2.2.3. Microhardness

Vickers microhardness was measured on polished TBC sections by optical microscope Neophot 2 (Zeiss, Oberkochen, Germany) equipped with a Hanemann head and Vickers indenter using 1 N load. The mean value of microhardness was calculated as an average from 20 indentations.

## **3. Results and Discussion**

## *3.1. Microstructure and Porosity*

After 1 h a CMAS residuum clearly covers the coating, Figure 1a. Some fine globular particles, however, developed deeper in the coating. The rod-shaped particles and globular particles have developed after 4 h in the interaction layer (Figure 1b), which looks, however, dense and not porous as in the literature [19]. Since the porosity in the coating was discontinuous and irregular, the infiltration was, expectably, inhomogeneous. After 4 h, Figure 1b, the fine globular particles are developed within the entire thickness of the residual CMAS and are visible also on the surface. After 8 h, Figure 1c, they disappear again. The cracks' character on the surface after 4 h and also 8 h correspond to the glassy character of the residual CMAS. After 10 h, Figure 1d, a significant whisker-like crystallization appears, whereas the matrix between the elongated faceted particles looks again glassy. Moreover, some fine globular particles appear again, and porosity is now opened to the surface (like in as-sprayed SrZrO3).

 **Figure 1.** *Cont*.

**Figure 1.** (**a**) SEM-BSE micrograph of polished cross section sample "1 h"; (**b**) SEM-BSE micrographs of polished sample "4 h": cross section (left) and surface; (**c**) SEM-BSE micrographs of polished sample "8 h": cross section (left) and surface; (**d**) SEM-BSE micrographs of polished– sample "10 h": cross section (left) and surface.

The thickness of the influenced layer increased from 222 μm after 1 h CMAS attack to 333 μm after 8 h. Cai et al. [19] reported only 11 μm after 1 h and 102 μm after 12 h for La2Zr2O7-SrZrO3 composite coating, although the annealing schedule was very similar to our case.

Berker et al. [14] reported 30 μm influenced depth after 3 h at 1300 ◦C for Yb-doped SrZrO3 coating. In these comparisons the initial growth of the influenced layer in our SrZrO3 coatings looks rather large. This was, first of all, because of the high porosity of the as-sprayed coating and due to this fact by easy proliferation of CMAS into the deeper layers of the coating. In contrast, after CMAS attack the coating surface became rather smooth and dense. Sometimes the published experiments were done with a lower concentration of CMAS per unit area of the coating, e.g., 10 mg·cm<sup>−</sup><sup>2</sup> [24]. In such a case, any CMAS residuum is formed on the top of the coating and—just in the beginning—the whole amount of CMAS is consumed for the interaction. Which approach is more realistic, from the application standpoint, is doubtful.

Evolution of porosity parameters, Table 1, indicates a general trend to the coarsening of porosity (pores coagulation) becoming more globular. Porosity area fraction increased and number of voids per mm<sup>2</sup> decreased (i.e., fewer but larger pores), equivalent diameter increased and circularity as well. However, these general trends are affected with certain oscillations—i.e., highest porosity and largest pores after 4 h. Porosity of the SrZrO3 as-sprayed coating was rather high, 23.8%, and after 1 h at 1250 ◦C it shifted upwards, 26.6%. After 4 h dwell time was it even higher, 32.1%. After 8 h dwell time approached again approximately the as-sprayed level, 25.1%. After 10 h was it even higher, 27.7%. Such oscillations could also be viewed as a sacrificial layer development and "ingestion". Repeatedly, a contiguous layer is formed on the surface, sealing the open porosity, but the total porosity is not influenced. The character of CMAS is too aggressive to serve as a simple liquid sintering aid for the base coating. Furthermore, in case of TBC application of SrZrO3, CMAS not only decreases the strain tolerance [2] of the coating but also left the surface opened for the possible next CMAS attack even after 10 h. Observation of microstructures indicates a "transitive" character of the sample after 4 h. Some CMAS remains always on the SrZrO3 coating surface after CMAS attack at 1250 ◦C, resulting in the difficulty to identify all the reaction products (hidden below) by XRD.

Coagulation of pores manifested itself by the increase of *E.D.* (as-sprayed–8.94 μm; 1 h–9.36 μm; 10 h–9.53 μm) and decrease of no. of pores per square millimeter (as-sprayed–5771; 1 h–4341; 10 h–4957). Circularity of pores correspondingly increased, indicating round shape of pores, physically favored after long-term annealing (as-sprayed–0.511; 1 h–0.695; 10 h–0.760), Table 1.


**Table 1.** Image analysis results.

Figure 1 shows that some globular particles disperse in the CMAS melt after 4 h, whereas earlier they were present only in the depth of about 50 μm. Moreover, distinct columnar-like areas with different grey levels developed in the CMAS melt, see in larger details in Figure 2a. Figure 2b shows that, based on the EDX measurement, darker columns are Al-rich, lighter columns are Sr-rich and the fine globular particles as well as rod-shaped fine particles (both white) are Zr-rich. At the beginning (1 h) a Si-rich amorphous superficial layer based on molten CMAS is formed, Figure 1a, with traces of monoclinic ZrO2, crystalline SrSiO3 and two forms of SiO2, i.e. cristobalite and quartz, Table 2. After 4 h, Al-rich (probably spinel MgAl2O4) columns are combined with the earlier developed features, but an amorphous content is still predominant. Interestingly, the Al-rich columns are free of the fine

cracks present in the neighboring Sr-rich columns, Figure 2a. After 8 h the globular Zr-rich particles "sunk" deeper down to the "glassy matrix", Figure 1c. The content of amorphous material decreased as the reaction time increased from 1 to 10 h, see below XRD-based results. The Al-rich and Si-rich columns and globular Zr-rich particles are interpenetrated by residual CMAS melt, Figure 1d. SrZrO3 seems to decompose to SrO and ZrO2. Sr is depleted in the interaction zone, dissolving into the glass while ZrO2 precipitated. After 10 h the surface looks recrystallized, Figure 1d. It contains white globular Zr-rich particles, but the main feature is whisker-like crystals that correspond roughly to SrSiO3 phase, Figure 3. The resting "matrix" is most probably amorphous, since according to EDX it contains besides Si, Sr and Zr also between 1% and 2% of each: Al, Mg, Na and Ca.

**Figure 2.** (**a**) SEM-BSE micrograph, sample 4 h; (**b**) Element map, sample 4 h. Scale bar 30 μm. Area of Figure 2 indicated on Figure 1b by red color.

**Figure 3.** Element maps corresponding to Figure 1d–surface. Scale bar 100 μm.


**Table 2.** XRD phases—semi-quantitative content (%).

\* Producer information. \*\* Most probably Ca, Na, Al or Si -stabilized (on Zr atom position).

## *3.2. Phase Composition*

X-ray diffraction (XRD) results are summarized in Table 2. ZrO2 in both—monoclinic "baddeleyite" P21/c and tetragonal P42/nmc—forms has been found in the irradiated volume. After 1 h, Figure 4, the surface is practically totally amorphous with only few percent of crystalline phases—SrSiO3, SiO2 and ZrO2 monoclinic. This material is CMAS-based glass. Because of its composition, CMAS behaves as a glass, i.e., the heating/cooling sequence in the furnace is although too fast for its crystallization. After 4 h annealing, Figure 5, SrSiO3 (20%), ZrO2 monoclinic (12%) and ZrO2 tetragonal (4%) were detected. ZrO2 monoclinic started to be the main component after 8 h (content 28%), Figure 6. SrSiO3 content dropped (16%) and ZrO2 tetragonal did as well (2%). After 10 h, Figure 7, SrSiO3 was again the most important constituent (42%), whereas ZrO2 monoclinic content dropped (21%) and tetragonal did as well. Always, the rest of the material was amorphous, with the content progressively decreasing with the annealing time (the intensity of peaks increased with time), but being until 8 h higher than 50 percent. XRD results deal only with surface of the samples. Amorphization of the originally crystalline coating is, however, the main aspect of the interaction with CMAS.

**Figure 4.** XRD patterns of the sample "1 h".

**Figure 5.** XRD patterns of the sample "4 h".

**Figure 6.** XRD patterns of the sample "8 h".

**Figure 7.** XRD patterns of the sample "10 h".

Based on reported [25] thermal analysis results, the temperature of the orthorhombic-to-tetragonal phase change has a value of 848 ◦C for SrZrO3. A local minimum of weight gain [25] was recorded by us for the same temperature. Concerning the main phase changes the predominant phenomenon is an inverse run of the reaction described in literature [26] on samples left at 1073 ◦C for one hour.

$$\rm{SrSiO\_3} + \rm{ZrOO\_2} = \rm{SrZrO\_3} + \rm{SiO\_2} \tag{1}$$

In our case the process proceeds as:

$$\text{SrZrO}\_3 + \text{SiO}\_2 = \text{SrSiO}\_3 + \text{ZrO}\_2 \tag{2}$$

ZrO2 tetragonal (a "frozen" high-temperature phase) from the TBC coating transformed to ZrO2 monoclinic under 1170 ◦C due to slow cooling in the furnace.

We sugges<sup>t</sup> that the content (13 percent) of tetragonal ZrO2 from the as-sprayed coating was converted at the presence of CMAS to monoclinic ZrO2, whereas the new tetragonal ZrO2 was transformed from SrZrO3 by its decomposition via chemical reaction (Equation (2)). When SrZrO3 will be decomposed to 50 molar percent of SrO and 50 molar percent of ZrO2, the 50 molar percent of SrO could react with SiO2 completely to SrSiO3 [27].

To identify the reaction products between the CMAS and the SrZrO3 coating, a mixture of the CMAS powder as the solvent and the SrZrO3 feedstock powder as the solute (weight ratio: 1:1) was homogeneously mechanically blended and then heat-treated at 1250 ◦C for 10 h. Besides monoclinic and tetragonal ZrO2, a Na4Ca4(Si6O18) combeite phase appeared. SrZrO3 in the La2Zr2O7-SrZrO3 composite coating [19] reacted much less intensively with CMAS compared to La2Zr2O7. Absence of CMAS-induced through-thickness cracks in our samples could be ascertained mainly to the identical thermal expansion of SrZrO3 [19] and SrSiO3 [28], 10.9 × 10−<sup>6</sup> K−1.

Concerning the "oscillations" in the character of the influenced layer(s) we conclude that the amorphous materials formed preferably on the surface from CMAS crystallizes preferably "from the bottom", i.e., in interaction with the coating. The crystallization opens new pores (via dimensional shrinkage). New proportions of CMAS melt fits into these new pores and sinks deeper into the coating. This repeated process stops after 10 h where finally not enough residual CMAS on the surface exists.

From this standpoint SrZrO3 is not advantageous TBC material since the crystallization front seems to move slowly. Better CMAS blocking layers or sacrificial layers would form when the crystallization is as fast as possible. In this view La2Zr2O7 in the system La2Zr2O7-SrZrO3 [19] served as a "crystallization promoter" and SrZrO3 itself is well CMAS-resistant under ultra-short exposure but started to be CMAS-nonresistant after at least 1 h.

The most stable and most desired phase in the TBC structure among the newly developed phases is tetragonal ZrO2. It stems from SrZrO3 by its decomposition. However, amorphous material is always present in high amounts, and evidently monoclinic baddeleyite crystallizes preferably from the amorphous phase. Disadvantageous is also the fact that admixtures are preferably incorporated in the lattice of tetragonal ZrO2. Among them so called silica-substituted zirconia is rather stable [29]. However, in case of reacting with CMAS, rather Ca, Al or Na atoms would occupy the place of Si, since Si elements form in large quantities the SrSiO3 phase.

## *3.3. Mechanical Properties*

We could summarize microhardness data from Table 3: The SrZrO3 coating was very similar to the as-sprayed samples with values between 5.6 and 6.1 GPa.


**Table 3.** Microhardness.

\* Different character of sample, non-comparable statistics.

E ffect of sintering was negligible in the base material since the annealing temperature was markedly below the sintering temperature of SrZrO3. The influenced superficial layers, visibly changed by interaction with CMAS, were similarly hard but the duration of annealing led to progressively decreased microhardness (GPa): (1 h–6.5; 4 h–5.9; 8 h–5.6; 10 h–5.5). The mixture of powders sintered in a crucible exhibited a very similar value to the coatings: 5.5 GPa. It contained globular (ZrO2-rich) particles and glassy matrix (CMAS-rich) that was in fact a "liquid-phase sintering" agent. In this arrangement, i.e., individual particles of SrZrO3 in contact with the CMAS melt, the interaction process is faster. Among the components the particles were harder, 6.0 GPa than the matrix, 5.0 GPa. SrZrO3 bulk ceramic microhardness was reported to increase after 100 h at 1450 ◦C to 5.5 GPa compared to 4.5 GPa for the as-sintered sample [30].

Combining microhardness with thickness and porosity evaluations we see: after 4 h the influenced layer is thinner and weaker than after 1 h. After 8 h the influenced layer is even less hard but thick and less porous—with the lowest porosity and smallest pore size. The influenced layer is again thinner and maintained its hardness but the coating itself is harder due to pore sealing.

## *3.4. Optical Properties*

The scattering coe fficient of SrZrO3 is reported to be higher than that of YSZ, which is beneficial for the TBC application because most of the incident radiation will be reflected back to the hot gas stream instead of being absorbed by the coating or penetrating the coating leading to the increase of temperature on the base superalloy [31].

In plasma-sprayed TBCs, the scattering is mainly due to the refractive index mismatch between the ceramic coating material and the interior of TBC pores and cracks. This was proved by the decrease of scattering/reflectance (or increase of transmittance) through infiltrating TBC with epoxy or CMAS, whose refractive index is close to the coating material [32]. The decrease of the scattering coe fficient with the increase of the wavelength is due to the decrease of the relative size of the scattering centers (internal pores and cracks) compared with the wavelength [33].

The brown-green color of the CMAS "glaze" is subsequently, with increasing annealing time, turned towards the yellow color typical for SrZrO3 powder and as-sprayed coatings. That is because the residual CMAS is consumed for the reaction, as described above. The yellow reflection as a response on a deep-violet illumination is called luminescence. The sintered mixed powder sample was more luminescent than the coating surfaces, Figure 8. The longer the dwell time, the lower CMAS amount directly at the surface and the highest the luminescence of the "coating itself". The highest luminescence is seen for the sintered mixture of powders. Here, CMAS transformed into a "brown-green glaze", surrounding fine yellow particles of the initial SrZrO3, serving partly as a "liquid-phase sintering" agent, and by this way decreasing the sintering temperature. The appearance, Figure 9, is more or less like a two-component composite with "intact" particles embedded in a fully remelted matrix. In fact, the particles are not intact—Sr from SrZrO3 is transferred to the matrix and the fine white globular particles should be ZrO2 (based on XRD and EDX).

**Figure 8.** From the left: Coatings after CMAS interaction for 1, 4, 8 and 10 h and the cross-sectioned bottom of a crucible with sintered mixture of powders, respectively. Samples illuminated by the 312 nm lamp (i.e., 3.97 eV).

Combeite, a component detected by XRD in the sintered mixed powders and having composition Na6Ca3(Si6O18), was reported to exhibit luminescence centered at about 615 nm [34]. SrZrO3 plasma sprayed coating was luminescent only after annealing in air at 1350 ◦C for 2 h [22]. SrZrO3 is good radiation energy absorber for energies above 3.8 eV [35]. Presence of SrSiO3 (in coatings) and combeite (in mixed powders) shifted the necessary annealing temperature for obtaining luminescent behavior down to 1250 ◦C. SrSiO3 with band gap energy 4.60 eV (higher than SrZrO3 with value this between 3.37 and 4.53 eV [35]) is itself less efficient luminescence material than SrZrO3 but could be formed with a much lower thermal load.

**Figure 9.** SEM-BSE of the sintered mixture of powders, polished cross section.

Similarly to the coatings, Figure 2, the sintered mixture of powders, Figure 9, contained globular Zr-rich particles with sizes under 10 μm. They originated evidently from SrZrO3, since the big "ball" in the left-bottom quadrant of Figure 9 is a particle of the spray feedstock powder (with diameter about 80 μm). The dark grey and pale grey "matrix" is evidently CMAS-based and similarly to coatings could be divided into a crack-free Si-rich darker component and cracked Sr-rich lighter component. Let us see that the Sr-rich component much more frequently surrounds the spray feedstock big "balls" that is coincident with the assumption of the Sr-depletion. Here, contrary to the coatings, Al concentration

in such a matrix was always low, under 3 atomic percent. MgAl2O4 spinel seems to be absent in the sintered mixture of powders.

In general, to tailor the spraying of a TBC to the "optimum porosity" of about 15% (or having it even higher as in the present case) and in the same way make the TBC well resistant to the penetration of molten CMAS, is a problem. A way is to close the superficial porosity as much as possible. Thin barrier layer formed by e.g., laser remelting [36,37] of few superficial micrometers of as-produced TBC would be helpful. However, this additional step makes the production of TBCs even more expensive. Our strontium zirconate, initially even more porous, is on the beginning of interactions with CMAS rather resistant, due to the glassy character of CMAS, but after disappearing the residual CMAS from the surface is the coating surface again porous. However, the dangerous through-thickness cracks were not induced by CMAS within the microstructure of the coating.

Luminescence could be, in principle, used in the diagnostics of the TBC at room temperature. Let us imagine the SrZrO3-covered functional parts of a turbine. The turbine was in service subjected to CMAS attack. After cooling down at the maintenance interruption of the service, change (i.e., the increase) in luminescence would clearly signalize the reaction of the TBC with CMAS. Such a check would be fast and inexpensive. If the turbine was in service, but evidently not subjected to CMAS attack, the luminescence would help at diagnostics of the maximum service temperature—since pure SrZrO3 started to be luminescent after heat exposure at 1350 ◦C [22]. Of course, in case of more complex TBCs, like La2Zr2O7-SrZrO3, these aspects would be less clear, and this is one interesting avenue for future research to explore.
