Introduction

Thermal barrier coatings (TBCs), along with advanced film cooling technology safeguard hot-section components (such as combustor, blade, vane, and flame tube) to operate in the most demanding environment, where gas temperature is well beyond the withstanding limitation of superalloy1,2,3. As a result, component lifetime can be significantly prolonged. Despite the attractive performance exhibited by TBCs, premature spallation during service is still an overriding concern, as it may result in exposure of bare metal to dangerously high-temperature gases, potentially bringing about a catastrophic accident4,5. Based on the damage mode, the failure of TBCs can be broadly divided into two categories: thermal cycling-relevant failure governed by strain misfits is also referred to ‘intrinsic failure’, whose inducements include sintering6,7, thermal mismatch8,9 and thermally grown oxide10,11, while damage caused by foreign particles (typically fine particulates<75 μm) is the leading cause contributing to ‘extrinsic failure’12,13,14.

Typical TBC system is comprised of a single-crystal substrate, a metallic bondcoat with MCrAlY (M=Ni, Co) and an outer ceramic layer functioning as thermal insulation15,16. Ceramic material is generally chemically inert, however, performance deterioration induced by environmental deposits which are introduced with the inlet air has already found in the past several decades. Among these environmental deposits, sulfate-vanadate mixtures (Na2SO4 and V2O5) are of early interest17,18,19. Since the formation of sulfate-vanadate mixtures is a result of reaction occurring in the combustion of low-quality fuel, the aforementioned degradation occurs exclusively in marine or power-generation engines, whose maximum working temperature is approximately 1000 oC. The salts adhered on coating surface are highly prone to melt during service, and then penetrate into TBCs via defects, eventually leading to irreversible phase transformation of common-used 6 ~ 8 wt.% Y2O3-stabilized zirconia (YSZ)20,21. Alternatively for aircraft engines with somewhat higher operating temperatures, molten silicate deposits otherwise termed ‘CMAS’ (calcium-magnesium-aluminum-silicate) poses a significant threat to the reliability and durability of YSZ coatings22,23,24. More specifically, once those ingested debris melt and infiltrate into coating at elevated temperature (~1250 oC), they would rapidly destabilize YSZ grains by dissolution-reprecipitation reaction, accompanied with phase transformation and coating bending/buckling25,26,27,28; on the other hand, the filling of pores remarkably weakens the strain tolerance of coating. This, in turn, elevates the level of thermal stress during thermal cycling, making coating more susceptible to detach from the metal substrate22,29,30.

Lately, the officers from Office of Naval Research (ONR) and Naval Air Systems Command (NAVAIR) announced the occurrence of a suspected CMAS-related attack in shipboard gas-turbine components after a routine inspection, which brought the issue of molten deposits to the cusp once again31,32. Therewith a series of questions was thrown up: what caused this CMAS-related attack in ship engine, whose operation temperature was well lower than the melting point of typical CMAS as we knew it? did fuel/marine species mixed with dust and debris lessen the triggering temperature of CMAS corrosion, making it more likely to happen? To answer above questions, the scholars have made some efforts in this issue to date. Li et al.33,34 experimentally analyzed the crystallization behavior of CMAS+sea salt mixtures, and found that the addition of sea salt into CMAS would narrow the crystallization temperature. In our previous study35, a noticeable drop in melting point from 1201.2 oC for simulated CMAS to 938.3 oC for CMAS+NaVO3 mixtures (CN) can be seen as the NaVO3 content increased to 28.6 wt.%. Given that degradation occurs only when the deposits become liquid, the temperature triggering coating degradation would be drastically lowered under synergetic effect of CMAS and molten salt. This conclusion well-accounts for the observed phenomenon by ONR and NAVAIR. Our research further pointed out that the introduction of NaVO3 into CMAS can potentially change the viscosity and basicity of the melt. The former has an influence on infiltration behavior, and the latter determines thermos-chemical activity. In addition to these, corrosion behavior of CMAS+molten salt has been also studied by other scholars36,37,38, but the nature of attack mechanism is still much less explored.

In this context, corrosion behavior of CN was experimentally investigated by adopting conventional TBCs/the melt interaction test and self-developed ceramic particles/the melt interaction test (summarized in Table S1), and more emphasis was placed on the thermos-chemical attack. Subsequently the corresponding simulation based on density functional theory (DFT) was carried out to reveal transport behavior during high-temperature reaction. The objective of this research is to comprehensively grasp the corrosion behavior of CN that can gain insights into the mechanism behind thermos-chemical attack, and accordingly provides a theoretical support for developing CN-phobic TBCs.

Results and discussion

Morphology and phase analysis after TBCs/the melt interaction test at 1300 oC

Figure 1 gives the cross-sectional backscattered electron (BSE) images of YSZ coatings subjected to CMAS and CN at 1300 oC for 0.5 h, and corresponding Si mapping. CN-1, CN-2, CN-3 and CN-4 were referred to CMAS+NaVO3 mixtures with different NaVO3 content, which was described in detail in the methodology. The first notable observation was that the upper regions in all tested coatings have almost lost their identity, with initial splats decomposing into globular grains due to the well-known dissolution-reprecipitation reaction25,38. In contrast, these deeper regions remained intact. The XRD pattern depicted in Fig. 2 confirmed that there was no other product reprecipitated apart from m-ZrO2. This was a somewhat surprising result, since one can expect the possible appearance of YVO3 as a result of the reaction between Y2O3 and NaVO339,40.

Fig. 1: Microscopic morphologies of YSZ coatings after attacked by the melts at 1300 oC for 0.5 h, and corresponding Si mapping.
figure 1

a1, b1, c1, d1, e1 The cross section of specimens attacked by CMAS, CN-1, CN-2, CN-3 and CN-4, respectively. a2, b2, c2, d2, e2 Si mapping where blue spots represent Si penetration in the coating.

Full size image
Fig. 2: XRD patterns of YSZ coatings after attacked by the melts at 1300 oC for 0.5 h and 2 h.
figure 2

a 1300 oC for 0.5 h. b 1300 oC for 2 h.

Full size image

Furthermore, EDS analysis was adopted to trace the melts front. Limited sensitivity of EDS mapping analysis was likely to influence the accuracy of the obtained values, but could still provide comparative information with respect to permeability of CMAS and CN. The data from EDS measurement (Fig. 1) demonstrated that the molten CMAS have just penetrated into YSZ specimen to a depth of ~250 µm, while Si element traveled across the whole cross-section of CN-interreacted specimen with the infiltrate depth exceeding ~350 µm under identical conditions. Self-evidently, the molten CN exhibited better permeability.

The appearance of the corroded coatings is provided in the polished cross-section of Fig. 3, corresponding to the specimens exposed to the deposits at 1300 oC for 2 h. The prolonged duration contributed to an extensive interaction, and the regions especially near the pores and micro-cracks those were filled with the melts suffered from severe degradation, leading to noticeable separation of the grains. The increasing intensity of m-ZrO2 characteristic peak as shown in Fig. 2 also suggested an exacerbated degradation, accompanied with the t-ZrO2 transforming into m-ZrO2. When the insight further shifted to the bottom of specimens, we can clearly observe the crystallization of some large rounded grains in CN-attacked specimens (Fig. 3(b–d)). The according formation mechanism will be elaborated below. Table 1 gives the chemical compositions of this rounded grain, together with the ideal composition of as-sprayed YSZ coating identified by EDS analysis. It suggested Y-enriched in the zirconia grains and some Ca incorporated into solid solution. Soon after, the Y- and Ca-rich zirconia was unambiguously determined to be single crystals of cubic ZrO2 by TEM studies in conjunction with the selected area electron diffraction patterns (SAEDPs), as shown in Fig. 4. In addition to these, the spaces between the grains were full of amorphous phase, basically recognized as the glassy deposits (Fig. 4B).

Fig. 3: Microscopic morphologies of YSZ coatings after attacked by the melts at 1300 oC for 2 h.
figure 3

a CMAS. b CN-1. c CN-2. d CN-3. e CN-4.

Full size image
Fig. 4: Bright-field TEM images and indexed SAEDPs of the products formed in coating bottom.
figure 4

c-ZrO2 and amorphous phase observed in coating bottom.

Full size image
Table 1 EDS analysis of the grains in Fig. 3 (mol.%)
Full size table

It is well established that Y-depleted and Y-enriched zirconia both crystallize from the melts, and existing research such as that conducted by Krause41 et al. further proposed that Y content in zirconia grain was closely related to local chemistry. Figure 5 gives the schematic diagram drawing the interaction between molten deposits and YSZ coatings. First considering the formation of Y-poor ZrO2 at the upper region of coatings, and the case was analogous in either CMAS or CN corrosion. The excess deposits accumulating on the coating surface was just like a large pool, and ZrO2 leached into the sink over the time scale of the experiment, as well as Y2O3. Owing to the large amount of Zr4+ available in YSZ and low solubility of Zr4+ 25,41,42, the solubility limit of Zr4+ in the melt was reached in a short order, allowing zirconia grains with Y-depleted to reprecipitate. In this case, phase transformation was inevitable upon cooling, from tetragonal to monoclinic. Deeper within the TBC alternatively, the t-ZrO2 grains were exposed to CMAS/CN “pockets” (small volume of the melt), making it possible to be saturated with Y3+ in the melt. Admittedly, it may take long duration and consume large amounts of t-ZrO2 grains. Critical evidence was the emergence of relatively small grains (marked as P2) in Fig. 3, implicating on-going dissolution into the melt to feed the growth of the neighboring and much larger cubic grains. EDS examination showing a large value of Y:Zr ratio in the melt (P3) was in good accordance with our above analysis. That is to say, the formation of c-ZrO2 grains is supposed to occur in specimen subjected to CMAS, just as it occurred in CN-corroded one. However, this phenomenon was not observed in CMAS corroded-sample as expected.

Fig. 5: Schematic diagram drawing the interaction between molten deposits and YSZ coatings.
figure 5

Figure A shows the formation of m-ZrO2. Figure B indicates the formation of c-ZrO2.

Full size image

Studies over the past two decades have provided important information on the contributing factors affecting crystallization behavior. Several researchers43,44,45 introduced a parameter named ‘Basicity Index’ to be an indicator for CMAS corrosion behavior, and it was defined as the molar ratio of basic and acidic oxides. Considerable evidence indicated that zirconia grains with high stabilizer (Y3+ and/or Ca2+) content preferred to form in more basic CMAS. Na2O, as it happens, is a basic oxide, and the Basicity Index of CMAS can naturally increase due to its addition. In a subsequent paper46, Krause et al. intrinsically attributed different crystallization behavior to the Y3+ solubility, which was highly sensitive to composition changes of the deposits. And it was further found that interreacting with Na-rich CMAS would result in a lower Y-solubility and higher Y-content in zirconia grains in comparison to CMAS with less Na or without Na. It implied that Y-solubility was the decisive factor influencing Y content in zirconia grain in this study.

The reduction in solubility of Y in CN may make it more dangerous and threatening, since the crystallization of Y-rich apatite (Ca4Y6(SiO4)6O), which was commonly considered as an effective product to mitigate CMAS infiltration, would be significantly suppressed due to lack of sufficient Y3+ in the molten CN47,48,49. Consequently, some CMAS-resistant TBC materials (e.g., 2ZrO2·Y2O3) would become ineffective against CN. Considering the difficulty in estimating Y-solubility, the experiment evaluating crystallization capability of apatite in the molten CMAS and CN was carried out via heat-treatment of several mixtures with increasing Y2O3:the deposits ratios at 1300 °C for 12 h. The experiment was not stopped until the characteristic peak of apatite detected by XRD for the first time, and the corresponding results are plotted in Fig. 6. Noted that with NaVO3 content increasing in the deposits, more Y2O3 was required to promote the formation of apatite phase. It meant that the apatite phase crystalized with much difficulty in CN because of NaVO3 addition, which indirectly verified our hypothesis well. That is, the addition of NaVO3 into CMAS would significantly lower Y-solubility in the melts, thereby influencing Y content in zirconia grain in this study. Of course, another effort is urgently needed to estimate whether solubility limit of other rare-earth cations (such as La3+ and Gd3+) also change thanks to the coexistence of CMAS and molten salt.

Fig. 6: Statistical result indicating the formation of apatite in different deposits.
figure 6

The red crosses represent that there is no apatite formed in this mole ratio of Y2O3 and the deposits, and the pink dots represent the formation of apatite. The pink area indicates that apatite phase can be formed in this mole ratio of Y2O3 and the deposits.

Full size image

Microstructural evolution in ceramic particles /the melt interaction test

Owing to lack of well-defined interface between degradation layer and intact layer (as seen in Fig. 3) as well as the constant change of deposits loading on specimen surface induced by infiltration, it was difficult to determine that the addition of NaVO3 had an impact on thermos-chemical aggressiveness of the molten deposits. In this context, another high-temperature test involving ceramic particles and the melt was developed. Figure 7 provides the SEM images illuminating microstructural evolution of YSZ particles exposed to the melts at 1300 oC as functions of exposure duration, and YSZ particles with a diameter of 30 μm were picked out to observe for comparison. It was clear to see that YSZ particles suffered from severe attack driven by dissolution-reprecipitation and intergranular corrosion, and degradation zone comprising of little grains was formed at the periphery of the particles. The occurrence of intergranular corrosion in ceramic particles /the melt interaction test was closely related to the natural defects and high-energy state of grain boundaries, while the interface between splats was preferentially corroded in actual coatings. Data collected from TEM (Fig. 8) confirmed that this kind of product belonged to m-ZrO2. What was noteworthy to highlight was that CN reacted with YSZ materials more vigorously than CMAS, presenting a superior severity of degradation under identical condition. Taking the case for 0.1 h as an example, YSZ particle showed limited interaction albeit impregnated by CMAS, whereas degradation zone in specimens exposed to CN increased in thickness with NaVO3 content, from 1.34 μm for CN-1 to 4.64 μm for CN-3 and 6.75 μm for CN-4. Self-evidently, the area in which phase transformation has occurred also increased accordingly, as indicated in Fig. 8. The different attack behavior led us to draw conclusion without hesitation that, the coupling of NaVO3 and CMAS would strengthen thermos-chemical corrosiveness of the deposits profoundly.

Fig. 7: Microstructural evolution of YSZ samples after exposure to simulated deposits at 1300 oC.
figure 7

a CMAS. b CN-1. c CN-2. d CN-3. e CN-4. a1, b1, c1, d1, e1 0.1 h. a2, b2, c2, d2, e2 0.2 h. a3, b3, c3, d3, e3 0.5 h. a4, b4, c4, d4, e4 1 h. a5, b5, c5, d5, e5 2 h.

Full size image
Fig. 8: Bright-field TEM images and indexed SAEDPs of the YSZ samples after exposure to CMAS and CN-4 at 1300 oC for 0.1 h.
figure 8

a The specimen attacked by CMAS. b The specimen attacked by CN-4.

Full size image

Computational results analysis

Table 2 lists the calculated total energy of CN, CMAS and NaVO3. The formation energy (Ef) of CN can be defined by the following equations50:

$${CMAS}\left({{Ca}}_{9}{{Mg}}_{3}{{Al}}_{4}{{Si}}_{12}{O}_{42}\right)+{NaV}{O}_{3}\to {CN}\left({{Ca}}_{9}{{Mg}}_{3}{{Al}}_{4}{{Si}}_{12}{{{Na}}_{1}{V}_{1}O}_{45}\right)$$
(1)
$${E}_{f}={E}_{{CN}}-{E}_{{CMAS}}-{E}_{{NaV}{O}_{3}}$$
(2)

in which \({{\boldsymbol{E}}}_{{\rm{CN}}}\), \({{\boldsymbol{E}}}_{{\rm{CMAS}}}\) and \({{\boldsymbol{E}}}_{{{\rm{NaVO}}}_{3}}\) are the total energy of CN, CMAS and NaVO3 unit cell, respectively. Accordingly, the defined formation energy for CN can be calculated to be -0.433 eV per formula. Besides, the density of states of CMAS and CN are plotted in Fig. 9. The featured valley commonly referred to as the pesudogap can be clearly found in CN model, suggesting the presence of strong directional covalent bonding in CN51,52. Combined with negative formation energy and the presence of pesudogap, it can be concluded that the formation of CN was thermodynamically feasible if these two environmental deposits meet in marine engines.

Table 2 Calculated total energy E (in eV)
Full size table
Fig. 9: The density of states (DOS).
figure 9

(Above) The DOS of CMAS. (Below) The DOS of CN. The dotted lines represent the Fermi level.

Full size image

As demonstrated above, the addition of NaVO3 into CMAS enhanced thermos-chemical corrosiveness towards YSZ material. To unravel the underlying corrosion mechanism microscopically, we attempted to simulated the process of high-temperature corrosion by establishing corresponding interface model, and subsequently the transport behavior was invested by calculating diffusion coefficient (D), which was derived from following formula:

$$D=\frac{1}{6{N}_{\alpha }}\mathop{\mathrm{lim}}\limits_{t\to \infty }\frac{d}{{dt}}\mathop{\sum }\limits_{i=1}^{{N}_{\alpha }}\left\langle {\left[{r}_{i}\left(t\right)-{r}_{i}\left(0\right)\right]}^{2}\right\rangle$$
(3)

where Nα represents the number of diffusive atoms, and ri (t) is defined as the position of the ith atom at time t. Figure 10 plots the calculated the mean square displacement as functions of simulation time, and the corresponding diffusion coefficient. Noted that except for Mg atom, the diffusion coefficients of all other atoms in CN/YSZ model were much larger than those in CMAS/YSZ model. In general, the essence of thermos-chemical attack is the migration and incorporation of Zr and Y from zirconia matrix into the melt. The pronouncedly promoted transport behavior indicated an active reaction at CN/YSZ interface, enabling the melt more likely to react with YSZ material and thus aggravating the degradation of YSZ coating. Therefore, CN performed more destructively.

Fig. 10: Calculated mean square displacement of several atoms in CMAS/YSZ and CN/YSZ model.
figure 10

a Calculated mean square displacement of Ca. b Calculated mean square displacement of Mg. c Calculated mean square displacement of Al. d Calculated mean square displacement of Si. e Calculated mean square displacement of Zr. f Calculated mean square displacement of Y. The slope of linear fitting indicates the diffusion coefficient.

Full size image

Morphologies analysis after TBCs/the melt interaction test at 1100 oC

Figure 11 shows the cross-sectional morphology of YSZ coatings after attacked by CMAS and CN at 1100 oC for 2 h. Because the annealing temperature did not reach the melting point of CMAS (starting to melt at 1222 oC)35, the specimen exposed to CMAS showed little evidence of attack, and no traces of the melt could be found inside the coatings. Concerning the attack of CN, the significant damage at the upper region of coatings can be noticed; furthermore, examination of the coating microstructure demonstrated that the molten CN readily “wet” YSZ coating. The apparent voids, micro-cracks and splat interfaces filled with the gray glassy melt (denoted by red arrows), indicating that the viscosity of CN at 1100 oC was low enough to travelled across the entire coating. According to our previous investigation35, the viscosity of CN-4 in 1100 oC was estimated to be 5.28 Pa·s, which was comparable to that of CMAS in 1300 oC (4.88 Pa·s). It sounds the alarm for the application of TBCs in marine engine even though used at relatively low operation temperature.

Fig. 11: Microscopic morphologies of YSZ coatings after attacked by the melt at 1100 oC for 2 h.
figure 11

a CMAS. b CN-1. c CN-2. d CN-3. e CN-4. The dotted lines indicate the area of microstructure degraded. Black arrows indicate the presence of melt.

Full size image

Methodology

Conventional TBCs/the melt interaction test

Conventional TBCs/the melt interaction test was conducted at 1300 oC and 1100 oC, respectively. In order to preclude any interference led by severe degradation of substrate and bondcoat at such high temperature, free-standing coating was employed in this work, which can be obtained by preparing a ceramic layer on a carbon steel substrate according to the spraying parameters in Table 3, and then immersing into acid. Figure 12 gives the microscopic images of free-standing YSZ coatings and corresponding binary image. Typical features of air plasma sprayed (APS) TBCs such as unmelted particles, pores and micro-cracks can be clearly visible. This porous characteristic gives coating a desired thermal insulation capacity, but it in turn makes the coating particularly susceptible to molten deposits attack.

Fig. 12: Morphology of free-standing YSZ coatings prepared by APS.
figure 12

Magnification factor of cross-sectional feature area: (a) 250x, (b) 500x. c Binary diagram converted by ImageJ software showing the porosity of the coatings.

Full size image
Table 3 Spraying parameters of YSZ coatings
Full size table

In present study, totally five kinds of synthetic deposits were used, including CMAS and CN with different weight ratio of CMAS and NaVO3 as listed in Table 4. Figure 13 provides a schematic illustrating experimental procedures, which has been described elsewhere in detail53,54. To put it simply, the resulting CMAS/CN frit was mixed with ethanol to form slurry and then applied to the surfaces of free-standing coatings with an area density of 20 mg/cm2. To avoid edge effects, the slurry was spread at the center, leaving a 2 mm salt-free area around the edges.

Fig. 13: Schematic diagram drawing the process of TBCs/the melt interaction test.
figure 13

There are four steps in total, including coating preparation by APS, acid immersion to obtain free-standing coatings, the deposits applied with an area density of 20 mg/cm2 and heat treatment.

Full size image
Table 4 EDS results of simulated deposits prepared in present study (wt.%)
Full size table

Ceramic particles/the melt interaction test

To visually compare thermos-chemical aggressiveness of CMAS and CN towards YSZ material, another experiment was elaborately designed where YSZ ceramic particles were immersed into the melts at 1300 oC. Prior to the test, the commercially available 8YSZ feedstocks those were used to fabricate coatings in Section 4.1 were plasma-densified by utilizing plasma system, and this process is schematically plotted in Fig. 14. It can be seen from inset image that the particle after remelting in the plasma arc had a completely dense microstructure. After thoroughly mixing simulated deposits and resultant particles with weight ratio of 10:1.5, the mixtures were placed into a crucible and then heated at 1300 oC for various duration (0.1, 0.2, 0.5, 1 and 2 h). As illustrated in Fig. 15, the microstructure observation was performed after grinding and polishing cross-sectional specimens according to standard metallographic process.

Fig. 14: Schematic diagram drawing the process of powder densification by plasma system.
figure 14

Green dots represent YSZ powders. The liquid in the beaker is deionized water.

Full size image
Fig. 15: Schematic diagrams illustrating the interaction between ceramic particles and the melt.
figure 15

Pink dots represent YSZ powders, and the brown area represent the melt.

Full size image

Computational method

Figure 16 presents the theoretical models used in this study. The theoretical corrosion model was primarily comprised of molten deposits model with amorphous cell and YSZ surface model. Taking computation and efficiency into account, CMAS can be simplified to Ca9Mg3Al4Si12O42 based on the chemical composition of simulated CMAS given in Table 4. And accordingly, CN model was built with the composition of Ca9Mg3Al4Si12Na1V1O45. Afterwards the construction of amorphous models was achieved by successively relaxing at 1823 K and 1573 K employing an Ab initio molecular dynamics (AIMD) module. The work on building tetragonal zirconia model (t-ZrO2) and the subsequent surface model was informed by the following literature54,55,56. Briefly, a Y4Zr12O30 supercell was first proposed with the oxygen vacancies occupying the next nearest neighbor position, and then cleaved along {1 1 1} Miller index to obtain YSZ surface model as (1 1 1) lattice face has been proved to be predominant and stable in ZrO2 crystal by experimental and theoretical investigation57,58.

Fig. 16: The model built in DFT calculations.
figure 16

a CMAS amorphous model. b CN amorphous model. c CMAS/YSZ interface model. d CN/YSZ interface model.

Full size image

All our simulations demonstrated in this work were carried out using DFT framework as implemented in the DMol3 package, and thereinto generalized gradient approximation as parameterized Perdew-Burke-Ernzerhof functional, in conjunction with double numerical plus polarization basis sets, were defined to perform the geometry optimization and energy calculation. The vacuum space of 25 Å along the non-periodic direction was set to avoid the possible interaction between two neighboring cells. Besides, the energy of 2.0×10−5 Ha, maximum force of 0.004 Ha/Å, maximum displacement of 0.005 Å and self-consistent field tolerance of 1.0×10−5 were chosen as the convergence criterions to achieve good convergence. When it came to MD calculation, the simulation was run in NVT ensemble with Nosé-Hoover thermostat controlling the constant temperature at 1573 K.

Characterization

The characterization of cross-sectional morphologies as well as chemical composition analysis was performed by utilizing energy dispersive spectrometer (EDS)-equipped scanning electron microscope (SEM, FEI Quanta FEG 250, USA). For more detailed examinations, another focused ion beam system (FIB, Thermo Scientific Helios-G4-CX, USA) was employed to prepare samples of interest and then analyzed using a transmission electron microscope (TEM, JEOL F200, Japan) to identify element distribution and phase structure. The specimen after corroded were subjected to X-ray diffraction (XRD, Bruker D8 ADVANCE DAVINCI, Germany) to estimate phase constitution in the 2θ range of 10° to 80° with a scanning rate of 5.8o/min.