8H–10H Stacking Periodicity Control in Twinned Hexagonal

6 days ago - The approach combining simulated annealing and Rietveld refinement with compositional constrain indicates that the 10H Ba8Zr0.4Ti2.6Ta4O2...
0 downloads 6 Views 2MB Size
Download PDF
Article pubs.acs.org/IC

Cite This: Inorg. Chem. XXXX, XXX, XXX−XXX

8H−10H Stacking Periodicity Control in Twinned Hexagonal Perovskite Dielectrics Weiwei Cao,† Xiaoyan Yang,† Fengqi Lu,† Wenfeng Zhu,† Laijun Liu,† Xiaojun Kuang,*,† and Mathieu Allix*,‡,§ †

MOE Key Laboratory of New Processing Technology for Nonferrous Metal and Materials, Guangxi Universities Key Laboratory of Non-ferrous Metal Oxide Electronic Functional Materials and Devices, College of Materials Science and Engineering, Guilin University of Technology, Guilin 541004, PR China ‡ UPR3079 CEMHTI, 1D Avenue de la Recherche Scientifique, Orléans CEDEX 2 45071, France § Université d’Orléans, Faculté des Sciences, Avenue du Parc Floral, Orléans CEDEX 2 45067, France S Supporting Information *

ABSTRACT: Isovalent substitution of Zr4+ for smaller Ti4+ was performed in the 8-layer twinned hexagonal perovskite (referred to as 8H) tantalate Ba8Ti3Ta4O24, which stabilizes a 10-layer twinned hexagonal perovskite (referred to as 10H). The formation of the 10H phase occurs at low substitution concentration (x = 0.1) in Ba8ZrxTi3−xTa4O24 at 1300 °C and reverts back to the 8H phase upon heating at elevated temperatures. Such a 10H-to8H phase transformation is suppressed at higher Zr-substitution contents (x > 0.1). The approach combining simulated annealing and Rietveld refinement with compositional constrain indicates that the 10H Ba8Zr0.4Ti2.6Ta4O24 (x = 0.4) composition adopts a simply P63/mmc disordered structure with Zr cations preferably located in corner-sharing octahedral (CSO) sites compared to face-sharing octahedral (FSO) sites. This 8H-10H phase competition, dependent on the substitution of Zr4+ for Ti4+ and firing temperature, is discussed in terms of the FSO B−B repulsion controlled by the cationic size, as well as the stacking periodicity which affects the thermodynamic stability. Both 8H- and 10H-phase pellets of Ba8ZrxTi3−xTa4O24 exhibit comparable and poorer microwave dielectric properties than the parent 8H Ba8Ti3Ta4O24, which is characterized by cationic disorder and FSO B−B repulsion. The 8H and 10H Ba8ZrxTi3−xTa4O24 ceramics display electrical insulator behavior but with electrically heterogeneous microstructure on the bulk grains. This study demonstrates the opportunity to control the stacking periodicity for the twinned hexagonal perovskites via tuning the B-cationic size and the firing temperature.

■

INTRODUCTION Complex perovskite tantalates Ba3MTa2O9 (M = Zn and Mg) containing 2:1-ordered B-cations are fascinating dielectric materials with applications as telecommunication resonators and filter devices owing to their excellent microwave dielectric performance, i.e., high permittivity (εr ≈ 30), superior quality factor (Qf > 100 000 GHz) and near-zero temperature coefficient of resonant frequency τf.1−3 However, these complex perovskites require prolonged high-temperature firing owing to the slow kinetic of the B-cation ordering that is essential for accessing high quality factor. Such complex and cost-ineffective ceramic processing is a significant hindrance to the large-scale production of complex perovskite tantalate dielectrics.3−8 Recently, B-site deficient hexagonal perovskite oxides, which consist of the same building units of close-packed AO3 layers in perovskite but with mixed cubic (c)-hexagonal (h) close packing, received growing attention for the development of new microwave dielectric ceramics due to their good dielectric performance, i.e., high permittivity and low dielectric loss, accessible under relatively simple process, as well as their diversely structural chemistry.5,9,10 For the B-site deficient hexagonal perovskites AnBn−1O3n, there are two common types © XXXX American Chemical Society

of twinned shifted structures where the stacking of AO3 layers shows an alternation of single and two consecutive hexagonal AO3 layers, forming the stacking sequences of (cn−1h)2 and (cn−2hh)m (m = 1 or 3) and showing periodically polysynthetically twinned and shifted and characters of corner-sharing octahedral (CSO) blocks, respectively.10 Among B-site deficient hexagonal perovskite dielectrics, the 8-layer twinned Ba8MTa6O24 (M = Co, Ni, Zn) compounds are particularly interesting owing to their high Qf values (∼60 000−90 000 GHz), close to those of complex perovskites.11−14 In the 8-layer twinned structure Ba8MTa6O24, the close-packed BaO3 layers are stacked in a (hccc)2 sequence (Figure 1a, referred to as 8H).15 These 8H Ba8MTa6O24 hexagonal perovskites feature partially ordered M cations and vacancies on face-sharing octahedral (FSO) B-sites, which play important role to their high quality factors in microwave frequency.10,16 More recently, the Ba8MnTa6O24 composition was unexpectedly found to form the first 14-layer twinned hexagonal perovskite, i.e., Ba14Mn1.75Ta10.5O42,17 with a (hcccccc)2 stacking Received: February 2, 2018

A

DOI: 10.1021/acs.inorgchem.8b00296 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

ZrO2 (99.99%, Aladdin), and Ta2O5 (99.99%, Aladdin) were weighed according to the correct stoichiometries and mixed thoroughly with ethanol in an agate mortar with a pestle. The mixtures were calcined at 1200 °C for 12 h at heating and cooling rates of 5 °C/min. The calcined powders were then reground with 5% PVA (poly(vinyl alcohol)) solution as an organic binder, then pressed into pellets under a 330 MPa uniaxial pressure with a diameter 10 mm. The pellets with different BZTT compositions were heated at 1300−1400 °C for 30− 35 h at heating and cooling rates of 5 °C/min to ensure reaching the phase equilibration. Dense Ba8Zr0.1Ti2.9Ta4O24 and Ba8Zr0.4Ti2.6Ta4O24 ceramics for microwave dielectric property measurements were prepared via firing 12 mm radius pellets (∼4 g) at 1500 °C for 27 h. The pellet densities were calculated using weight and dimension sizes. Structural and Electrical Property Characterizations. Phase formation was investigated by X-ray powder diffraction (XRPD), which was performed on a Panalytical X’Pert PRO diffractometer with Cu Kα radiation. High-quality XRPD data for Rietveld refinement were collected over the 5−120° 2θ range at room temperature. Variable-temperature XRPD (VT-XRPD) data were collected under vacuum on a D8 Advance Bruker laboratory diffractometer (Cu Kα radiation) using a Vantec-1 linear detector and an Anton Paar oven chamber (HTK1600N model). Diffractograms were acquired from room temperature (RT) up to 1525 °C using isothermal measurements every 25 °C temperature steps from 15 to 70° (2θ) with a 0.0327 step size. Synchrotron X-ray powder diffraction (SPD) data with high intensity and high resolution were collected on the 11BM diffractometer at the Advanced Photon Source (Argonne National Laboratory, USA). The sample powder was loaded in a 0.8 mm diameter kapton capillary, and data were collected at RT over the 0.5− 50° 2θ range using a 0.001° step size and a λ = 0.412642 Å wavelength. Both laboratory XRPD and SPD data were analyzed by the Rietveld method using the Topas academic 5.0 software.20 Selected-area electron diffraction (SAED) experiments were performed on a Philips CM20 transmission electron microscope (TEM) operating at 200 kV and fitted with an Oxford EDS (X-ray energy dispersive spectroscopy) analyzer, on which the elementary analysis was performed. The ceramic morphology sprayed with a thin gold conducting layer was examined using a Hitachi (Tokyo, Japan) S4800 scanning electron microscopy (SEM). Microwave dielectric properties were measured by the HakkiColeman dielectric resonator method21 with the TE011 mode using an Agilent network analyzer N5230A. Temperature coefficient of resonant frequency τf was measured from RT to 85 °C. Ac impedance spectroscopy (IS) measurements were carried out using a Solartron 1260A impedance/gain-phase analyzer within a frequency range from 10−1 to 107 Hz over the RT to 800 °C temperature range. Prior to the measurements, platinum paste was painted on both surfaces of the pellets and fired at 700 °C for 30 min to remove the organic component and to form electrodes.

Figure 1. Schematic plots of the 8H, 10H, and 14H structures with a (cnh)2 stacking sequence of the AO3 layers: (a) 8-layer, n = 3; (b) 14layer, n = 6; (c) 10-layer, n = 4. The blue, light green, and pink spheres denote A, B, and O atoms, respectively.

sequence for the BaO3 layers (Figure 1b, referred to as 14H), in contrast with 8H Ba8MTa6O24. 14H Ba14Mn1.75Ta10.5O42 features high-spin Mn2+ ordering in the central position within the five-consecutive corner-sharing octahedral (CSO) layers. The presence of the large-size Mn2+ in high-spin is key to both the stabilization of this 14H structure and the Mn2+ ordering, disfavoring the 8H structure.17 The stabilization of the 14H structure for the Ba8MnTa6O24 composition implies that the thicker consecutive cubic layers or the larger separation length of hexagonal layers in the twinned structure may be accessible via controlling the B-cationic size. The aim of this work is to achieve similar longer stacking periodicity of the twinned structure on the B-site deficient Ba8B7O24 compositions. Apart from the 14H Ba8MnTa6O24, the twinned structure with a periodicity longer than 8H for the Bsite deficient compositions is the 10-layer twinned hexagonal perovskite (hcccc)2 (Figure 1c, referred to as 10H). So far, 10H phases have been observed only for Ba10M0.25Ta7.9O30 (M = Mg, Co)18 and Ba10TixTa8−0.8xO30 (x = 0.6−1.2)9,19 compositions which contain more B-site vacancies than the 8H Ba8MTa6O24 and Ba8Ti3Ta4O24 materials. Although 10H phase can been observed as a secondary phase during the preparation of Ba8ZnTa6O24, this 10H phase from the BaO− ZnO−Ta2O5 system is hardly accessible, probably owing to the uncontrollable volatilization of zinc oxide during the preparation.11 Here we demonstrate that a limited amount (∼3 mol %) of Zr4+-substitution for smaller Ti4+ in the 8H Ba8Ti3Ta4O24 enables to control the stacking sequence and therefore to reach a new 10H phase. This phase shows an irreversible 10Hto-8H phase transformation for the low-substituted composition Ba8Zr0.1Ti2.9Ta4O24 upon heating at elevated temperatures. The crystallographic data and electrical properties of these new 10H phases are compared with the 8H structures and the new 8H−10H phase competition is discussed based on both the Bcationic size and FSO B-cationic repulsion owing to the Zr4+substitution for the smaller Ti4+ cation as well as the stacking periodicity.

■

■

RESULTS XRPD Data. Figure 2a shows XRPD patterns of BZTT samples synthesized at 1300 °C. At this temperature, the Zrsubstituted compositions form a new single-phase material from x = 0.1. The related XRPD pattern can be indexed with a hexagonal cell (a ∼ 5.782 Å, c = 23.61 Å), indicating the possible formation of a 10H structure.18,19 For substitution contents below x = 0.1, the Zr-substituted BZTT compositions form a mixture of 8H and 10H phases, indicating that the solubility of Zr in the 8H structure is limited. At 1300 °C, the single 10H phase BZTT composition range extends up to x = 0.8. A secondary cubic BaZrO3 phase is observed for the compositions x = 1 and 1.2, and a Ba6TiTa4O18-like 18-layer hexagonal perovskite (referred to 18R) secondary phase also appears for x ≥ 1.4. The fully substituted Ba8Zr3Ta4O24 composition forms a mixture of BaZrO3 and 18R Ba6ZrTa4O18 phases. The rather linear increase of the 10H-phase cell

EXPERIMENTAL SECTION

Synthesis. Polycrystalline Ba8ZrxTi3−xTa4O24 (BZTT, 0 ≤ x ≤ 3) samples were synthesized by high-temperature solid-state reaction in air. Starting materials BaCO3 (99%, Aladdin), TiO2 (99%, Aladdin), B

DOI: 10.1021/acs.inorgchem.8b00296 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 3. (a) Ex situ room-temperature XRPD data of the Ba8Zr0.1Ti2.9Ta4O24 (x = 0.1) samples, fired at 1300 and 1400 °C, and showing the 10H-to-8H phase transformation upon heating. (b) Selected in situ VT-XRPD patterns from 1275 to 1500 °C of the same Ba8Zr0.1Ti2.9Ta4O24 sample. The relative intensities for the reflections of 8H phase collected in situ at high temperature appear different with those at room temperature, which is ascribed to preferred orientation effects.

to-8H phase transformation was investigated further by in situ VT-XRPD. The VT-XRPD data confirms that the 10H-to-8H phase transformation is a gradual procedure that essentially starts at 1350 °C and is complete at 1500 °C during this in situ experiment (Figure 3b). Structure. TEM-EDS analysis of the new 10H BZTT was performed on the x = 0.4 sample to confirm the Ti for Zr substitution. The Ti and Ba elements cannot be distinguished properly by EDS measurement due to their close energy edges (4.469 keV for Ba(L) and 4.509 keV for Ti(K)), with the 0.040 keV energy difference (Δ) being smaller than the usual resolution of EDS of 0.135 keV (Figure S1). However, the Ta(M) edge peaking at 1.709 keV is well separated from the Zr(L) edge peaking at 2.042 keV (Figure S1d, Δ = 0.333 keV compared to the EDS resolution of 0.135 keV). Therefore, as the Zr signal can be clearly distinguished, spectral deconvolution is possible and quantification of the Zr content (compared to Ta) is possible for the x = 0.4 composition. Only the Zr/Ta ratio was analyzed here, from 20 crystallites. The average Zr and Ta atomic percentages correspond to 90(1)% Ta and 10(1)% Zr, which is in very good agreement with the Ba8Zr0.4Ti2.6Ta4O24 (x = 0.4) nominal composition. The small Zr/Ta ratio deviation from one grain to another suggests compositional homogeneousness and confirms that Zr4+

Figure 2. (a) XRPD patterns of BZTT, 0 ≤ x ≤ 1 samples heated at 1300 °C for 25 h. (b) a and c cell parameters of the 10H phase in the BZTT, x = 0.075−1.6 samples heated at 1300 °C. (c) XRPD patterns of BZTT, 0 ≤ x ≤ 0.6 samples heated at 1400 °C for 30 h.

parameters versus the Zr content observed in materials synthesized at 1300 °C (Figure 2b) agrees well with a continuous substitution of Zr4+ for the smaller Ti4+, obeying the Vegard’s law22 within the 0 ≤ x ≤ 0.8 compositional range. Interestingly, when the BZTT samples were fired at 1400 °C, the single 10H solid solution formed in a narrower range, 0.2 ≤ x ≤ 0.4, while a 8H solid solution formed within an extended range, 0 ≤ x ≤ 0.1 (Figure 2c), compared to the samples fired at 1300 °C. This result indicates that increasing the firing temperature is favorable for enhancement of the Zr solubility in the 8H structure but unfavorable for stabilizing the 10H structure. A Zr substitution level higher than x = 0.4 results in the presence of the BaZrO3 secondary phase (Figure 2c) at 1400 °C. The structural change upon heating from 1300 to 1400 °C highlights an irreversible 10H-to-8H phase transformation for the x = 0.1 composition (Figure 3a). Such a 10HC

DOI: 10.1021/acs.inorgchem.8b00296 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

cell. The reflection conditions suggest the P63/mmc, P6̅2c or P63mc possible space groups. Extra weak diffuse streaks (marked by red arrows in Figure 4a) were observed along c*, indicating the occurrence of short-range ordering along the abplane, whereas ordering would not be clearly established along the c axis. The long-range ordering along the ab-plane may lead to a triple ordered structure, which has been frequently observed in the B-site deficient twinned hexagonal perovskites, e.g., 8H Ba8ZnTa6O2411 and 14H Ba14Mn1.75Ta10.5O42.17 TEM examination revealed that in the 10H Ba8Zr0.4Ti2.6Ta4O24 sample some crystallites show such short-range ordering at nanometer scale, but some others do not, therefore implying the existence of localized nanometer-scale ordering. Rietveld refinement of the Ba8Zr0.4Ti2.6Ta4O24 10H structure was performed based on SPD data using a simple 10H structural model with the highest symmetric space group P63/ mmc without considering the short-range ordering revealed by the electron diffraction in some areas. The structural model contains three crystallographically distinct B-sites: two CSO sites (B1 and B2) and one FSO site (B3) in Table 1, among which B1 sites are the central CSO sites in the CSO block and are sandwiched by two B2 sites neighboring the FSO B3−B3 dimers. Given the complex B-site occupancy associated with three cations and vacancies, the occupancies of Ta, Ti, and Zr of these sites were refined by a simulated annealing procedure and subject to a compositional constraint according to their nominal compositions and B-site vacancy confinement on the FSO site only. This approach combining simulated annealing and Rietveld refinement with compositional constraint is effective to reduce the probability of false minima in the refinement by randomly setting the initial occupancies parameters in each refinement cycle and to search for the best solution with the minimum R factors from thousands of refinement cycles.20 Such an approach has proved reliable and successful on determining complex structures from powder diffraction data.16,23 The refinement converged to Rwp ∼ 7.75% and RB ∼ 1.74%. Figure 5 shows the Rietveld fit of SPD data, showing excellent fit to the overall reflections. Tables 1 and 2 list the final refined structrual paramteters and selected bond lengths for the 10H Ba8Zr0.4Ti2.6Ta4O24 in P63/mmc. The refined structure shows a larger thickness of ∼2.47 Å for the FSO B3O6 layer than those (∼2.29 Å) for the CSO B1O6 and B2O6 layers, consistent with the presumption of B-site vacancy on the FSO site only. In order to examine the possible shortrange B-cationic ordering in the 10H structure, the lower symmetry model in acentric P63mc (Table S1) was also

contributes to the formation of the Ba8Zr0.4Ti2.6Ta4O24 10-layer phase. Figure 4 shows representative SAED patterns of the 10H Ba8Zr0.4Ti2.6Ta4O24 sample, which confirm the 10H hexagonal

Figure 4. SAED patterns of 10H Ba8Zr0.4Ti2.6Ta4O24. Diffuse streaks observed along c* in (a) and marked with red arrows are assigned to short-range ordering of cations and vancancies while spots marked with a white arrows in (b) arise from double diffraction.

Table 1. Final Refined Structural Parameters for 10H Ba8Zr0.4Ti2.6Ta4O24 (Ba10Zr0.5Ti3.25Ta5O30) in P63/mmc from Rietveld Refinement of SPD Dataa atom

site

x

y

z

occupancyb

Biso (Å2)

Ba1 Ba2 Ba3 B1 B2 B3 O1 O2 O3

2d 4e 4f 2a 4f 4f 6h 12k 12k

2/3 0 2/3 0 2/3 2/3 0.8217(4) 0.3332(5) 0.4973(3)

1/3 0 1/3 0 1/3 1/3 0.1783(4) 0.1665(3) 0.9946(5)

0.25 0.34023(2) 0.55218(3) 0 0.39955(2) 0.79915(3) 3/4 0.4511(1) 0.6460(1)

1 1 1 0.672(2) Ta, 0.24(4) Ti, 0.086(6) Zr 0.575(1) Ta, 0.342(4) Ti, 0.083(5) Zr 0.3389(7) Ta, 0.351(2) Ti, 0.002(1) Zr 1 1 1

0.75(1) 1.80(1) 0.87(1) 0.15(1) 0.49(1) 0.36(1) 1.49(9) 0.56(6) 0.79(6)

a a = 5.792563(5) Å, c = 23.65078(3) Å, and V = 687.253(1) Å3; Z = 1 for the Ba10Zr0.5Ti3.25Ta5O30 formula. bThe B-site occupancies were refined by using the approach combining simulated annealing and Rietveld refinement with compositional constraint.

D

DOI: 10.1021/acs.inorgchem.8b00296 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 5. Rietveld refinement plot of SPD data for 10H Ba8Zr0.4Ti2.6Ta4O24 in P63/mmc. The reliability factors are Rwp ∼ 7.75%, Rp ∼ 5.35% and RB ∼ 1.74%. The Bragg diffraction positions are shown by vertical marks. The insets enalrge the fitting to the weak reflections in the low 2θ range 1.8−7.2° (top) and high 2θ range 25− 30° (bottom).

Figure 6. SEM observations of the surface morphologies of 8H (x = 0.1) and 10H (x = 0.4) BZTT pellets.

corresponding parent 8H Ba 8 Ti 3 Ta 4 O 24 24 and 10H Ba10TixTa8−0.8xO30 (x = 0.6−1.2) analogue compounds,9 indicating that the incorporation of Zr in the 8H and 10H structure has little influence on their εr and τf values. The ac impedance spectroscopy measurements were performed on the BZTT ceramics in order to determine their conductivity and electrical microstructure. The impedance measurements showed that both 8H (x = 0.1) and 10H (x = 0.4) BZTT pellets are highly resistive below 400 °C, and their bulk conductivity varies within 10−9−10−5 S/cm in the temperature region 500−800 °C, showing activation energies ∼1.3−1.5 eV (Figure S3), similar to the parent Ba8Ti3Ta4O24. Figure 7a shows complex impedance plot of a 10H Ba8Zr0.4Ti2.6Ta4O24 pellet measured at 700 °C, which exhibits one large and one small semicircular arcs. The large arc displays a capacitance of ∼4−6 pF/cm, signing for the bulk response.25 The small arc shows a capacitance of ∼10−11−10−8 F/cm, corresponding to the response from the thinner grain boundary regions.25 Careful examination of the bulk response arc of the 10H Ba8Zr0.4Ti2.6Ta4O24 pellet reveals that the semicircular arc is asymmetric and shows two capacitance plateaus (inset in Figure 7a) of ∼6 pF/cm (in 103−104 Hz) and ∼4 pF/cm (in 104−106 Hz), indicating two components for the bulk response arc. This is further confirmed by the existence of two imaginary electrical modulus M″ peaks26 (Figure 7b). Thus, the impedance data reveals that the Zr-substituted 10H pellet is an excellent insulator but possesses electrically heterogeneous microstructures of the bulk grains. The imaginary electrical modulus as a function of frequency for the 8H x = 0 and 0.1 samples at 650 °C (Figure S4) also shows similar behavior of electrically heterogeneous microstructure of the bulk grains.

Table 2. Bond Lengths for the 10H Ba8Zr0.4Ti2.6Ta4O24 in P63/mmc bond

length (Å)

bond

length (Å)

Ba1−O1(×6) Ba1−O3(×6) Ba2−O1(×3) Ba2−O3(×6) Ba2−O2(×3) Ba3−O2(×6) Ba3−O2(×3)

2.899(3) 2.959(2) 2.785(1) 2.915(2) 3.108(2) 2.897(3) 2.920(2)

Ba3−O3(×3) B1−O2(×6) B2−O2(×3) B2−O3(×3) B3−O1(×3) B3−O3(×3)

2.795(2) 2.034(2) 2.069(2) 1.966(2) 1.942(2) 2.138(2)

considered by Rietveld refinement as each 4f site in the P63/ mmc model derives into two different 2b sites in the P63mc model, which allows refining the cationic ordering inside the FSO dimers and the CSO sites. The remaining possible space group P6̅2c was not considered for such short-range ordering refinement as the crystallographically distinct atomic positions in P6̅2c model have exactly the same multiplicities as those in the P63/mmc model. The P63mc refinement gave essentially the same R factors (e.g., Rwp ∼ 7.67%, RB ∼ 1.31%) and did not show apparent evidence of B-cationic ordering (Table S1). Therefore, the centric P63/mmc model is used to describe the 10H structure of x = 0.4 sample. Electrical Properties. Dense pellets of 8H Ba8Zr0.1Ti2.9Ta4O24 and 10H Ba8Zr0.4Ti2.6Ta4O24 were fired at 1500 °C, reaching ∼92−95% of theoretical X-ray densities. XRPD data confirms that Ba 8 Zr 0 . 1 Ti 2 . 9 Ta 4 O 2 4 and Ba8Zr0.4Ti2.6Ta4O24 pellets formed 8H and 10H structures, respectively, consistent with the phase formation of the samples fired at 1400 °C. The surface morphologies of 8H Ba8Zr0.1Ti2.9Ta4O24 and 10H Ba8Zr0.4Ti2.6Ta4O24 pellets were examined by SEM (Figure 6), showing similar compact stacking of flat grains with side surfaces exposed, leading to column-like shape along the pellet surfaces. The elemental mapping analyses (Figure S2) confirm the compositional homogeneousness in these Zr-substituted samples. The microwave dielectric properties for these BZTT pellets are εr ∼ 39, Qf ∼ 13223 GHz and τf ∼ 77 ppm/°C for 8H Ba8Zr0.1Ti2.9Ta4O24 and εr ∼ 35, Qf ∼ 11052 GHz, τf ∼ 68 ppm/°C for 10H Ba8Zr0.4Ti2.6Ta4O24. The εr and τf values are close to those reported for the

■

DISCUSSION Stabilization of the hexagonal perovskite structures is predominated by the electrostatic repulsion between the FSO B-cations, owing to the short B−B distances.10,16,27 Introducing vacancies over the FSO sites is favorable for reducing the FSO B−B cationic repulsion, which frequently stabilizes the hexagonal perovskite structures of the B-cationic deficient compositions.27 In contrast, formation of hexagonal perovskites became more and more difficult with the increase of the stacking periodicity of h- and c- AO3 layers.17 It is well-known E

DOI: 10.1021/acs.inorgchem.8b00296 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

repulsion as it lowers the probability of FSO Ta5+−Ta5+ dimers with strong repulsion. The substitution of Zr4+ for the FSO Ti4+ in the 8H Ba8Ti3Ta4O24 material is thus expected to form FSO Zr4+−Ta5+ dimers and therefore to deteriorate the FSO B−B cationic repulsion owing to the much larger size of Zr4+ (0.72 Å)28 compared to Ti4+ and Ta5+, which could destabilize the 8H structure. Therefore, the preference of Zr4+ substitution for the CSO Ti4+ sites over the FSO Ti4+ sites may be expected in 8H Ba8Ti3Ta4O24 although the refinement on the lightly substituted composition 8H Ba8Zr0.1Ti2.9Ta4O24 cannot firmly confirm this preference owing to the low Zr substitution level. In the 8H structure, there are only two consecutive CSO layers thus each CSO site is close to the FSO site (Figure 1a). In the case of Zr substitution on the CSO site in 8H structure, the large corner-sharing ZrO6 could induce local stress to the facesharing octahedra and therefore could also indirectly deteriorate the FSO B−B repulsion. The Zr distribution in the extended CSO block (three consecutive CSO layers) of the 10H structure (Figure 1c) has more site choice over both B1 (next to the FSO B3 site) and B2 sites (in the central position of CSO block and has no direct interaction with the FSO B3 site), as revealed by the structure refinement of 10H Ba8Zr0.4Ti2.6Ta4O24 (Table 1). Therefore, the Zr presence in the CSO block tends to have lower deterioration effect to the FSO B−B repulsion, compared with its substitution effect in the 8H structure. In contrast, as the B-site vacancy concentration in the FSO sites of the 10H structure (31.25%) is higher than the one observed in the 8H structure with the same Ba8B7O24-type composition (25%), the reduction of the FSO B−B cationic repulsion is therefore favored in the 10H structure. Thus, the structural advantage of the extended CSO block combined with the increase of FSO B-site vacancies could make the 10H structure more favorable for the Zrsubstituted BZTT composition over the 8H structure. The expectation that the FSO sites do not favor large cations such as Zr4+ is consistent with an apparent preference of large Zr cations for CSO sites over FSO sites, as revealed by the Rietveld refinement of the 10H Ba8Zr0.4Ti2.6Ta4O24 structure (Table 1). Compared to other known 10H phases18,19 and the 14H Ba14Mn1.75Ta10.5O42 (Ba8MnTa6O24)17 phase which contains essentially 50% vacant FSO sites, thus avoiding the FSO B−B electrostatic repulsion and which belong to the AnBn−2O3n-type structure,10,17 the 10H BZTT materials contain lower amounts of vacant FSO sites (31.25%) and thus more FSO B−B dimers in short contact. Therefore, this 10H Ba8ZrxTi3−xTa4O24 is expected to be less stable than the AnBn−2O3n-type 10H structures. The irreversible 10H-to-8H phase transformation observed for the Ba8Zr0.1Ti2.9Ta4O24 composition upon heating (Figure 3) indicates that the Zr-substituted 10H structure is indeed metastable and that the 8H structure is more thermodynamically stable than the 10H structure. This result emphasizes again that the thermal stability of hexagonal perovskites decreases with the increase of the stacking periodicity. Therefore, the FSO B−B electrostatic repulsion and stacking periodicity are competing factors on controlling the 8H−10H phase option in BZTT. The stacking periodicity plays a dominant role on the thermodynamic stability over the electrostatic repulsion at elevated temperatures for the low Zrsubstitution, while the suppression of the 10H-to-8H phase transformation for the highly Zr-substituted composition Ba8Zr0.4Ti2.7Ta4O24 at 1400 °C indicates that the enhanced electrostatic repulsion owing to the presence of large Zr4+ takes

Figure 7. (a) Complex impedance plot and (b) Z″ and M″ spectroscopic plots of 10H Ba8Zr0.4Ti2.6Ta4O24 pellet at 700 °C. Rb and Rt denote bulk and total resistivity, respectively. The numbers denote the logarithms of the selected frequencies marked with filled symbols. The inset in (a) shows the frequency dependency of capacitance, highlighting two capacitance plateaus in the high frequency region.

that hexagonal perovskites usually accommodate compositions with a tolerance factor greater than 1. This means that the Bcations are relatively smaller than the octahedral interstices between AO3 layers in the cubic perovskite block, i.e., the CSO block is in tension, which may be balanced by the stress of the FSO blocks. Expansion of consecutive c layers between h layers for the longer periodicity of stacking sequence (hcn)2 in the twinned structure will separate further or dilute the h-AO3 layers, which could affect the balance between the CSO and FSO blocks, and therefore tend to reduce the thermodynamic stability for longer stacking periodicity in the twinned structure. Here, the formation of the 10H structure for limited substitution levels of larger Zr4+ for Ti4+ and the 10H-to-8H structural transformation upon heating in the Ba8Ti3Ta4O24based compositions may be interpreted based on the FSO B−B cationic repulsion and stacking periodicity. Among the known 8H tantalates, Ba8Ti3Ta4O24 is a typical example with small Bcationic size and charge difference between Ta5+ and Ti4+, thus resulting in more disordered B-cationic than that in Ba8MTa6O24, although Ba8Ti3Ta4O24 remains a vacancypartially ordered triple 8H structure.19 The smaller size (0.605 Å) and lower charge of Ti4+ compared to Ta5+ (0.64 Å)28 make Ti4+ cations slightly preferable for FSO-site over CSO site, which is favorable on reducing the FSO B−B cationic F

DOI: 10.1021/acs.inorgchem.8b00296 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

opportunity for accessing new twinned hexagonal perovskites with longer periodicity.

charge of the phase competition/formation, favoring the 10H structure with less FSO B−B repulsion when the Zr4+ substitution is increased. Compared with the parent 8H Ba8Ti3Ta4O24,24 both the 8H and 10H Zr-substituted Ba8Ti3Ta4O24 pellets possess lower Qf values. Site/charge disorder and FSO B−B repulsion have been proposed as two major factors deteriorating the dielectric loss for the B-site deficient hexagonal perovskites and their contribution to the dielectric loss depends on the composition.16,18 The enhanced FSO B−B repulsion related to the large Zr incorporation may contribute to the lower Qf values of the 8H Ba8Zr0.1Ti2.9Ta4O24 pellet compared with the undoped Ba8Ti3Ta4O24.24 Although the FSO B−B repulsion was reduced in the 10H structure, the Qf value of the 10H Ba 8 Zr 0 . 4 Ti 2 . 6 Ta 4 O 2 4 pellet is lower than the 8H Ba8Zr0.1Ti2.9Ta4O24 pellet, which could be ascribed to more disordered B-cations in the 10H structure compared with the 8H structure: The SAED patterns show only short-range ordering in the 10H phase (Figure 4), in contrast with the ordered triple 8H superstructure. The cationic disorder could have a predominant role to the dielectric loss over the FSO B− B repulsion in the 10H Ba8Zr0.4Ti2.6Ta4O24. Last, it is interesting to note that to the best of our knowledge the electrically heterogeneous microstructure of the bulk grains in the Ba8Ti3Ta4O24-based ceramics revealed by impedance spectroscopy measurements has not been previously observed on the 8-layer hexagonal perovskites. As both the parent and Zr-substituted compositions displayed two components for the bulk response arcs in the impedance data and as the Zr distribution is homogeneous in the samples, there has to be other factors responsible for such electrically heterogeneous microstructure rather than the Zr-substitution. A possible mechanism could be associated with the cationic ordering. As revealed by the TEM characterization, given that some crystals show local ordering whereas some others do not, the presence of both partially ordered and nonordered crystallites may lead to the heterogeneity observed by electrical measurements. These results are foreseen to stimulate further investigation.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.8b00296. TEM-EDS spectrum and SEM-EDS elemental mapping images of Ba8Zr0.4Ti2.6Ta4O24; bulk conductivities of BZTT (x = 0, 0.1, 0.4) pellets; Z″ and M″ spectroscopic plots for BZTT (x = 0, 0.1) pellets; crystallographic information in CIF file for Ba8Zr0.4Ti2.6Ta4O24 (PDF) Accession Codes

CCDC 1582871 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif, or by emailing data_ [email protected], or by contacting The Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: +44 1223 336033.

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Xiaojun Kuang: 0000-0003-2975-9355 Mathieu Allix: 0000-0001-9317-1316 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The National Natural Science Foundation of China (Nos. 21361008 and 21511130134), Guangxi Program for Hundred Talents for Returned Scholars and Guangxi Key Laboratory for Advanced Materials and New Preparation Technology (No. 12AA-11) are acknowledged for the financial support, as well as CNRS for financial support on the bilateral TransLight project (PICS07091). Use of the Advanced Photon Source at Argonne National Laboratory was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357. The authors thank the CRMD laboratory for TEM access. Prof. Liang Fang is acknowledged for help on the microwave dielectric measurements.

■

CONCLUSIONS Limited isovalent substitution (∼3 mol %) of Zr4+ for smaller Ti4+ in the 8H Ba8Ti3Ta4O24 enables stabilization of a 10H hexagonal perovskite, which transforms back to 8H phase at high temperature. The stabilization of the 10H phase by Zr substitution, and the 10H-to-8H phase transformation at low Zr-substitution (x = 0.1), as well as its absence for higher Zrsubstitution levels upon heating, reveal two competing factors: FSO cationic repulsion and stacking periodicity, which control the 8H-10H phase stability. In contrast with the ordered triple 8H Ba8Ti3Ta4O24, the 10H Ba8Zr0.4Ti2.6Ta4O24 composition adopts a simply disordered structure described in the P63/mmc space group. The approach combining simulated annealing and Rietveld refinement with compositional constrain indicates that the Zr cations in this 10H structure prefer to occupy CSO sites rather than FSO sites. The 8H and 10H Ba8ZrxTi3−xTa4O24 pellets demonstrate electrical insulator character with electrically heterogeneous microstructure on the bulk grains and possess comparable, although poorer microwave dielectric properties, than the 8H Ba8Ti3Ta4O24 parent material. These results demonstrate the possibility to control the stacking sequence for the twinned hexagonal perovskites via tuning the B-cationic size and the firing temperature, offering more

■

REFERENCES

(1) Nomura, S.; Toyama, K.; Kaneta, K. Ba(Mg1/3Ta2/3)O3 Ceramics with Temperature-Stable High Dielectric Constant and Low Microwave Loss. Jpn. J. Appl. Phys. 1982, 21, L624−L626. (2) Lufaso, M. W. Crystal Structures, Modeling, and Dielectric Property Relationships of 2:1 Ordered Ba3MM′2O9 (M= Mg, Ni, Zn; M′=Nb, Ta) Perovskites. Chem. Mater. 2004, 16, 2148−2156. (3) Reaney, I. M.; Iddles, D. Microwave Dielectric Ceramics for Resonators and Filters in Mobile Phone Networks. J. Am. Ceram. Soc. 2006, 89, 2063−2072. (4) Zeng, S.; Ni, G.; Kuang, X. Tuning the temperature coefficient of resonant frequency for 8-layer twinned hexagonal perovskite Ba8ZnTa6O24 ceramics. J. Mater. Sci.: Mater. Electron. 2016, 27, 10078−10081. (5) Thirumal, M.; Davies, P. K. Tunable high Q perovskite dielectrics in the BaO−NiO−Ta2O5 system. J. Mater. Sci. 2011, 46, 4715−4718. G

DOI: 10.1021/acs.inorgchem.8b00296 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry (6) Cava, R. J. Dielectric materials for applications in microwavecommunications. J. Mater. Chem. 2001, 11, 54−62. (7) Hughes, H.; Iddles, D. M.; Reaney, I. M. Niobate-based microwave dielectrics suitable for third generation mobile phone base stations. Appl. Phys. Lett. 2001, 79, 2952−2954. (8) Wersing, W. Microwave ceramics for resonators and filters. Curr. Opin. Solid State Mater. Sci. 1996, 1, 715−731. (9) Baranov, A. N.; Oh, Y. J. Microwave frequency dielectric properties of hexagonal perovskites in the Ba5Ta4O15 − BaTiO3 system. J. Eur. Ceram. Soc. 2005, 25, 3451−3457. (10) Trolliard, G.; Ténèze, N.; Boullay, P.; Mercurio, D. TEM study of cation-deficient-perovskite related AnBn−1O3n compounds: the twinshift option. J. Solid State Chem. 2004, 177, 1188−1196. (11) Moussa, S. M.; Claridge, J. B.; Rosseinsky, M. J.; Clarke, S.; Ibberson, R. M.; Price, T.; Iddles, D. M.; Sinclair, D. C. Ba8ZnTa6O24: a high-Q microwave dielectric from a potentially diverse homologous series. Appl. Phys. Lett. 2003, 82, 4537−4539. (12) Thirumal, M.; Davies, P. K. Ba8ZnTa6O24: A New High Q Dielectric Perovskite. J. Am. Ceram. Soc. 2005, 88, 2126−2128. (13) Kan, A.; Ogawa, H.; Yokoi, A.; Ohsato, H. Microwave Dielectric Properties of Perovskite-Like Structured Ba8Ta6(Ni1−xMx)O24 (M Co, Cu, and Zn) Solid Solutions. Jpn. J. Appl. Phys. 2006, 45, 7494− 7498. (14) Kawaguchi, S.; Ogawa, H.; Kan, A.; Ishihara, S. Microwave dielectric properties of Ba8Ta6(Ni1−x Mx)O24 (MZn and Mg) ceramics. J. Eur. Ceram. Soc. 2006, 26, 2045−2049. (15) Abakumov, A. M.; Tendeloo, G. V.; Scheglov, A. A.; Shpanchenko, R. V.; Antipov, E. V. The Crystal Structure of Ba8Ta6NiO24: Cation Ordering in Hexagonal Perovskites. J. Solid State Chem. 1996, 125, 102−107. (16) Cao, J.; Kuang, X.; Allix, M.; Dickinson, C.; Claridge, J. B.; Rosseinsky, M. J.; Iddles, D. M.; Su, Q. New 8-Layer Twinned Hex agonal P er ov sk ite M icr owa ve Die lec tr ic Ce ra mics Ba8Ga4−xTa4+0.6xO24. Chem. Mater. 2011, 23, 5058−5067. (17) Tao, F.; Genevois, C.; Lu, F.; Kuang, X.; Porcher, F.; Li, L.; Yang, T.; Li, W.; Zhou, D.; Allix, M. First 14-Layer Twinned Hexagonal Perovskite Ba14Mn1.75Ta10.5O42: Atomic-Scale Imaging of Cation Ordering. Chem. Mater. 2016, 28, 4686−4696. (18) Mallinson, P.; Claridge, J. B.; Iddles, D.; Price, T.; Ibberson, R. M.; Allix, M.; Rosseinsky, M. J. New 10-Layer Hexagonal Perovskites: Relationship between Cation and Vacancy Ordering and Microwave Dielectric Loss. Chem. Mater. 2006, 18, 6227−6238. (19) Shpanchenko, R. V.; Nistor, L.; Vantendeloo, G.; Vanlanduyt, J.; Amelinckx, S.; Abakumov, A. M.; Antipov, E. V.; Kovba, L. M. Structural Studies on New Ternary Oxides Ba8Ta4Ti3O24 AND Ba10Ta7.04Ti1.2O30. J. Solid State Chem. 1995, 114, 560−574. (20) Topas Academic, v5; Coelho: Brisbane, Queensland, Australia, 2012. (21) Hakki, B. W.; Coleman, P. D. A Dielectric Resonator Method of Measuring Inductive Capacities in the Millimeter Range. IEEE Trans. Microwave Theory Tech. 1960, 8, 402−410. (22) Denton, A. R.; Ashcroft, N. W. Vegard’s Law. Phys. Rev. A: At., Mol., Opt. Phys. 1991, 43, 3161−3164. (23) Evans, I. R.; Howard, J. A. K.; Evans, J. S. O. α-Bi2Sn2O7 - a 176 atom crystal structure from powder diffraction data. J. Mater. Chem. 2003, 13, 2098−2103. (24) Yokoi, A.; Ogawa, H.; Kan, A. Microwave dielectric properties of BaO−Ta2O5 − TiO2 system. J. Eur. Ceram. Soc. 2006, 26, 2069−2074. (25) Irvine, J. T. S.; Sinclair, D. C.; West, A. R. Electroceramics: Characterization by Impedance Spectroscopy. Adv. Mater. 1990, 2, 132−138. (26) Morrison, F. D.; Sinclair, D. C.; West, A. R. Characterization of lanthanum-doped barium titanate ceramics using impedance spectroscopy. J. Am. Ceram. Soc. 2001, 84, 531−538. (27) Lu, F.; Wang, X.; Pan, Z.; Pan, F.; Chai, S.; Liang, C.; Wang, Q.; Wang, J.; Fang, L.; Kuang, X.; et al. Nanometer-scale separation of d10 Zn2+ - layers and twin−shift competition in Ba8ZnNb6O24 - based 8 layered hexagonal perovskites. Dalton Trans. 2015, 44, 13173−13185.

(28) Shannon, R. D. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1976, 32, 751−767.

H

DOI: 10.1021/acs.inorgchem.8b00296 Inorg. Chem. XXXX, XXX, XXX−XXX