltage of 75 V, spraying distance of 75 mm, two traverse speeds of 0.75 m/s and 1 m/s, powder feeding rate of 30 g/min, primary hydrogen carrier gas flow of 80 SCFH and secondary argon carrier gas flow of 15 SCFH.

2.2 Infiltration experiments

For the infiltration experiments, the geosciences department of the national autonomous university of Mexico (UNAM) provided volcanic ashes (VAs) from the Colima volcano. In this process, 20 mg of Colima VAs were applied per square centimeter of the ceramic surface. Subsequently, the samples were subjected to heat treatment at 1250 °C for 1 h and 10 h, with heating and cooling rates set at 10 °C/min. This temperature was chosen based on previous TGA–DSC analyses of these VAs [5], ensuring that the molten VAs infiltrate through the microstructures of the TBCs.

2.3 Materials characterization

Particle size characterization of the commercial GZO powder was carried out using a HELOS/BR-OM laser diffractometer (Sympatec, Germany) under standard dry condition. Micrographs of the cross-sections of the free-standing as-sprayed coatings were recorded with scanning electron microscope (SEM, JSM-7610F) coupled with secondary electron (SE) and back-scattering electron (BSE) detectors. Relative densities and TBCs’ thicknesses of the as-sprayed TBCs were determined using imaging software (ImageJ).

Common metallographic techniques were used to prepare the cross-sections of the infiltrated TBCs, which were then used to study the TBC system’s infiltration depths, reaction layers, and microstructures using SEM. The chemical compositions of the reaction products were measured by energy dispersive X-ray spectroscopy (EDS) coupled to the SEM. Micro-Raman spectroscopy measurements (Alpha 300RA, WiTec, Germany) were performed with a 355 nm laser excitation source, a constant laser power of 4.7 mW, an integration time of 60 s, and 2 accumulations to identify the reaction products and extent of infiltrated samples. The identification of the crystalline phases was performed using X-ray diffraction (XRD) employing a diffractometer Rigaku SmartLab with a copper anode (Cu Kα1 = 1.54 Å) operating at 40 kV and 44 mA, the scanning range in 2θ and scanning speed were from 20º to 70º and 10 º/min, respectively.

Statistical analysis of the experiments was conducted using Minitab® software to evaluate the influence of key input factors in GZO TBCs, such as relative density and infiltration time, on the depth of the melt infiltration across the TBC microstructure and the Ca:(Gd + Zr) ratio, which is associated with the stoichiometry of apatite compound formation at the reaction front. The analysis of variance (ANOVA) was carried out at 95% confidence level (α = 0.05), and the responses exhibited a correlation coefficient (R2adj) of > 70%. The analysis included main effect plots to examine the influence of the factors at the selected levels. Contour plots were used to identify trends in the responses individually, and finally, a simultaneous optimization approach of the two responses was analyzed using an overlaid contour plot.

3 Results

In this section, the results of the sequential characterization of the investigated materials are presented. Initially, the description of the microstructural and structural properties of the feedstock powders and coatings is presented. The next subsection provides a detailed description of the microstructural characteristics and a semi-quantitative elemental analysis of the reaction front of the coatings after the infiltration tests. A thorough examination of the phases identified in the infiltrated samples is included. Finally, an analysis of the influence of the TBCs infiltration test characteristics on Ap formation and infiltration depth is presented using statistical analysis.

3.1 Feedstock powder and thermal barrier coatings’ characterization

Figure 1 shows the microstructural and structural characteristics of GZO commercial powders. The SEM micrographs in Fig. 1a and b, show the microstructure of the particles, typical features of agglomerated and sintered feedstock powders can be observed. The morphology of most of the particles tends to be spherical, which is ideal for improving flowability during the spray process and promoting a homogeneous deposition rate. Particles with apple-like morphology can also be observed and some fractured agglomerates exhibit a hollow microstructure. The particle size distribution plot of Fig. 1c is characterized by percentiles d10 = 45.2 µm, d50 = 99.1 µm and d90 = 135.8 µm, which is in good agreement with the size of the particles observed in the micrographs and represents a suitable distribution for the coating fabrication using the APS process [15]. The XRD pattern for GZO powders is shown in Fig. 1d. The diffraction peaks at 2θ = 29.4° (222), 34.0° (400), 48.9° (440), 58.0° (622), and 61.0° (444) confirm the presence of the pyrochlore phase Gd2Zr2O7 (ICSD # 165813). The peak at 2θ = 37°, diffraction plane (331), is characteristic of the pyrochlore superstructure and indicates its ordering degree [14, 16]. No additional phases were identified in the diffraction patterns of the powders.

Fig. 1
figure 1

SEM micrographs showing the surface morphology of agglomerated and sintered particles at two magnifications (a, b), particle size distribution plot (c) and XRD pattern of the commercial GZO powders (d)

Full size image

The SEM images in Fig. 2a and b show the cross-sectional microstructure of the GZO TBCs deposited at a traverse speed of 0.75 m/s and 1 m/s, respectively. Both coatings exhibit the characteristic layered microstructure usually observed in thermal spray methods, as shown in Fig. 2c and d. This microstructure is formed by the buildup of splats, which are produced when particles in a molten state impact the substrate. In addition, the presence of lamellar and circular porosity, and some cracking can be observed in the microstructure, which is typically observed in TBCs deposited by APS. Porosity plays a key role as a stress concentrator, promoting cracks formation and subsequent fracture [17, 18]. However, in TBCs it can be beneficial in reducing thermal conductivity. At high operating temperatures, diffusion sintering mechanisms are activated, generating a denser ceramic that increases mechanical strength, while reducing its thermal insulation capacity [18]. The traverse speed of 0.75 m/s and 1 m/s produced coatings with average thickness of 228.61 ± 8.63 µm and 168.52 ± 10.05 µm, respectively. Higher traverse speed leads to a reduced number of particles reaching the substrate, hence yielding a thinner coating.

Fig. 2
figure 2

Cross sectional SEM micrographs recorded at different magnifications, of the free-standing GZO coatings deposited by APS at 0.75 (a, c) and 1 m/s (b, d)

Full size image

3.2 Microstructure of the infiltrated thermal barrier coatings’ cross sections

Figure 3 shows the cross-sectional SEM micrographs of the reaction layers of the infiltrated GZO coatings, with relative densities of 91.71% and 89.93%, during 60 min (Fig. 3a and b) and 600 min (Fig. 3c and d). The infiltrated regions present different morphologies of the reaction products as a function of infiltration time and relative density of the TBC. The infiltrated TBC’s microstructure shows the glass and the GZO zones. The reaction layers between the GZO and glass zones show the morphology of embedded globular and needle-like phases. The observed phase morphologies match those reported for ZrO2 and Ap, respectively [5, 14, 19]. These phases, Ap and ZrO2, reprecipitate forming a compact layer that stops infiltration. The reaction layer seals the infiltration channels of the TBC at different infiltration depths, depending on the infiltration time and relative density. As infiltration time increases from 60 to 600 min, the infiltration depth increases and a thicker reaction layer forms above the reaction front. A SEM image at high magnifications of the reaction layer of the infiltrated TBC with a relative density of 89.83% and 60 min, Fig. 3e, shows a closer representation of the microstructure of the reaction products, revealing the presence of globular zirconia and needle-like apatite phases.

Fig. 3
figure 3

Cross-sectional SEM images of the infiltrated free-standing GZO coatings with 91.7% and 89.9% of relative density after 60 min and 600 min of infiltration at 1250 °C, showing the reaction front (a, b, c and d). Enlarged image of (b) showing the microstructure of the reaction products (e)

Full size image

Figure 4a shows a SEM image of the cross-section of an infiltrated TBC (Fig. 3b, 60 min with 89.93% relative density). The volume of the dissolved TBC can be observed. Figure 4b shows a magnified SEM image of the infiltrated sample, which presents an infiltration depth of 48.48 µm, determined by image analysis (ImageJ). EDS mappings, Fig. 4c, reveal the distribution of the main elements involved throughout the reaction layer. Si is mainly distributed in the molten glass of the Colima VA; Zr and Gd elements are mainly observed in the TBC. However, the latter two elements are also observed in the reprecipitated ZrO2 and Ap phases [19, 20]. To confirm the presence of these phases, Tables 1 and  2 show the normalized chemical composition of Ap and ZrO2 phases, respectively, measured by EDS. The Ap phase, Ca2(Gd,Zr)8((Si,Al)O4)6O2 can incorporate cations of Zr4+ and Gd3+ in Al3+ and Si4+ sites, respectively [20]. For longer infiltration times, from 60 to 600 min, it is observed that TBCs with 91.71% relative density, the contents of Si, Ca and Zr tend to decrease, and meanwhile Gd amount tends to increase. However, the amount of Si shows the sharpest decrease, going from 61.24% to 54.74%. On the other hand, the relative density of the TBC increased from 89.83% to 91.71%. In the TBCs infiltrated during 600 min, the Si, Ca and Zr contents tend to increase, while Gd3+ contain tends to decrease.

Fig. 4
figure 4

Cross sectional SEM micrographs of an infiltrated GZO coating sample during 60 min with 89.93% relative density (a), a magnification of the micrograph in the reaction front zone and the corresponding EDS chemical compositional maps of Si, Zr and Gd (b)

Full size image
Table 1 Elemental compositions of Aps measured by point EDS
Full size table
Table 2 Elemental compositions of zirconias measured by point EDS
Full size table

Table 2 exhibits the normalized chemical compositions of the globular phases observed in the reaction layers of the TBC, where it is confirmed that these compositions match with those reported for cubic zirconia (c-ZrO2) [5, 19, 21]. It has been reported that the c-ZrO2 phase incorporates up to 16% of Gd3+ in solid solution [19]. Furthermore, it is observed that as infiltration time increases, the Gd3+ contained in c-ZrO2 slightly decreases around 3%.

The graph in Fig. 5, shows the average infiltration depths of TBCs as a function of time. Two stages of can be seen; stage I, which corresponds to time less than 60 min, shows fast infiltration rates, 0.30 µm/min and 0.80 µm/min for TBCs with relative densities of 91.71% and 89.83%, respectively. For times longer than 60 min, stage II, the infiltration rates decrease to 0.10 µm/min and 0.05 µm/min. At 600 min, TBCs with relative densities of 91.71% and 89.83% reach infiltration depths around 71 µm and 78 µm, respectively. These differences in the infiltration rates are due to the kinetic formation of the reaction products. In stage I, the melt penetrates along the TBC porosities dissolving the TBC components into Gd3+ and Zr4+, subsequently the melt becomes saturated with these cations and reaction products like Ap and c-ZrO2 begin to reprecipitate and gradually grow to seal the infiltration channels. In stage II, the melt continues to penetrate in the TBCs, but, the infiltration rates, 0.1 µm/min and 0.05 µm/min, decrease considerably due to the formation of more Ap and c-ZrO2 above the reaction front which contributes to the sealing of the infiltration channels. These phenomena will be discussed in detail later.

Fig. 5
figure 5

Plot of the infiltration depth kinetics of the GZO coatings with 91.7 and 89.8% relative density infiltrated with Colima volcano ashes at 1250 °C

Full size image

3.3 Phase identification of the thermal barrier coatings’ reaction layers

Figure 6a shows Raman spectra as a function of infiltration depth performed on the reaction layer of an infiltrated GZO TBC, with a relative density of 91.71% for 600 min. The Raman spectra shows bands of Ap, c-ZrO2, t-ZrO2, and GZO, confirming the presence of these phases along the reaction layer. At low infiltration depths (15 µm), bands of c-ZrO2, at 247 cm−1, 294 cm−1, 436 cm−1 and 709 cm−1; t-ZrO2, at 275 cm−1, 326 cm−1, 408 cm−1, 456 cm−1; GZO at 306 cm−1, 335 cm−1, 410 cm−1, 544 cm−1 and 610 cm−1 are observed [22,23,24]. However, a slight band of Ap is also present at 875 cm−1, which corresponds to the stretching vibrational mode of the Si–O tetrahedra in the Ap structure [25]. As infiltration depth increases, from 15 µm to 50 µm, the Ap band stands out progressively indicating a major amount of this phase. At infiltration depths higher than 50 µm, the zirconias and Aps phase bands disappear, with the GZO bands remaining, indicating the maximum infiltration depth reached. The band deconvolution of the Raman spectrum measured at 30 µm of infiltration depth is shown in Fig. 6b and exhibits the main vibrational modes of t-ZrO2, c-ZrO2, GZO and Ap [14, 22, 24].

Fig. 6
figure 6

Raman spectra as a function of infiltration depth of Colima VA with GZO coatings with 91.7% relative density infiltrated after 600 min (a) and band deconvolutions at 30 µm of infiltration depth (b). The main vibrational modes of t-ZrO2, c-ZrO2, Ap and GZO are identified

Full size image

3.4 Analysis of the factorial 22 design of experiment

This section examines the reaction products observed in the infiltrated ceramics. The distribution and morphologies of the precipitated phases were altered by the infiltration process of the REZ. To understand the interaction of the reaction products at the infiltration depth, various methodologies were used to study the properties of the reaction layer.

The performance of GZO TBCs to prevent failure during infiltration of Colima VAs was evaluated using DoE analysis, considering the relative density (%) of the coating and the infiltration time (min) at 1250 °C as input factors. The effect of the studied factors on the responses infiltration depth (µm), and the Ca:(Gd + Zr) ratio related to the formation of stable apatite compounds during the infiltration tests is discussed. In the main effect plots presented in Fig. 7a and b, it can be observed that the input factor with the most significant effect of the two studied responses was the infiltration time. Relative density, on the other hand, has the least pronounced effect on the infiltration depth, and it can be seen that the effect of the factors on the responses is the opposite. Figure 7d and e show, respectively the contour plots of the infiltration depth (µm) and Ca:(Gd + Zr) ratio responses, for the experimental region defined by the infiltration time (min) and the relative density of the coating (%). In the contour plots the response values can be predicted by a combination of the factors; and they can be individually optimized by setting the tendency to maximize or minimize. For the infiltration depth, the lowest values can be achieved with high relative densities and infiltration times, while the opposite trend is observed for the Ca:(Gd + Zr) ratio, where the range of interest has been reported to be 0.25 to 0.6 [26]. An overlaid contour plot (Fig. 7f) as a function of the studied factors has been obtained to visualize the feasible region (white) where the two responses are simultaneously analyzed and defined by setting low and high limits shown by dashed and solid lines, respectively. The infiltration depth and Ca:(Gd + Zr) ratio were optimized in the ranges from 45 μm to 70 μm and μm 0.25 to 0.4 μm, respectively. The local optima of the input factors that allow mitigating the degradation of the coating indicate the conditions under which the depth of the melt penetration is restricted, and the formation of stable apatite phases is promoted. In the optimizer response of Fig. 7c can be visualized the effects of the relative density and infiltration time on the ca:(Gd + Zr) ratio and infiltration depth, and the combination of factor levels that result in the optimal response, which was defined by setting the objectives of maximizing or minimizing the response variable. By analyzing the feasible region of the overlaid contour plot with the multiple response optimizer tool, it was predicted that in a TBC with a relative density of 89.8%, the most effective Ap to stop infiltration reprecipitates at an infiltration depth and time of 66.3 µm and 338 min, respectively, and exhibits a Ca:(Gd + Zr) of 0.33, which is closer to the nominal composition of Ap [26]. As observed in the response optimizer (Fig. 7c), the achieved desirability for single and composed responses was 1, indicating that the defined levels allow achieving the predicted responses.

Fig. 7
figure 7

Main effect plots (a, b) and contour plots (d, e) of the analyzed responses: Ca:(Gd + Zr) and Infiltration depth as a function of the factors: Infiltration time and relative density. Overlaid contour plot (f) showing the experimental region, feasible zone, and optimized condition for the input factors, determined with the response optimizer (c)

Full size image

4 Discussion

In this section, the influence of the reaction products on the infiltration kinetics of GZO TBCs is discussed. In Fig. 5, at stage I (< 60 min), there is a high infiltration rate, compared to stage II, of both manufactured TBCs, presenting infiltration rates of 0.3 µm/min and 0.8 µm/min for TBCs with relative densities of 91.71% and 89.83%, respectively. These high rates at stage I are due to the dissolution, diffusion, and reprecipitation reactions taking place [21, 27]. Initially, the melt dissolves the TBC, and its main constituents, Gd3+ and Zr4+, diffuse away from the reaction into the melt; however, as the infiltration continues, the melt becomes saturated and reaches local concentrations to promote the crystallization of zirconia and Aps, which gradually and partially seal the infiltration channel [21, 27, 28]. Therefore, these differences in the relative densities of TBCs directly influence the flowability of the melt through their porous structures, with the low relative density (89.83%) penetrating easily compared to high relative density TBCs (91.71%). On average, a major amount of melt penetrates the low relative density TBCs; however, the local saturation in the melt to promote the reprecipitation of the reaction products depends on the amount of dissolved cations in the TBCs, which may vary due to the relative density of the TBCs and the infiltration depth reached. In stage II, the infiltration rates decrease to 0.10 µm/min and 0.05 µm/min for high and low relative density GZO TBCs, respectively. These changes in infiltration rates occur because an significant amount of Ap and zirconia reaction products have already crystallized above the reaction front at the first 60 min; therefore, a large amount of these reaction products prevent high infiltration rates [21] due to the partial sealing of the reaction front in the TBCs. Nonetheless, after 600 min of infiltration time, both samples present similar values of infiltration depths, around 75 µm and 80 µm; this indicates that the initial relative density of the TBC, infiltration depth, and time influence the formation of reaction products to stop the degradation process.

The Ap phase has been reported to be the main reaction product stopping infiltration in REZ-based TBCs [14, 27]. The formation of a stable Ap depends on several factors, such as RE3+ cation size, Ca:(Gd + Zr) ratio, etc. It has been reported that Aps exhibiting the nominal Ca:Gd around 0.25 have is the most stable Ap in GZO TBCs [26], representing a defect-free stoichiometry in the Ap structure [29]. In this sense, understanding the formation of a stable Ap in GZO is relevant to in turn understanding its connection with the infiltration kinetics. Figure 7 shows the influence of infiltration time and initial TBC’s relative density on the Ap stability ratio and infiltration depth. Figure 7a shows that the reprecipitation of a stable Ap is favorable with longer infiltration time, while it is unfavorable with TBCs with higher relative density. On the other hand, the infiltration depth increases with both longer times and higher relative densities. Figure 7f shows the optimal region in which stable Ap with a Ca:(Gd + Zr) ratio greater than 0.4 at infiltration depths below 70 µm. The area of this region tends to decrease with increasing relative density and infiltration time. Statistical analysis predicts that, in a GZO TBC with a relative density of 89.8%, a stable Ap with a Ca:(Gd + Zr) ratio of 0.33 is reprecipitated after 338 min at an infiltration depth of 66.3 µm. However, higher relative density TBCs (91.2%—91.7%) require longer infiltration times to reprecipitate stable Aps with a lower Ca:(Gd + Zr) ratio compared to lower relative density TBCs (89.8%—90.3%), which require shorter infiltration times to promote the reprecipitation of a stable Ap that effectively stops infiltration. These results indicate that relative density and infiltration time play an important role in promoting the crystallization of stable Aps due to the different amounts of Gd3+ cations dissolved in the melt.

5 Conclusions

Free-standing GZO TBCs deposited by APS, showing different microstructural characteristics such as densification level and thickness as a function on the deposition rates were used for infiltration experiments.

Compared to YSZ, the GZO TBCs show excellent resistance to infiltration of molten Colima VAs, and present infiltration depths around 75 µm and 80 µm. On the other hand, the infiltration kinetics show two infiltration rate regimes, fast infiltration rates during the first 60 min and slow infiltration rates between 60 to 600 min of infiltration time. These changes in infiltration rates are attributed to partial sealing of the channels due to the formation of stable apatite phases.

Applying a systematic DoE methodology, the impact and correlation between key input parameters of the infiltration TBCs were investigated to understand their interactions with the Colima VA. The interaction between microstructural characteristics of GZO TBCs and apatite formation was studied to identify the best conditions for stopping the infiltration of molten ash. In the GZO coatings of this work, the stable Ap phase reprecipitates at infiltration depths around 66.3 µm after 338 min of Colima volcano ash infiltration time.