Ytterbium Disilicate/Monosilicate Multilayer Environmental Barrier Coatings: Inﬂuence of Atmospheric Plasma Spray Parameters on Composition and Microstructure

: SiC/SiC ceramic matrix composites (SiCf/SiC CMCs) are regarded as the new materials for the hot-section components of aircraft gas turbine engines, since they have one-third of the density of metallic superalloys, a higher temperature capability, good mechanical strength, and excellent thermal shock resistance. However, high-temperature water-vapor-rich combustion gases can induce severe surface recession phenomena in SiC/SiC leading to component failure. For this reason, it is necessary to design protective coatings, i


Introduction
Innovation in aeronautical engines requires new materials able to meet several requirements such as low density, high thermomechanical properties, and high-temperature corrosion resistance.The ever-increasing demand for an improvement in thermodynamic efficiency and a reduction in fuel consumption requires higher temperatures in combustion chambers and, as a consequence, higher gas turbine inlet temperatures, as well as engines of lighter weight [1,2].The achievement of these goals is quite challenging since it requires continuous technological improvements regarding materials, coatings, and cooling systems.
Superalloy-based components reached their limit in terms of thermomechanical performances, even if protected by cooling systems and thermal barrier coatings (TBCs) [3,4].For this reason, there is an increasing interest in high-performance ceramic materials able to meet highly demanding performance criteria.Obviously, low fracture toughness is an important issue for ceramics, and only a selection of them are suitable for this kind of application.The materials selected for such components are ceramic matrix composites (CMCs) [5][6][7].In particular, CMC materials used in aeronautical gas turbines engines consist of silicon carbide (SiC) fibers embedded in a SiC matrix (SiC f /SiC CMCs), and they are characterized by low density (one third compared with metallic superalloys), good mechanical properties at high temperature, oxidation resistance in high-temperature oxidizing As a consequence, the YSZ layer was replaced by a compound of barium-strontiumaluminosilicate (BSAS), since this material showed a better thermomechanical compatibility with the SiC substrate [30] and a good resistance to water vapor [22,31].The result was a more crack-resistant coating with higher durability than YSZ in a combustion environment [23,27].However, even the top coat made of BSAS revealed some critical issues like the volatilization of the material in high-velocity combustion environments [29].
Then, at the beginning of the 2000s, several studies observed that some rare-earth (RE) silicates (e.g., Y, Yb, Gd, Er, and Lu) showed better chemical stability than BSAS in combustion environments, as well as greater resistance to high temperatures, leading to the current development of RE silicates-based advanced EBC systems [22][23][24]32,33].
Rare-earth silicates have been the object of great interest because of their high-temperature stability and water vapor corrosion resistance.Moreover, these materials can form various phases with different properties suitable for several applications such as thermal and environmental barrier coatings for silicon-based ceramics or thermal barrier coatings (TBC) for Ni-based superalloys [34].For instance, X2-RE 2 SiO 5 , γ-RE 2 Si 2 O 7 , and β-RE 2 Si 2 O 7 are potential candidates to protect silicon-based ceramics [35], while the high coefficient of thermal expansion of RE 9.33 (SiO 4 ) 6 O 2 is of great interest for superalloys [34,36].
In particular, rare-earth monosilicates (RE 2 SiO 5 ) and rare-earth disilicates (RE 2 Si 2 O 7 ) are the most promising compounds for environmental barrier coatings.
RE disilicates have the advantage of a lower CTE, closer to that of the SiC substrate, but also the disadvantage of a lower stability when exposed to water vapor at high temperature.Furthermore, most of the RE 2 Si 2 O 7 show polymorphism in the temperature range of 25-1700 • C (Figure 1); thus, the volumetric variations due to the phase transitions can induce deleterious residual stress inside the microstructure [37,38].
Coatings 2023, 13, x FOR PEER REVIEW 3 of 16 Ytterbium stabilized zirconia (YSZ) was selected as a protective top layer for mullite because of its good stability in water vapor.However, its CTE (10 × 10 −6 K −1 ) was almost twice that of SiC or mullite, and these multilayer coatings showed severe cracking and delamination after a few hundred hours at 1300 °C [25,29].
As a consequence, the YSZ layer was replaced by a compound of barium-strontiumaluminosilicate (BSAS), since this material showed a better thermomechanical compatibility with the SiC substrate [30] and a good resistance to water vapor [22,31].The result was a more crack-resistant coating with higher durability than YSZ in a combustion environment [23,27].However, even the top coat made of BSAS revealed some critical issues like the volatilization of the material in high-velocity combustion environments [29].
Then, at the beginning of the 2000s, several studies observed that some rare-earth (RE) silicates (e.g., Y, Yb, Gd, Er, and Lu) showed better chemical stability than BSAS in combustion environments, as well as greater resistance to high temperatures, leading to the current development of RE silicates-based advanced EBC systems [22][23][24]32,33].
Rare-earth silicates have been the object of great interest because of their high-temperature stability and water vapor corrosion resistance.Moreover, these materials can form various phases with different properties suitable for several applications such as thermal and environmental barrier coatings for silicon-based ceramics or thermal barrier coatings (TBC) for Ni-based superalloys [34].For instance, X2-RE2SiO5, γ-RE2Si2O7, and β-RE2Si2O7 are potential candidates to protect silicon-based ceramics [35], while the high coefficient of thermal expansion of RE9.33(SiO4)6O2 is of great interest for superalloys [34,36].
In particular, rare-earth monosilicates (RE2SiO5) and rare-earth disilicates (RE2Si2O7) are the most promising compounds for environmental barrier coatings.
RE disilicates have the advantage of a lower CTE, closer to that of the SiC substrate, but also the disadvantage of a lower stability when exposed to water vapor at high temperature.Furthermore, most of the RE2Si2O7 show polymorphism in the temperature range of 25-1700 °C (Figure 1); thus, the volumetric variations due to the phase transitions can induce deleterious residual stress inside the microstructure [37,38].Ytterbium disilicate has the great advantage of inherent phase stability up to the melting point and good chemical and mechanical compatibility with SiCf/SiC and Si substrate [40]; thus, in this work, it has been selected and investigated as a candidate for an EBC system.Rare-earth monosilicates are more stable at high temperature when exposed to water vapor thanks to their higher resistance to the formation of volatile products.However, the high CTE mismatch between RE2SiO5 and the substrate leads to crack formation when exposed to high temperature, allowing for the diffusion of oxygen and water vapor Ytterbium disilicate has the great advantage of inherent phase stability up to the melting point and good chemical and mechanical compatibility with SiC f /SiC and Si substrate [40]; thus, in this work, it has been selected and investigated as a candidate for an EBC system.Rare-earth monosilicates are more stable at high temperature when exposed to water vapor thanks to their higher resistance to the formation of volatile products.However, the high CTE mismatch between RE 2 SiO 5 and the substrate leads to crack formation when exposed to high temperature, allowing for the diffusion of oxygen and water vapor throughout the coating.RE 2 SiO 5 can crystallize in two monoclinic polymorphs depending on the ionic radius of the rare-earth element (RE 3+ ): the X1 phase is characteristic of RE elements with a higher ion radius and thus for lanthanides from Gd to La, while the X2 phase is the crystal structure typical of RE with a lower ion radius (from Lu to Dy).RE 2 SiO 5 compounds based on Tb and Y can crystallize in both phases, depending on the crystallization temperature: X1 for lower temperature, and X2 for higher temperature [34,41,42].The presence of a single stable phase in the temperature range of interest is an interesting property that Coatings 2023, 13, 1602 4 of 15 facilitates the avoidance of volumetric variation due to polymorphism.The architecture of coatings investigated in this work was designed considering the features and the behavior of previously mentioned silicate materials when exposed to harsh environments.In this experimentation, a multi-layer coating was designed and manufactured with three different layers with specific protective functions: (i) a Si-based bond coat whose function is to protect the substrate from oxidation and improve the adhesion of the upper layer to the SiC substrate; (ii) an intermediate layer made of ytterbium disilicates (Yb 2 Si 2 O 7 ), which improves the thermomechanical compatibility and reduces the volatilization of Si-based compounds; (iii) a top coat based on ytterbium monosilicates (Yb 2 SiO 5 ) increasing the thermal insulation and improving high-temperature stability in the presence of water vapor and resistance to CMAS attack (mixtures of CaO-MgO-Al 2 O 3 -SiO 2 ).
EBC multilayer coatings were deposited by using the atmospheric plasma spray technique (APS), which is already widely chosen for the deposition of thermal barrier coating systems.This strategy is of great interest, because it allows for the manufacturing of EBC employing facilities already used for thermal barrier coatings with considerable costs saving.
Numerous research works have studied ytterbium monosilicate and disilicate as protective layers for EBC [17,24,40,[43][44][45][46], but only some studies have presented the possibility of obtaining a tri-layer coating using Yb 2 Si 2 O 7 /Yb 2 SiO 5 /Si [47][48][49]; Xin et al. [50] presented a work in which this tri-layer coating is deposited via APS.In the present research, the multilayer coating Yb 2 Si 2 O 7 /Yb 2 SiO 5 /Si was deposited by using APS with three different torch powers, and the influence of the deposition parameters on the coating properties was deepened.In particular, crack density was deeply analyzed because it can compromise the in-service performances and durability of coatings, leading to a decay in thermal and thermomechanical performances [51].Moreover, through-the-thickness cracks in the outer layer of the system provide a channel of access for oxidative and corrosive substances coming from combustion environments, like oxygen, water vapor, and corrosive impurities from fuel, thus reducing the durability of protective coatings.Thus, it is of great importance to carefully optimize the APS deposition parameters in order to reduce the presence of these defects, which have negative influences on the performances of these systems.Furthermore, the porosity and the chemical composition of the coatings were also studied.In fact, the presence of high porosity can facilitate the coalescence of cracks and the failure of the coating.The APS process produced thermal-sprayed coatings with a chemical composition different from the initial powders, and it is important to understand the evolution of the compounds in the final coatings.

Materials and Methods
Powders of the raw materials suitable for the atmospheric plasma spray technology were provided by several suppliers: silicon powders (particle size: 75/20 µm) were supplied by H.C. Starck (Goslare, Germany), while fused and crashed powders of Yb 2 Si 2 O 7 (d 50 = 35.4µm) and Yb 2 SiO 5 (d 50 = 34.9µm) were provided by Treibacher Industrie AG (Althofen, Austria).
The deposited and investigated EBC architecture is a tri-layer coating, with silicon as the bond coat, Yb 2 Si 2 O 7 as the intermediate layer (YbDS), and Yb 2 SiO 5 as the upper layer (YbMS), deposited onto sintered silicon carbide (SiC) substrates provided by IPS Ceramics (Newcastle-under-Lyme, UK).SiC substrates are considered adequate for this study because they guarantee the same chemical compatibility as, and a similar CTE to, a SiC/SiC CMC substrate.The deposition process was carried out by Borga Meccanica S.R.L. located in Chiarano (TV), Italy, using an atmospheric plasma spray (APS) (Oerlicon Metco, Wohlen, Switzerland) technique in a facility equipped with an F4-MB torch (Oerlikon Metco, Wohlen, Switzerland).
Substrates were grit blasted using 60 µm SiC grit provided by Florence Abrasives & Tools s.r.l.(Florence, Italy); then, they were cleaned in an ultrasonic bath with ethanol to remove surface contamination and improve the mechanical anchorage of the coating.
Thermal spray parameters deeply affect the composition, microstructure, and defects of the coating layers: in the present work, starting from the literature [14,24,43,46,[51][52][53][54], an experimental array was developed with three different torch powers.For each layer (Si, YbDS and YbMS), the feeding rate (40 g/s) and the number of spraying passes (10) were kept constant and then arranged to obtain the desired thickness.A processing issue for silicate EBCs is the deposition of stoichiometric coatings, since during the spraying process, there is a preferential volatilization of SiO 2 from molten particles; thus, Si-depleted coatings are obtained when compared with the starting powders.Obviously, the higher the temperatures and/or the longer the exposure times of the particles, the more intense is the effect of SiO 2 volatilization.For this reason, coatings obtained using APS show different microstructures, compositions and porosities depending on the process parameters [52].Furthermore, RE silicates form an undesired amorphous phase at solidification temperatures lower than 1000-1200 • C [14,24,35,52].
For this reason, the multi-layer coatings were deposited by heating substrates with the plasma plume: a special pre-heating program was used in order to reach a temperature of approximately 950-1050 • C.During this process, the temperature of the samples was monitored using a K-type thermocouple placed below the substrates.
In order to obtain high-performance coatings, the deposition process required an optimization of the thermal spray parameters: torch power, stand-off distance and strategy of deposition.
The power of the torch can be adjusted according to the current intensity, the voltage, and the gases' flow rate.An increase in torch power can also lead to an increase in the plasma plume's velocity, with a consequent reduction in the particles' residence time in the plasma [55].The variations in temperature and in the plasma plume's velocity in turn affect the heating rate of the particles and, in the case of silicates, can also influence the evaporation of SiO 2 .
For this experimentation, the primary plasma gas (Ar) flow rate and current intensity of the torch were taken as constant, while three different torch powers were set, varying the secondary plasma gas, H 2 .The stand-off distance was kept constant and equal to 135 mm for the silicate layers.In Table 1, the matrix of the deposition parameters adopted for each material is reported.The Si bond coat was sprayed with a torch power of 25 kW and a stand-off distance of 100 mm: these parameters allow for the deposition of ~150 µm of Si with good cohesion.This bond coat was then deposited on all the SiC substrates.The silicate layers were first deposited with 10 passes.Then, according to the different thickness per pass (Table 1), a number of passes was selected to obtain thicknesses of approximately 150 µm.
EBC samples were mounted in resin and polished in order to be observed with a scanning electron microscope (Tescan Mira 3, Brno, Czech Republic) equipped with an EDS detector (EDAX/Ametek Inc., Pleasanton, CA, USA).BSE micrographs and EDS analyses allowed observation of the microstructure and phase distribution inside the silicate layers.Porosity was measured by observing samples with an optical microscope (Nikon Eclipse L150) and using the image analysis software LUCIA Measurements (v.4.80 Laboratory Imaging S.r.o.Praha, Czech Republic).XRD analyses were performed by using a Philips X'Pert X-ray device (PANalytical B.V., Almelo, The Netherlands), operating at 40 KV and 40 mA with CuKα1 radiation.The samples were scanned with a scan range of 20 • -80 • , a step size of 0.02 • , and a counting time of 2 s.
Crack analysis was performed using an appropriately developed image processing and analysis MATLAB routine.First, the region of interest (ROI) is selected from the original SEM picture (Figure 2a).The selected ROI is made binary according to a proper color threshold in the greyscale.This step allows the identification of cracks and porosities, which turn black and are easily distinguished from the rest of the layer under analysis, which is represented in white (Figure 2b).The image is then processed to differentiate porosities from cracks, according to their shape: porosities are removed and considered as white background so that only cracks in black are left for analysis (Figure 2c).By means of manual selection of their apex, cracks are identified (red lines in Figure 2d) and their length is calculated.The routine output is reported in Figure 2e, where red lines corresponding to cracks are superimposed on the original SEM picture, in order to prove effective matching.Several parameters can be calculated from the image processing and analysis process: the number of cracks; their mean length (1); crack density (2); overall specific crack length (3).

Mean length
Crack density Overall specific crack length mm mm 2 = where A is the surface area of the analyzed coating cross-section, n is the total number of cracks, and L i is the length of each identified crack.

Microstructural Characterization: SEM Micrographs and EDS Analysis
The raw materials were characterized by performing scanning electron microscopy (FE-SEM Tescan Mira-3) and energy dispersive X-ray spectrometry (EDS) to evaluate their morphology and their composition.Then, X-ray diffraction (XRD) (Philips X'Pert X-ray device (PANalytical B.V., The Netherlands)) analyses were performed to determine their phase composition.Yb 2 Si 2 O 7 powders have impurities of silicon (Figure 3a), while Yb 2 SiO 5 ones have impurities of silicon and Yb 2 O 3 (Figure 3b).However, these compounds were not detected by XRD measurements in ytterbium disilicate powders; the spectrum presents only diffraction peaks belonging to monoclinic Yb 2 Si 2 O 7 (C2/m, 25-1345) (Figure 3c).By contrast, the XRD analysis for the ytterbium monosilicate powders shows the presence of monoclinic Yb 2 SiO 5 (I2/a, 40-0386) and of SiO 2 and Yb 2 O 3 , which are residue precursors for the formation of ytterbium monosilicate (Figure 3d).Si was identified by the SEM micrographs, but it was not detected by XRD analysis, probably because it was in an amorphous state.micrographs, but it was not detected by XRD analysis, probably because it was in an amorphous state.All the deposited samples were characterized by using SEM and EDS analysis in order to observe their microstructure and to assess the composition and distribution of the different phases in the coatings.As a first step, only the YbDS coatings were deposited in order for us to understand the deposition efficiency for each program and adjust the number of passes required to obtain a comparable thickness.These coatings were also used for the XRD analysis.Then, the multi-layer coatings (YbDS and YbMS) were deposited.
Figure 4 shows the cross-section of the YbDS coatings deposited with a stand-off distance of 135 mm and different torch powers.The comparison of Figure 4a,c,e shows the lower deposition efficiency for sample DS-16, while the higher magnification micrographs in Figure 4b,d,f show the difference in the cracks' morphology: cracks are wider in sample DS-20 and lighter in sample DS-16.In these samples, areas having different gray contrast were investigated using EDS (Figure 5): as discussed in several studies [24,52,56], dark splats have a composition closer to the powder feedstock, while bright splats show a lower Si content.This observation confirms an expected phenomenon: during the deposition process, the SiO2 partially volatilizes, and its depletion promotes, according to the phase All the deposited samples were characterized by using SEM and EDS analysis in order to observe their microstructure and to assess the composition and distribution of the different phases in the coatings.As a first step, only the YbDS coatings were deposited in order for us to understand the deposition efficiency for each program and adjust the number of passes required to obtain a comparable thickness.These coatings were also used for the XRD analysis.Then, the multi-layer coatings (YbDS and YbMS) were deposited.
Figure 4 shows the cross-section of the YbDS coatings deposited with a stand-off distance of 135 mm and different torch powers.The comparison of Figure 4a,c,e shows the lower deposition efficiency for sample DS-16, while the higher magnification micrographs in Figure 4b,d DS-20 and lighter in sample DS-16.In these samples, areas having different gray contrast were investigated using EDS (Figure 5): as discussed in several studies [24,52,56], dark splats have a composition closer to the powder feedstock, while bright splats show a lower Si content.This observation confirms an expected phenomenon: during the deposition process, the SiO 2 partially volatilizes, and its depletion promotes, according to the phase diagram (Figure 1), the formation of Yb 2 SiO 5 .Furthermore, the observation of the BSE micrographs lightened the appearance of several shades of gray.These variations are related to the presence of splats with different amounts of very finely dispersed Yb 2 Si 2 O 7 and Yb 2 SiO 5 phases, as already observed in other studies [14,24,40,56].The quantity of volatilized SiO2 influences the composition of the final coating; for this reason, EDS quantitative analyses were carried out considering the atomic content of Yb and Si and calculating the ratio Si/(Si + Yb).The EDS analyses were carried out on areas of approximately 100 × 400 µm: for each coating, 10 areas were evaluated along all the sample, and then, the average values and the standard deviation were calculated.The results are reassumed in the histograms of Figure 6: the content of Si is very similar for the three YbDS coatings, even where it is observed that the higher the torch power is, the lower is the Si content.This means that, for YbDS, the SiO2 volatilization is more relevant for higher torch powers, and this phenomenon promotes the formation of phases with a lower Si content, like ytterbium monosilicate and ytterbium oxide.On the other hand, for YbMS, the behavior is in countertrend, so the higher the torch power, the higher the Si content.The quantity of volatilized SiO2 influences the composition of the final coating; fo this reason, EDS quantitative analyses were carried out considering the atomic content o Yb and Si and calculating the ratio Si/(Si + Yb).The EDS analyses were carried out on areas of approximately 100 × 400 µm: for each coating, 10 areas were evaluated along all the sample, and then, the average values and the standard deviation were calculated.The re sults are reassumed in the histograms of Figure 6: the content of Si is very similar for the three YbDS coatings, even where it is observed that the higher the torch power is, the lower is the Si content.This means that, for YbDS, the SiO2 volatilization is more relevan for higher torch powers, and this phenomenon promotes the formation of phases with a lower Si content, like ytterbium monosilicate and ytterbium oxide.On the other hand, fo YbMS, the behavior is in countertrend, so the higher the torch power, the higher the S content.The quantity of volatilized SiO 2 influences the composition of the final coating; for this reason, EDS quantitative analyses were carried out considering the atomic content of Yb and Si and calculating the ratio Si/(Si + Yb).The EDS analyses were carried out on areas of approximately 100 × 400 µm: for each coating, 10 areas were evaluated along all the sample, and then, the average values and the standard deviation were calculated.The results are reassumed in the histograms of Figure 6: the content of Si is very similar for the three YbDS coatings, even where it is observed that the higher the torch power is, the lower is the Si content.This means that, for YbDS, the SiO 2 volatilization is more relevant for higher torch powers, and this phenomenon promotes the formation of phases with a lower Si content, like ytterbium monosilicate and ytterbium oxide.On the other hand, for YbMS, the behavior is in countertrend, so the higher the torch power, the higher the Si content.In Figure 7a,c,e, several BSE micrographs show the cross-section of multilayer coatings with YbMS as the top layer, YbDS as the interlayer, and Si as the bond coat.The adhesion between layers is always good, even if some through-the-thickness cracks can be detected in the YbMS layer.BSE micrographs at higher magnification (Figure 7b,d,f) show that some cracks propagate from the YbMS layer into the YbDS layer, splitting into two horizontal cracks.In sample MS-20 (Figure 7d), the cracks propagate partially at the interface, while in the other samples, this phenomenon is not observed, and the crack propagates inside the layers.
The EDS analysis carried out on of the YbMS layers (Figure 8) shows different phases: the elements in the brighter areas are Yb and O (Si is not detected), while the elemental analysis of the darker zones suggests the presence of Yb2SiO5.In the YbMS top layer, a very small amount of Yb2Si2O7 can be detected, and also in this case, the volatilization of SiO2 during the deposition process leads to the formation of a phase with a lower amount of silicon (in this case Yb2O3).
The histogram in Figure 6b shows the correlation of the Si atomic content with the different torch powers.The results show that in this case, an increase in the torch power leads to a higher Si content in the YbMS layer.In Figure 7a,c,e, several BSE micrographs show the cross-section of multilayer coatings with YbMS as the top layer, YbDS as the interlayer, and Si as the bond coat.The adhesion between layers is always good, even if some through-the-thickness cracks can be detected in the YbMS layer.BSE micrographs at higher magnification (Figure 7b,d,f) show that some cracks propagate from the YbMS layer into the YbDS layer, splitting into two horizontal cracks.In sample MS-20 (Figure 7d), the cracks propagate partially at the interface, while in the other samples, this phenomenon is not observed, and the crack propagates inside the layers.In Figure 7a,c,e, several BSE micrographs show the cross-section of multilayer coatings with YbMS as the top layer, YbDS as the interlayer, and Si as the bond coat.The adhesion between layers is always good, even if some through-the-thickness cracks can be detected in the YbMS layer.BSE micrographs at higher magnification (Figure 7b,d,f) show that some cracks propagate from the YbMS layer into the YbDS layer, splitting into two horizontal cracks.In sample MS-20 (Figure 7d), the cracks propagate partially at the interface, while in the other samples, this phenomenon is not observed, and the crack propagates inside the layers.
The EDS analysis carried out on of the YbMS layers (Figure 8) shows different phases: the elements in the brighter areas are Yb and O (Si is not detected), while the elemental analysis of the darker zones suggests the presence of Yb2SiO5.In the YbMS top layer, a very small amount of Yb2Si2O7 can be detected, and also in this case, the volatilization of SiO2 during the deposition process leads to the formation of a phase with a lower amount of silicon (in this case Yb2O3).
The histogram in Figure 6b shows the correlation of the Si atomic content with the different torch powers.The results show that in this case, an increase in the torch power leads to a higher Si content in the YbMS layer.The EDS analysis carried out on of the YbMS layers (Figure 8) shows different phases: the elements in the brighter areas are Yb and O (Si is not detected), while the elemental analysis of the darker zones suggests the presence of Yb 2 SiO 5 .In the YbMS top layer, a very small amount of Yb 2 Si 2 O 7 can be detected, and also in this case, the volatilization of SiO 2 during the deposition process leads to the formation of a phase with a lower amount of silicon (in this case Yb 2 O 3 ).
The histogram in Figure 6b shows the correlation of the Si atomic content with the different torch powers.The results show that in this case, an increase in the torch power leads to a higher Si content in the YbMS layer.

Microstructural Characterization: Porosity and Cracks Analysis
YbDS and YbMS coatings are affected by several defects such as porosities and cracks which need to be analyzed in order to understand the quality of the deposited coatings.All these measurements were taken on the multilayer coatings in which the YbDS is the intermediate layer and the YbMS is the top layer: this strategy is fundamental in considering the interaction between the two layers.
The histogram in Figure 9 shows that for YbDS, the higher the torch power, the lower the porosity of the coating, while for YbMS, the porosity is higher for coatings deposited with medium torch power.However, the difference among the values of porosity for the samples is almost negligible, especially observing the standard deviations.In the micrographs in Figure 4, all the YbDS layers show the presence of cracks along the entire coatings.These cracks have different morphologies and distributions: some cracks propagate from the top to the interface (through-the-thickness cracks), and others are characterized by smaller dimensions and less depth.Coatings manufactured using higher torch powers show more vertical cracks coupled with branched ones (Figure 4b,d), while those manufactured using lower torch powers have cracks with less critical size and morphology but are widespread along the entire coating layer (Figure 4f).BSE micrographs in Figure 7 show the cross-section of the three-layer coatings with some cracks in the YbMS layer developing throughout the entire thickness.Furthermore, some vertical cracks in the YbMS layer are joined to horizontal cracks in YbDS layer (Figure 7b,d,f).As observed by Richards et al. [43], cracks in the YbMS frequently branch with a bifurcation into two cracks when they reach the YbDS layer.This phenomenon was ascribed to thermal residual stress resulting from the mismatch of coefficients of thermal expansion between the YbDS (CTE of 4.7 × 10 −6 °C−1 ) and the YbMS (CTE of 7.5 × 10 −6 °C−1 )

Microstructural Characterization: Porosity and Cracks Analysis
YbDS and YbMS coatings are affected by several defects such as porosities and cracks which need to be analyzed in order to understand the quality of the deposited coatings.All these measurements were taken on the multilayer coatings in which the YbDS is the intermediate layer and the YbMS is the top layer: this strategy is fundamental in considering the interaction between the two layers.
The histogram in Figure 9 shows that for YbDS, the higher the torch power, the lower the porosity of the coating, while for YbMS, the porosity is higher for coatings deposited with medium torch power.However, the difference among the values of porosity for the samples is almost negligible, especially observing the standard deviations.

Microstructural Characterization: Porosity and Cracks Analysis
YbDS and YbMS coatings are affected by several defects such as porosities and cracks which need to be analyzed in order to understand the quality of the deposited coatings.All these measurements were taken on the multilayer coatings in which the YbDS is the intermediate layer and the YbMS is the top layer: this strategy is fundamental in considering the interaction between the two layers.
The histogram in Figure 9 shows that for YbDS, the higher the torch power, the lower the porosity of the coating, while for YbMS, the porosity is higher for coatings deposited with medium torch power.However, the difference among the values of porosity for the samples is almost negligible, especially observing the standard deviations.In the micrographs in Figure 4, all the YbDS layers show the presence of cracks along the entire coatings.These cracks have different morphologies and distributions: some cracks propagate from the top to the interface (through-the-thickness cracks), and others are characterized by smaller dimensions and less depth.Coatings manufactured using higher torch powers show more vertical cracks coupled with branched ones (Figure 4b,d), while those manufactured using lower torch powers have cracks with less critical size and morphology but are widespread along the entire coating layer (Figure 4f).BSE micrographs in Figure 7 show the cross-section of the three-layer coatings with some cracks in the YbMS layer developing throughout the entire thickness.Furthermore, some vertical cracks in the YbMS layer are joined to horizontal cracks in YbDS layer (Figure 7b,d,f).As observed by Richards et al. [43], cracks in the YbMS frequently branch with a bifurcation into two cracks when they reach the YbDS layer.This phenomenon was ascribed to thermal residual stress resulting from the mismatch of coefficients of thermal expansion between the YbDS (CTE of 4.7 × 10 −6 °C−1 ) and the YbMS (CTE of 7.5 × 10 −6 °C−1 ) In the micrographs in Figure 4, all the YbDS layers show the presence of cracks along the entire coatings.These cracks have different morphologies and distributions: some cracks propagate from the top to the interface (through-the-thickness cracks), and others are characterized by smaller dimensions and less depth.Coatings manufactured using higher torch powers show more vertical cracks coupled with branched ones (Figure 4b,d), while those manufactured using lower torch powers have cracks with less critical size and morphology but are widespread along the entire coating layer (Figure 4f).BSE micrographs in Figure 7 show the cross-section of the three-layer coatings with some cracks in the YbMS layer developing throughout the entire thickness.Furthermore, some vertical cracks in the YbMS layer are joined to horizontal cracks in YbDS layer (Figure 7b,d,f).As observed by Richards et al. [43], cracks in the YbMS frequently branch with a bifurcation into two cracks when they reach the YbDS layer.This phenomenon was ascribed to thermal residual stress resulting from the mismatch of coefficients of thermal expansion between the YbDS (CTE of 4.7 × 10 −6 • C −1 ) and the YbMS (CTE of 7.5 × 10 −6 • C −1 ) [15].The results obtained with Matlab code for the evaluation of the crack density are summarized in Figure 9.For the DS layer, the crack density is quite similar for DS-25 and D-20, while a small decrement can be observed for samples deposited at lower power, DS-16.Several studies [57,58] have highlighted that lower plasma powers yield a lower crack density: the colder plasma jet reduces the cooling-down thermal stresses and leads to the lower volatilization of SiO 2 from molten particles and thus leading to a lower crack density.For the MS layer, the results are different, and a higher crack density was measured for layers deposited with lower power.In order to better understand the crack propagation phenomena of these samples, another kind of analysis was also carried out: the density of cracks propagating for at least 70% of the coating thickness (throughthe-thickness crack density or TTT crack density) was evaluated.The TTT crack density parameter was calculated as the number of TTT cracks per millimeter of coating length.In this case, sample MS-16 shows the lower value of TTT crack density (Figure 9); the higher number of micro-cracks probably allows for an efficient stress release with a consequent lower formation of TTT cracks [43].D-20, while a small decrement can be observed for samples deposited at lower power, DS-16.Several studies [57,58] have highlighted that lower plasma powers yield a lower crack density: the colder plasma jet reduces the cooling-down thermal stresses and leads to the lower volatilization of SiO2 from molten particles and thus leading to a lower crack density.For the MS layer, the results are different, and a higher crack density was measured for layers deposited with lower power.In order to better understand the crack propagation phenomena of these samples, another kind of analysis was also carried out: the density of cracks propagating for at least 70% of the coating thickness (through-the-thickness crack density or TTT crack density) was evaluated.The TTT crack density parameter was calculated as the number of TTT cracks per millimeter of coating length.In this case, sample MS-16 shows the lower value of TTT crack density (Figure 9); the higher number of micro-cracks probably allows for an efficient stress release with a consequent lower formation of TTT cracks [43].

X-ray Diffraction Analysis
The diffraction patterns belonging to the as-deposited YbDS samples DS-25, DS-20, and DS-16 are shown in Figure 10.The coatings appear partially amorphous with two broad humps at 25°-38° and 40°-70° and diffraction peaks belonging to monoclinic Yb2Si2O7 (C2/m, 25-1345).Moreover, in DS-20 and DS-16, there are peaks belonging to monoclinic Yb2SiO5 (I2/a, 40-0386).The area under the broad humps is very similar for DS-16 and DS-20, while it is higher for DS-25.This means that a higher torch power causes a higher formation of amorphous phases.Nevertheless, the wider humps corresponding to the amorphous phase in sample DS-25 can hide the effective presence of the Yb2SiO5, which shows its main characteristic peaks at 30.65° and 31.05°.This hypothesis is supported by the BSE SEM micrographs (Figure 4e,f), which show brighter areas corresponding to Yb2SiO5.
For faster quenching, the mobility of the molecules is reduced, and they are not allowed to assume a minimum energy equilibrium position, like a lattice place; therefore, they solidify as an amorphous glass [14,59].This is the case with samples manufactured working with higher torch powers.
In the XRD patterns of YbMS coatings (Figure 10), the two broad humps at 25°-38° and 40°-70° are totally similar to those observed in YbDS, but the area under the humps is higher compared to the YbDS spectra, showing a higher percentage of amorphous phase.The peaks correspond to monoclinic Yb2SiO5 (I2/a, 40-0386), and for MS-16, some peaks related to cubic Yb2O3 (Ia/3, 74-1981) are also detected.For faster quenching, the mobility of the molecules is reduced, and they are not allowed to assume a minimum energy equilibrium position, like a lattice place; therefore, they solidify as an amorphous glass [14,59].This is the case with samples manufactured working with higher torch powers.
In the XRD patterns of YbMS coatings (Figure 10), the two broad humps at 25  -70 • are totally similar to those observed in YbDS, but the area under the humps is higher compared to the YbDS spectra, showing a higher percentage of amorphous phase.The peaks correspond to monoclinic Yb 2 SiO 5 (I2/a, 40-0386), and for MS-16, some peaks related to cubic Yb 2 O 3 (Ia/3, 74-1981) are also detected.

Discussion
The presented research activity aimed to understand the influence of torch power both on the YbDS and the YbMS coatings when deposited as a multi-layer on a Si bond coat and a SiC substrate.The other deposition parameters were kept constant in order to reveal the effect of the torch power on the quality and composition of the coatings.The EDS analyses carried out on YbMS coatings show that the lower the torch power, the lower the Si content.For lower torch powers, the XRD data consistently show the formation of phases poorer in Si content, such as Yb 2 O 3 .
EDS results for YbDS coatings are different: a higher Si content is observed at lower torch power levels; at the same time, the XRD spectra show the presence of ytterbium monosilicate and ytterbium oxide only for low torch power levels (16 kW and 20 kW).This unexpected result can be due to the initial powder's composition (for example, for YbMS, the SiO 2 and Yb 2 O 3 impurities can react during the deposition, and this phenomenon can guarantee a higher content of monosilicate) or to the reliability of the measurements (Si content is very similar for the analysed YbDS coatings).The hypothesis of a further reaction of the SiO 2 and Yb 2 O 3 during the APS process can be supported by the higher content of the powder precursor oxides (SiO 2 and Yb 2 O 3 ) in the YbMS, which were identified by the XRD analysis.Nevertheless, it is expected that the higher torch power causes a higher volatilization of the SiO 2 because of the higher temperature of the torch plume.In any case, it is necessary to consider the velocity variation which can modify the residence time of the particles in the hottest zone of the plasma [60].Furthermore, the SEM micrographs show that all the coatings have few unmolten particles; thus, even for the deposition program with the lower torch power, the combination of temperature and in-flight time allows for the melting of the particles [60,61].Also, the geometry of the porosities can give important information about the thermal history of the thermal-sprayed particles: in this case, there is a prevalence of round porosities, which are due to the gas formed during the melting.By contrast, irregular porosities, due to unmolten particles, are rarely observed [61].
Porosity and crack density for YbDS and YbMS are in countertrend: for YbDS, the higher the torch power, the higher the porosity and crack density, while for YbMS, the higher the torch power, the lower the porosity and crack density.These defects can be due to poor adhesion between successively deposited layers and to the cooling-down residual stresses induced by the difference in YbDS and YbMS CTEs.It is noteworthy that for YbMS, the crack density is higher for a torch power of 16 kW, but for the same condition, the TTT crack density is lower: micro-cracking can relax the internal stress and avoid the formation of TTT cracks.SEM micrographs show that a lower torch power level leads to the formation of narrower and shorter cracks; moreover, in samples deposited at 16 kW, cracks do not propagate at the interface between YbMS and YbDS: a lower torch power seems to guarantee a better adhesion between the interlayer and the top layer.The higher torch power is more efficient only in minimizing the porosity in the YbMS and YbDS, probably because this condition can maximize the amount of molten particles during the in-flight path and their adhesion to the underneath layer.

Conclusions
EBC systems with a Si bond coat, an ytterbium disilicate as the interlayer, and an ytterbium monosilicate as the top coat were deposited onto SiC substrates using APS technologies with three different torch powers.The three-layer system was characterized by performing SEM/EDS analysis, and the defects, like cracks and porosities, were quantified with appropriate software.Several conclusions can be deduced: The phenomenon of the SiO 2 volatilization during the deposition process is not comparable for YbDS and YbMS, so a lower content of Si can be found in the as-sprayed YbDS for a higher torch power (DS-25), while a lower content of Si can be found in the as-sprayed Yb-MS for a lower torch power (MS-16); The crack density is influenced by the torch power, and for YbDS, the lower the torch power, the lower the concentration of defects (DS-16), while for YbMS, a higher torch power

Figure 2 .Figure 2 .
Figure 2. Representation of the main steps involved in the image processing routine for crack analysis: (a) ROI selected on the original SEM micrograph; (b) binarized image where black regions within the coating (in white) represent cracks and porosities; (c) result of the filtering on aspect ratio, which removed porosities; (d) cracks identification (red lines); (e) routine output with the identified and analyzed cracks superimposed on the original SEM micrograph, to demonstrate the good matching.3.Results Coatings 2023, 13, x FOR PEER REVIEW 8 of 16
,f show the difference in the cracks' morphology: cracks are wider in sample Coatings 2023, 13, 1602 8 of 15

Figure 6 .
Figure 6.Variation of Si content with respect to the torch power evaluated using EDS analysis for YbDS and YbMS layers.

Figure 6 .
Figure 6.Variation of Si content with respect to the torch power evaluated using EDS analysis for YbDS and YbMS layers.

Figure 6 .
Figure 6.Variation of Si content with respect to the torch power evaluated using EDS analysis for YbDS and YbMS layers.

Figure 9 .
Figure 9. Porosity, crack density, and through-the-thickness crack density for YbDS layers (on the left) and YbMS layers (on the right).

Figure 9 .
Figure 9. Porosity, crack density, and through-the-thickness crack density for YbDS layers (on the left) and YbMS layers (on the right).

Figure 9 .
Figure 9. Porosity, crack density, and through-the-thickness crack density for YbDS layers (on the left) and YbMS layers (on the right).
The diffraction patterns belonging to the as-deposited YbDS samples DS-25, DS-20, and DS-16 are shown in Figure 10.The coatings appear partially amorphous with two broad humps at 25 • -38 • and 40 • -70 • and diffraction peaks belonging to monoclinic Yb 2 Si 2 O 7 (C2/m, 25-1345).in DS-20 and DS-16, there are peaks belonging to monoclinic Yb 2 SiO 5 (I2/a, 40-0386).The area under the broad humps is very similar for DS-16 and DS-20, while it is higher for DS-25.This means that a higher torch power causes a higher formation of amorphous phases.Nevertheless, the wider humps corresponding to the amorphous phase in sample DS-25 can hide the effective presence of the Yb 2 SiO 5 , which shows its main characteristic peaks at 30.65 • and 31.05• .This hypothesis is supported by the BSE SEM micrographs (Figure 4e,f), which show brighter areas corresponding to Yb 2 SiO 5 .

Figure 10 .
Figure 10.XRD spectra for as-sprayed coatings (YbDS on the left and YbMS on the right).