High-entropy perovskite RETa3O9 ceramics for high-temperature environmental/thermal barrier coatings

Four high-entropy perovskite (HEP) RETa3O9 samples were fabricated via a spark plasma sintering (SPS) method, and the corresponding thermophysical properties and underlying mechanisms were investigated for environmental/thermal barrier coating (E/TBC) applications. The prepared samples maintained low thermal conductivity (1.50 W·m−1·K−1), high hardness (10 GPa), and an appropriate Young’s modulus (180 GPa), while the fracture toughness increased to 2.5 MPa·m1/2. Nanoindentation results showed the HEP ceramics had excellent mechanical properties and good component homogeneity. We analysed the influence of different parameters (the disorder parameters of the electronegativity, ionic radius, and atomic mass, as well as the tolerance factor) of A-site atoms on the thermal conductivity. Enhanced thermal expansion coefficients, combined with a high melting point and extraordinary phase stability, expanded the applications of the HEP RETa3O9. The results of this study had motivated a follow-up study on tantalate high-entropy ceramics with desirable properties.

The structure of HECs is similar to that of highentropy alloys (HEAs), that is, one or more Wyckoff sites in lattices are occupied by identical or near-identical atoms. In HECs, more than four types of atoms occupy one site, and the configurational entropy is higher than 1.5R (where R denotes the gas constant); the configurational entropy (ΔS conf ) is expressed as ΔS conf = R × lnN, where N is the number of elements in one sublattice [5,[25][26][27]. The configurational entropy increases with the number of types of atoms in one or more Wyckoff sites in the lattice. Herein, we regulate the configurational entropy by controlling the types of atoms in the lattice. Theoretically, the highest configurational entropy is obtained for some materials when all the elements in a sublattice have the same atomic fraction [28,29]. To date, HECs have been fabricated with many excellent properties, including enhanced hardness and toughness [19], regulation of thermal transportation properties [20,23], increased corrosion resistance [21], and strengthened hightemperature phase stability [10,17]. HECs of RE 3 TaO 7 [16], RE 2 Zr 2 O 7 [20,23], RE 3 NbO 7 [22], and RE 3 Al 5 O 12 [30] have also been investigated for use as TBCs, whereas HECs of RE 2 Si 2 O 7 [21] and RE 2 SiO 5 [31] have been investigated for use as EBCs. Currently, multifunctional thermal and environmental barrier integration coatings are used to improve the hightemperature performance of components of alloys and ceramic matrix composites. A few oxides meet the requirements of multifunctional E/TBCs. Perovskitetype ceramics typically have high melting points [17], low thermal conductivity [15,17,18], regulatable mechanical properties [17,32,33], etc. [34,35]. Accordingly, multifarious perovskite-type oxides have been fabricated to improve the properties and expand the applications [15,17,18,[33][34][35][36]. In this study, we optimize the properties of HECs of RETa 3 O 9 (RE 0.33 TaO 3 ) perovskites as high-temperature E/TBCs. Differences in ionic radii and atomic weights of constituent elements affect the crystal structures and thermophysical properties of HECs. Therefore, a two-step sintering process is used to prepare four different equiatomic high-entropy perovskite (HEP) tantalates, including (La 1/5 Pr 1/5 Dy 1/5 Ho 1/5 Tm 1/5 )Ta 3 O 9 (5HEP-1), (Gd 1/5 Dy 1/5 Ho 1/5 Er 1/5 Tm 1/5 )Ta 3 O 9 (5HEP-2), (La 1/6 Sm 1/6 Eu 1/6 Dy 1/6 Ho 1/6 Tm 1/6 )Ta 3 O 9 (6HEP-3), and (Eu 1/6 Gd 1/6 Dy 1/6 Ho 1/6 Er 1/6 Tm 1/6 )Ta 3 O 9 (6HEP-4). The ionic radii and atomic weights of La and Pr are considerably different from those of Dy, Ho, and Tm in 5HEP-1, whereas Gd, Dy, Ho, Er, and Tm in 5HEP-2 have similar ionic radii and atomic weights. That is, differences in the ionic radius and atomic weight among the RE elements in 5HEP-1 are more prominent than those in the 5HEP-2 sample. The aforementioned designs are reproduced in the 6HEP-3 and 6HEP-4 samples to investigate the impacts of differences in ionic radii and atomic weights of various RE elements.

1 Material fabrication
HEP RETa 3 O 9 powders were synthesized by a hightemperature sintering process without pressure. We used RE 2 O 3 (RE = La, Sm, Eu, Gd, Dy, Ho, Er, Tm), Pr 6 O 11 , and Ta 2 O 5 with 99.9% purity as raw powders, which had particle sizes less than 10 μm. The weighed powders were mixed by a ball mill (300 r·min -1 , 12 h) using alcohol as the dispersion medium, and then placed in a drier at 90 ℃ for 6 h to evaporate alcohol. The dried powders were maintained at 1600 ℃ for 10 h to obtain high-purity RETa 3 O 9 powders. The prepared powders were passed through a 500-mesh sieve, and approximately 2.5 g of the sieved powders were sintered by spark plasma sintering (SPS). The final SPS parameters were 1600 ℃, 100 MPa, 10-min sintering time, and a heating rate of 50 ℃·min −1 . For comparison, a dense yttria-stabilized zirconia (YSZ) bulk ceramic was synthesized and tested using the same methods.

2 Structure identification
The X-ray diffractometer (MiniFlex 600, Rigaku, Japan) was applied to confirm the crystal structures and lattice constants of the HEP RETa 3 O 9 . The X-ray diffraction (XRD) 2θ angle was ranged between 20° and 60°, and the scanning speed was 5 (°)·min −1 . The exact lattice constants, unit cell volumes, and theoretical densities of each sample were computed from the XRD patterns. The scanning electronic microscope (SEM, JSM-7800F, JEOL, Japan) and the energy-dispersive spectroscope (EDS, 6751A-6UUS-SN, Thermo Scientific, USA)

3 Mechanical property testing
The hardness (H) and fracture toughness (K IC ) were determined by the lengths of indentation diagonal (d) and the indentation crack (c) using a hardness tester (DHV-1000Z-CCD, SZNG, China). Each sample was tested 10 times, the minimum and maximum results were removed, and the remaining results were used to calculate the average values. The fracture toughness increased with the surface fracture energy (Γ ), and was calculated as below [37,38]: where F = 4.9 N, ζ = 0.016, and E denotes the Young's modulus measured using the ultrasonic equipment (UMS-100, TECLAB, France) [39]: where ρ is the sample density, and V l and V t are the longitudinal and transverse acoustic velocities, respectively.
To eliminate the effect of the microstructure on the mechanical properties, the nano-Young's modulus (E N ) and nano-hardness (H N ) were simultaneously measured using the nanoindentation device (iNano instrument, iMicro, USA). During the nanoindentation test, we applied both E and H mapping (three-dimensional (3D) mapping) and advanced dynamic E and H (single-point test) models, where the highest test load was 100 mN. A total of 30 × 30 = 900 points were tested over a 300 µm × 300 µm area at intervals of 10 µm during the 3D mapping measurement. The advanced dynamic E and H tests were used to determine how the hardness and Young's modulus varied with the indentation load and depth, and the results were analysed.

4 Thermal property testing
The thermal expansion coefficients (TECs) were calculated based on the elongation caused by increasing the temperature that was measured by a thermo-mechanical analysis device (TMA 402 F3, NETZSCH, Germany). The thermal diffusivity (λ, 25-900 ℃) was measured using the laser reflection device (LFA 457, NETZSCH, Germany), and the thermal conductivity (k) was calculated from λ, density (ρ), porosity (ϕ), and heat capacity (C p ) as follows [40]: The phonon mean free path (l) was calculated using the relationship between the thermal diffusivity (λ) and the mean acoustic velocity (V m ) [40][41][42]: The mean acoustic velocity (V m ) was calculated from V l and V t [39]: 3 Results and discussion

1 Crystal structures
Figure 1(a) shows that the main XRD peaks of the four HEP RETa 3 O 9 samples coincide with PDF#52-0567 with the P4/mmm space group. A RETaO 4 precipitate (PDF#24-0379) is detected in 5HPE-2, 6HEP-3, and 6HEP-4, whereas the minor peaks at 2θ angles of 25°-30° for 5HEP-1 are associated with a different RETaO 4 phase (PDF#19-1013). The weak peaks of RETaO 4 indicate that the RETaO 4 content is too low to dominate the thermophysical properties of the prepared samples. In Ref. [43], the crystalline temperature of YTaO 4 is reported to be approximately 840 ℃, and this low crystalline temperature may be the cause of the RETaO 4 precipitate in the prepared HEP samples. The lattice of RETa 3 O 9 (RE 1/3 TaO 3 ) with the P4/mmm space group is treated as a defect perovskite (A 1/3 BO 3 -type) structure when a 1/3 vacancy is created in the RE cationic sublattice, as shown in Fig. 1(b). Each Ta is surrounded by six oxygens to form a [TaO 6 ] octahedron, and the octahedra are connected to each other at the corners. These octahedrons are twisted structures, because the Ta-O bond lengths vary. The distortion degree of perovskites depends on the octahedral twist. Normally, the tolerance factor (t) is an indicator of the

Table 1 Design composition, sample name, lattice parameters (a, b, and c), unit cell volume (V), theoretical density (ρ 0 ), and porosity (ϕ) of HEP RETa 3 O 9 ceramics
where r A , r O , and r Ta are the effective ionic radii of RE 3+ , O 2− , and Ta 5+ ions, respectively. Perovskite oxides have a stable lattice when t ranges from 0.8 to 1, and they may crystallize into an orthorhombic phase for t less than 0.9 [29,32]. The tolerance factors of the HEP RETa 3 O 9 are proportional to the A-site ionic sizes, and are 0.871, 0.855, 0.866, and 0.856 for 5HEP-1, 5HEP-2, 6HEP-3, and 6HEP-4, respectively. Among the four prepared samples, 5HEP-2 has the maximum lattice distortion degree based on the minimum tolerance factor. The results presented in Table 1 and Fig. 1(b) indicate that the length of the c-axis is almost twice that of the a-and b-axes, because HEP RETa 3 O 9 is an orthorhombic phase. Usually, the lengths of the a-, b-, and c-axes of ABO 3 -type perovskites are almost equal (corresponding to a cubic phase), and the structures of the HEP RETa 3 O 9 are also called doubleperovskite structures [18]. Herein, the 2 × 2 × 1 unit cell of HEP RETa 3 O 9 is similar to that of cubic ABO 3 perovskites.

2 Microstructures
The four HEP bulk samples are polished using diamond sandpapers, and then held at 1400 ℃ for 3 h to observe the surface grains and elemental distributions by SEM and EDS. In Fig. 2, most grains are smaller than 5 µm, and the RE elements are homogeneously distributed in the prepared samples. The prepared HEP samples have smaller grains than the single-rare-earth RETa 3 O 9 , and grain refinement is achieved in the HEP RETa 3 O 9 samples [15]. The mean grain sizes are 5-7 µm for the single-rare-earth RETa 3 O 9 ceramics and less than 2 µm for the HEP samples. Thus, the corresponding reduction in the grain size is higher than 50%, and we investigate the effect of this reduction on the material properties. Some nanoscale grains are found in 5HEP-2, 6HEP-3, and 6HEP-4, and it is inferred that these grains constitute the RETaO 4 precipitate. We used the backscattered electron (BSE) model to analyse the grains, and no difference was found between nano-and micron-scale grains. The RETaO 4 and RETa 3 O 9 phases are comprised of the same elements, and the grains of these phases have the same atomic contrast. Furthermore, the EDS results show no elemental segregation. The XRD, SEM, and EDS analyses imply that we can neglect the influence of RETaO 4 on the properties of the prepared samples. In Table 1, the porosities of the prepared samples are 1%-3%, which are lower than those of the samples synthesized via a solid-state reaction [15]. Hence, dense and high-purity HEP RETa 3 O 9 ceramics can be fabricated via an SPS process. The grain size, grain boundary density, cracks, and pores affect the properties of structural ceramics, as discussed in the following sections.

3 Mechanical properties
As shown in  Fig. 4(d) and Table 2 indicate that the toughness is proportional to the surface fracture energy. Equation (3) shows that the surface fracture energy is related to the lengths of indentation and cracks, and is an intrinsic feature of these materials. To study the ductile and brittle features of the four samples, we calculate the G/B ratio, where G and B represent the shear and bulk moduli, respectively [39,45]: 2(1 ) where ν is the Poisson's ratio, as measured by ultrasonic equipment. Materials with a G/B ratio less than 0.57 are brittle, and materials with a G/B ratio higher than

modulus and nano-Young's modulus (E and E N ), hardness and nano-hardness (H and H N ), fracture toughness (K IC ), mean acoustic velocity (V m ), surface fracture energy (Γ ), and G/B ratio of HEP RETa 3 O 9 ceramics
Sample  0.57 are ductile [45].  [20], and close to those of the HECs of RE 2 SiO 5 (170 GPa) [31] and ABO 3 (180 GPa) [17]. The relatively low Young's modulus of the HEP samples is attributed to their parent single-rare-earth RETa 3 O 9 ceramics, where a comparatively low modulus promotes thermal stress reduction for high-temperature E/TBC applications [46].

4 Nanoindentation
The nano-Young's modulus (E N = 140-180 GPa) and hardness (H N = 10-11 GPa) are shown in Table 2 and Fig. 5. The slightly higher mean values measured by nanoindentation than those tested by normal methods (Fig. 5(a)) may be due to the indentation-size effect [47][48][49]. Figure 6 shows the 3D mapping of E N and H N , where there are slight variations over the testing area because microstructure defects (pores and cracks) and test error can produce fluctuations in E N and H N . In Fig. 7 where the investigated parameters include the hardness, Young's modulus, indentation depth, and load. The indentation depth is proportional to the load, which has a maximum value of 100 mN in the current study. The H N and E N values for indentation depths less than 100 nm are not considered here because the indenter appearance is not self-similar, and the contact area is indeterminate at small indentation depths (low indentation loads) [47]. H N and E N decrease as the depth and load increase, for test depths beyond 100 nm or loads heavier than 15 mN. During the nanoindentation continuous stiffness test (using the advanced dynamic E and H model), the decrease in the strain gradient results in weakened hardening as the indentation depth and load increase [48][49][50][51][52]. In addition to the strain gradient, other factors affect the hardness and modulus of some materials. The grain size (the grain boundary density), pore, crack, and surface roughness may affect the hardness, toughness, and modulus. Considering the relative density and purity of the HEP RETa 3 O 9 , macro-defects should have little effect on the hardness and modulus, that is, the strain gradient is the predominant factor. The relationships between the Young's modulus and depth/ load are shown in Figs. 5(d) and 5(f), respectively, and their slopes indicate the decreasing rate of the strain gradient in conjunction with increments of load/depth. In addition to the surface micro-defects, the internal micro-defect density affects the modulus and toughness of the prepared ceramics. The results of the advanced dynamic E and H tests show that the internal microdefect density of the HEP RETa 3 O 9 decreases in the order of 5HEP-1 > 5HEP-2 > 6HEP-3 > 6HEP-4, based on the decrease in the modulus with increasing depth. increase of temperature, and then clearly increases at high temperatures (≥ 500 ℃). Figure 8(b) shows that the thermal conductivity of the HEP RETa 3 O 9 exhibits a similar temperature dependence to that of the corresponding single-rare-earth samples, which is caused by a high concentration of A-site cation vacancies and thermal radiation [15,53]. Increases in thermal conductivity caused by thermal radiation have been reported for YSZ [34], La 2 Zr 2 O 7 [54], RETaO 4 [42], RE 2 SiO 5 [31], and ABO 3 perovskite ceramics [17]. Evidently, the prepared samples have lower thermal conductivity than those of SrZrO 3 [55], YSZ [44], and many   [6,10,14,15,17,23,42,44,[53][54][55][56][57][58].

5 Thermal properties
other oxides [57,58]. Note that HEP RETa 3 O 9 and HECs ABO 3 (Zn 0.1 Ca 0.1 Sr 0.4 Ba 0.4 ZrO 3 ) [17] exhibit similar thermal conductivity, which implies that amorphous-like thermal conductivity may be a common feature of HEP oxides. Furthermore, the thermal conductivity of the HEP samples is lower than those of HECs of RE 2 SiO 5 [31] and RE 3 Al 5 O 12 [30], and close to those of HECs of RE 2 Zr 2 O 7 [20]. The lower the thermal conductivity of an HEP sample, the higher the thermal insulation performance is. The temperature dependence of the phonon mean free path (l) shown in Fig. 8(c) is similar to that of the thermal diffusivity, which further suggests the presence of thermal radiation. Figure 8(d) shows that the TECs range from 8.0 × 10 −6 to 10.1 × 10 −6 K −1 (1473 K), which exceeds those of GdTa 3 O 9 (3.6 × 10 −6 K −1 ), SmTa 3 O 9 (5.0 × 10 −6 K −1 ), and DyTa 3 O 9 (7.5 × 10 −6 K −1 ), and lattice relaxation may be the predominant factor in the enhanced thermal expansion of the HEP samples [15]. The enhanced TECs of the HEP RETa 3 O 9 can expand applications of these materials, e.g., as EBCs on the surface of Al 2 O 3 fibre-enhanced Al 2 O 3 ceramic matrix composites (Al 2 O 3f /Al 2 O 3 with TECs of 8.5 × 10 −6 -9.0 × 10 −6 K −1 ) because of matched TECs [10]. In addition, no phase transition is detected during the TECs test up to 1200 ℃, which demonstrates the extraordinarily high-temperature phase stability of the prepared HEP RETa 3 O 9 ceramics. Obviously, the TECs of the RETa 3 O 9 are successfully regulated in the fabricated HECs, which exhibit outstanding phase stability. Figure 8(d) shows that the HEP RETa 3 O 9 oxides have higher TECs compared with those of RE 2 SiO 5 (6.3 × 10 −6 -9.0 × 10 −6 K −1 ) [58] and La 2 Zr 2 O 7 (9.0 × 10 −6 K −1 ) [6] and lower thermal conductivity compared with various E/TBCs [6,10,14,15,17,23,42,44,[53][54][55][56][57][58]. HECs typically have lower thermal conductivity compared with their parent ceramics; however, this trend is not borne out for HEP RETa 3 O 9 . The point defect phonon scattering principle is not applicable for analysing the thermal transport properties of the prepared samples because of (1) the formation of random solid solutions, (2) severe lattice distortion, and (3) the cocktail effect of HECs [59,60]. As the E/k ratio reflects the phonon scattering rate, an increase in E/k causes a decrease in the thermal conductivity. The E/k ratio is 113.8, 92.6, 96.9, and 114.1 GPa·W −1 ·m·K for 5HEP-1 (k: 1.52 W·m −1 ·K −1 ), 5HEP-2 (k: 1.50 W·m −1 ·K −1 ), 6HEP-3 (k: 1.63 W·m −1 ·K −1 ), and 6HEP-4 (k: 1.56 W·m −1 ·K −1 ), respectively. Herein, the prominent differences in the Young's modulus, E/k ratio, and acoustic velocity among the four HEP ceramics do not noticeably affect the thermal conductivity. Grain refinements are found in the prepared samples: Fig. 2 shows that most grains are micron-sized and are therefore considerably larger than the nanoscale phonon mean free path. Herein, the phonon scattering strength from the grain boundary is not taken into account. The increasing degree of lattice distortion may decrease the thermal conductivity of perovskites, where the distortion degree is reflected by the tolerance factor (t). In the HEP RETa 3 where c i denotes the atomic ratio of the i th RE element; A i e denotes the electronegativity of the i th RE element, and A e denotes the mean electronegativity of an A-site atom; A i r denotes the ionic radius of the i th RE element, and A r denotes the mean ionic radius of an A-site atom; A i m denotes the atomic mass of the i th RE element, and A m indicates the mean atomic mass of an A-site atom. Table 3 displays the values of the aforementioned parameters. The variation trends in the thermal conductivity with t, A e  , A r  , and A m  are shown in Fig. 9. No regular variation trends are found  for the thermal conductivity with the aforementioned four parameters; thus, more HEP specimens should be fabricated and tested to identify variation laws. The four investigated samples clearly exhibit low thermal conductivity and high TECs, and they are therefore candidate E/TBCs.

Conclusions
Four novel A-site equiatomic HEP RETa 3 O 9 ceramics are fabricated via a two-step sintering process, and the corresponding thermophysical properties are investigated and analysed as candidate E/TBCs for high-temperature applications. The contributions of this study are presented below: 1) Dense and high-purity HEP RETa 3 O 9 bulk ceramics are successfully fabricated via SPS, and XRD, SEM, EDS, and 3D nanoindentation results demonstrate the excellent compositional homogeneity of all the samples.
2) The tolerance factor of the prepared HEP samples depends on the ionic radius of RE 3+ , where the tolerance factor ranging from 0.8 to 0.9 shows that the samples have stable double-perovskite structures.