Phase, microstructural characterization and dielectric properties of Ca-substituted Sr₅Nb₄TiO₁₇ ceramics

Abstract

The effect of Ca substitution for Sr on the phase, microstructure and microwave dielectric properties of the Sr_5−x Ca _x Nb₄TiO₁₇ composition series was investigated using X-ray diffraction (XRD), scanning electron microscopy (SEM), an LCR meter, and vector network analyzer. Below 1450 °C, Sr_5−x Ca _x Nb₄TiO₁₇ (x = 1, 2, 3, or 4) compositions formed single-phase Sr₄CaNb₄TiO₁₇, Sr₃Ca₂Nb₄TiO₁₇, Sr₂Ca₃Nb₄TiO₁₇, and SrCa₄Nb₄TiO₁₇ ceramics, respectively. At x = 0 and 5, Sr₅Nb₄TiO₁₇ and Ca₅Nb₄TiO₁₇ formed, but along with Sr₂Nb₂O₇ (at x = 0) and CaNbO₃ and CaNb₂O₆ (at x = 5) secondary phases. Above 1450 °C, all the compositions formed two-phase ceramics. At low frequencies, a phase transition was observed in the composition Sr₅Nb₄TiO₁₇. The substitution of Ca for Sr enabled processing of highly dense Sr₂Ca₃Nb₄TiO₁₇, with ε_r ~ 53.4, τ_f ~ −6.5 ppm/°C and Q _u  × f _o  ~ 1166 GHz. Further investigations are required to improve the quality factor of these ceramics for possible microwave applications.

Introduction

Ceramic dielectric resonators (DRs) used at microwave frequencies have been utilized for the last few decades in mobile telecommunications devices. The three key parameters needed for a material to be used as a microwave DR are (a) a high dielectric constant (especially for handsets) to reduce the size of the microwave component, (b) low dielectric loss (implying a high quality factor, Q _u , where Q _u  = 1/tanδ) for fine tuning and (c) a near-zero temperature coefficient of resonant frequency to ensure temperature stability of the microwave component. Only a few materials meet this industrial requirement. Examples include Ba(Zn_1/3Ta_2/3)O₃, BaTi₄O₉, Ba₂Ti₉O₂₀, (Zr,Sn)TiO₄, and Ba_6−3x Re_8+2x Ti₁₈O₅₄ (Re = Nd, Sm, La), all of which are used at microwave frequencies in various filters and oscillators in telecommunication devices [1]. Generally, a high dielectric constant material exhibits high loss and vice versa [2, 3]; therefore, it is difficult to have materials exhibiting all the above mentioned three properties simultaneously and there is a constant demand for new materials with better dielectric properties.

A number of dielectric ceramics belonging to the A _n B _n O_3n+2 series are known to exhibit good microwave properties. For example, (CaLa₄)Ti₅O₁₇ with n = 5 and CaLa₈Ti₉O₃₁ with n = 4.5 (i.e., (Ca_0.5La₄)(Ti_4.5)O_15.5) are layered perovskites, with dielectric constants ε_r = 55.2 and 54.9, quality factors Q _u  × f _o  = 17359 GHz (at 3.67 GHz) and 19345 GHz (at 3.65 GHz), and temperature coefficients of resonant frequency τ_f = −20 and −6 ppm/°C, respectively [2]. Substitution of 1 wt% of Ca by Zn in CaLa₄Ti₅O₁₇ has been reported to give ε_r = 57.6, Q _u  × f _o  = 17100 GHz and τ_f ~ 4.9 ppm/°C [3]. Similarly, the addition of 0.5 wt% CuO as a sintering aid to Ca_0.99Zn_0.01La₄TiO₁₇ resulted in ε_r = 57, Q _u  × f _o  = 15000 GHz and τ_f ~ −8.16 ppm/°C at 1450 °C in comparison with 1500 °C for pure Ca_0.99Zn_0.01La₄TiO₁₇ [4]. The substitution of 1 wt% of Ca by Mg in CaLa₄Ti₅O₁₇ gave ε_r = 56.3, Q _u  × f _o  = 12300 GHz and τ_f ~ −9.6 ppm/°C [5]. Anjana et al. [6] reported that Ca₅Nb₄TiO₁₇ and Ca₃Mg₂Nb₄TiO₁₇ have ε_r = 45, Q _u  × f _o  = 17600 GHz and τ_f ~ −113 ppm/°C and ε_r = 37.5, Q _u  × f _o  = 22500 GHz and τ_f ~ −4.3 ppm/°C, respectively. More recently, Joseph et al. [7] reported Ca₅Nb₄TiO₁₇ and Ca₅Ta₅TiO₁₇ to have ε_r = 44.9, Q _u  × f _o  = 17600 GHz and τ_f ~ −112.9 ppm/°C and ε_r = 40.1, Q _u  × f _o  = 16450 GHz and τ_f ~ −53.6 ppm/°C, respectively.

Slobodyanik et al. [8] synthesized Ca_5−x Sr _x TiNb₄O₁₇ with 0 ≤ x ≤ 5 via co-precipitation of hydroxocarbonates, observed the beginning of Sr₅TiNb₄O₁₇ formation at temperatures above 1300 °C, and noted the requirement of temperatures well above 1400 °C for the formation of single-phase Sr₅TiNb₄O₁₇ ceramics. They did not measure the dielectric properties. Iqbal and Reaney [9] reported the dielectric constant ε_r ~ 57 (at 10 kHz), Q _u  × f _o  ~ 1070 MHz and TCε ~ −0.0007 ppm/°C for Sr₅Nb₄TiO₁₇ with no Ca. The crystal structure of Sr₅Nb₄TiO₁₇ is reported to be orthorhombic with space group Pnnm and lattice parameters a = 5.6614 Å, b = 32.515 Å, c = 3.9525 Å, and Z = 2 [10].

Keeping in view the high loss of Sr₅Nb₄TiO₁₇, the effect of Ca substitution on the phase, microstructure and dielectric properties of Sr₅Nb₄TiO₁₇ has been investigated as an attempt to improve Q _u  × f _o for possible microwave applications.

Experimental procedure

Sr_5−x Ca _x Nb₄TiO₁₇ compositions (x = 0, 1, 2, 3, 4, and 5) were prepared via a mixed-oxide solid-state route. Laboratory reagent grade SrCO₃, CaCO₃ (Fisher Scientific, > 99%), Nb₂O₅ (BDH Chemicals Ltd., > 99.9%) and TiO₂ (Aldrich Chemicals, 99 + %), powders were weighed according to the stoichiometric ratios. The mixtures were milled for 60 h in disposable polyethylene mill jars with cylindrical Y-toughened ZrO₂ balls as grinding media and 2-praponol as lubricant using a conventional horizontal ball mill. The slurry was dried in an oven at ~95 °C overnight and the resulting powder samples were sieved to dissociate agglomerates (if any). Thermo-gravometric (TG) analysis and differential thermal analysis (DTA) were performed from room temperature to 1200 °C at 5 °C/min (for x ≤ 1) and 10 °C/min (for 2 ≥ x ≥ 5) for as-mixed Sr_5−x Ca _x Nb₄TiO₁₇ compositions to determine the weight loss and phase transformation temperatures. The Sr₅Nb₄TiO₁₇ composition was calcined at 990 °C for 2 h. The other batches were calcined at 1350 °C for 2 h. Heating/cooling rates were 10 °C/min. The calcined powders were re-milled for 30 min and dried before uniaxial pressing into 13 mm diameter pellets at 140 MPa in a stainless steel die. The pellets were sintered in air at 1400–1550 °C for 2 h at heating/cooling rate of 10 °C/min. Sintered pellets were pulverized in a pestle and mortar system and ground into fine powders for XRD. A STOE PSD X-ray diffractometer with CuKα₁ radiation (λ = 1.540598 Å) was used for phase analysis and measurement of lattice parameters. The sintered samples were finely polished and thermally etched at temperatures 10% less than the corresponding sintering temperatures and gold coated for microstructural analysis. Microstructural characterization was performed using a JEOL 6400 SEM operating at 20 kV, equipped with an energy dispersive X-ray spectroscopy (EDS) detector (Link, Oxford Instruments). The apparent density of the sintered pellets was measured using the Archimedes method. The theoretical densities of the compositions were calculated by using Eq. (1);

$$ {\text{R}}_{\text{th}} = \left( {ZM} \right)/\left( {A_{g} V} \right) $$

(1)

where Z is the number of formula units per unit cell, M is the molecular weight, V is the volume of the unit cell and A _g is the Avogadro number (6.022 × 10²³ atoms/mole).

The faces of the sintered pellets were coated with gold paste and heated to 800 °C for 2 h at 10 °C/min. The dielectric constant was measured from 1 kHz to 1 MHz at temperatures ranging from 20 to 800 °C using a HP 4284A LCR meter. For loss tangent measurements, the sintered pellets were mounted on a low-loss quartz single crystal at the centre of a 20 mm diameter cavity made of brass to avoid conduction loss. The microwave energy was coupled to the test piece using two ports. Measurements were taken in transmission mode using a network analyzer (R3767CH, Advantest). The average temperature coefficient of resonant frequency, τ _f , between 20 and 70 °C was calculated using Eq. (2);

$$ \tau_{\text{f}} = \left( {f_{ 2} - f_{ 1} } \right)/\left( {f_{ 1} \Updelta T} \right) $$

(2)

where f ₁ is the resonant frequency at 20 °C and f ₂ at 70 °C, and ∆T is the difference between the initial (~20 °C) and final temperature (~70 °C) of measurement.

Results and discussion

TG/DTA analysis

TG/DTA curves from as mix-milled 3SrCO₃:2CaCO₃:2Nb₂O₅:1TiO₂ powders are shown in Fig. 1. A careful examination of the TGA curve indicated the beginning of mass loss at 650 °C, which continued till 750 °C, due to the decomposition of CaCO₃ [11]. Another downward slope was observed just after 750 °C which continued until 902 °C, probably due to the decomposition of SrCO₃ [12]. The observed increase in the decomposition temperature of SrCO₃ from the previously reported [13] 880–902 °C in this study may be due to the increase in the heating rate from 5 to 10 °C/min [11, 14]. Two endotherms were observed in the DTA curve, at 750 and 902 °C, consistent with the temperatures at which the downward sloping of TGA curves ended. A total of ~18.60% mass loss was recorded in the entire heating cycle from 30 to 1200 °C which is consistent (within ± 1%) with the total CO₂ forming during the reaction given in Eq. (3)

$$ 3 {\text{SrCO}}_{ 3}{:}2 {\text{CaCO}}_{ 3}{:}2 {\text{Nb}}_{ 2} {\text{O}}_{ 5}{:}1 {\text{TiO}}_{ 2} \to {\text{Sr}}_{ 3} {\text{Ca}}_{ 2} {\text{Nb}}_{ 4} {\text{TiO}}_{ 1 7} + 5 {\text{CO}}_{ 2} $$

(3)

Fig. 1

TG/DTA curves from the as mix-milled 3SrCO3:2CaCO3:2Nb2O5:1TiO2 powders

X-ray diffraction

The room-temperature XRD patterns recorded for the calcined and sintered (at 1450 and 1500 °C) Sr_5−x Ca _x Nb₄TiO₁₇ (x = 0–5) samples are shown in Figs. 2 and 3. The major phase observed in the sample with x = 0 and calcined at 990 °C/2 h was Sr₅Nb₄TiO₁₇; however, a few XRD peaks matched JCPDS Card# 28–1247 for Sr₂Nb₂O₇ indicating incomplete reaction. XRD patterns from compositions with x = 1–4 calcined at 1350 °C/2 h were similar and matched JCPDS Card# 87–1170 for Sr₅Nb₄TiO₁₇ [10] indicating the formation of single-phase ceramics. It is noticeable that no JCPDS cards could be found for Ca₅Nb₄TiO₁₇ but the observed major XRD peaks for the composition with x = 5, matched JCPDS Card# 51–412 for Ca₅Nb₅O₁₇. Being isostructural with Ca₅Nb₅O₁₇, the peaks due to Ca₅Nb₄TiO₁₇ could be indexed according to a model analogous to the Ca₅Nb₅O₁₇ reported in JCPDS Card# 51–412. A few low intensity peaks matching the JCPDS Card# 47–1668 for orthorhombic CaNbO₃ and JCPDS Card# 39–1392 for CaNb₂O₆ were also observed which indicated second phase formation at x = 5.

Fig. 2

XRD patterns from Sr_5−x Ca _x Nb₄TiO₁₇ samples with x = 0–5 sintered at 1450 °C for 2 h, showing the presence of single-phase compounds for x = 0, 1, 2, 3, and 5 and a mixture of layered perovskite (Sr_1−x Ca _x ) _n (Nb,Ti) _n O_3n+2 phases with n = 5 and 6 for the Sr_5−x Ca _x Nb₄TiO₁₇ composition with x = 4

Fig. 3

XRD patterns from Sr_5−x Ca _x Nb₄TiO₁₇ samples with x = 0–5 sintered at 1500 °C for 2 h, showing the presence of a mixture of (Sr_1−x Ca _x ) _n (Nb,Ti) _n O_3n+2 compounds with n = 5 and 6

The XRD of the samples sintered at 1450 °C for 2 h revealed the formation of single phases for the compositions with x = 0, 1, 2, 3, and 5, while a mixture of n = 5 and 6 layered perovskites was found for the composition with x = 4. No reference data cards could be found for the Ca versions of the compounds expected from the compositions with x = 4; however, these XRD peaks matched JCPDS Card# 87–1170 for Sr₅Nb₄TiO₁₇ and 49–544 for Sr₆Nb₄Ti₂O₂₀ with appropriate shifting of peak positions due to Ca substitution. Thus, SrCa₄Nb₄TiO₁₇ and (Sr,Ca)₆Nb₄Ti₂O₂₀, being isostructural with the compounds Sr₅Nb₄TiO₁₇ and Sr₆Nb₄Ti₂O₂₀, respectively, were indexed according to the models analogous to these compounds reported in JCPDS Card# 87–1170 and 49–544 (Fig. 2). Above 1450 °C, even the compositions with x = 1, 2, 3, and 5 were not single phase and XRD of samples sintered at 1500 °C for 2 h revealed the presence of a mixture of (Sr_1−x Ca _x ) _n (Nb,Ti) _n O_3n+2 perovskite compounds with n = 5 and 6, i.e., Sr₅Nb₄TiO₁₇ (JCPDS Card# 87–1170) and Sr₆Nb₄Ti₂O₂₀ or Ca₆Nb₄Ti₂O₂₀ (JCPDS Card# 49–544) for x = 0 or 5, respectively. It is noticeable that at 1500 °C, the major phases observed in the compositions with x = 1, 2, 3, and 4 were Sr₄CaNb₄TiO₁₇, Sr₃Ca₂Nb₄TiO₁₇, Sr₂Ca₃Nb₄TiO₁₇, and SrCa₄Nb₄TiO₁₇, respectively; however, the nature of the secondary phases, i.e., the layered perovskites with n = 6 in compositions with x = 1−4, could not be identified. The crystal structure of the compounds with x = 0−4 was orthorhombic (Pnnm) and the compound with x = 5 crystallized into a monoclinic structure with space group P2₁/b or P2₁/c [8]. Manan et al. [13, 15] and Levin et al. [16] have also reported the formation of a mixture of Sr _n (Nb,Ti) _n O_3n+2 compounds with n = 5 and 6 in the SrTiO₃–Sr₂Nb₂O₇ system with no Ca⁺² substitution. The decomposition into secondary phases is more visible in the compositions with x = 4 and 5 as compared to the compositions with x = 0−3 sintered at 1500 °C/2 h (Fig. 3).

The compounds in the system Sr_5−x Ca _x Nb₄TiO₁₇ crystallize into A₅B₅O₁₇ structure where the large Sr/Ca cations occupy the A site with coordination of 12 and the Ti/Nb cations occupy the B site with coordination of 6. The resulting structure consists of perovskite-like blocks each of which is five octahedra thick and separated by an extra layer of oxygen. With increasing Ca⁺² content from x = 1 to 5, the XRD peaks shifted toward higher diffraction angles, i.e., smaller interplanar spacings due to the smaller size of Ca⁺² (1.34 Å) substituted for the larger Sr⁺² (1.44 Å) ion [17, 18] resulting in a decrease in the lattice constants of the unit cell (Fig. 4).

Fig. 4

The variation of unit cell parameters with composition for the Sr_5−x Ca _x Nb₄TiO₁₇ ceramics

The unit cell parameters of the Sr_5−x Ca _x Nb₄TiO₁₇ ceramics refined by the least squares method are given in the Table 1. The lattice parameters ‘a’, ‘b’, and ‘c’ decreased from 5.6614 to 5.5360 Å, 32.515 to 32.151 Å, and 3.9525 to 3.8727 Å, respectively, with increase in ‘x’ from 0 to 4. For x = 5, ‘a’ and ‘c’ increased to 7.6889 and 32.253 Å, respectively, and ‘b’ decreased to 5.4763 Å. This appears consistent with the observed change in structure from orthorhombic (Pnnm) [10] to monoclinic (P2₁/c) [19]. Consequently, the normalized cell volume (V _m  = V _{unit cell} /Z) of the Sr_5−x Ca _x Nb₄TiO₁₇ unit cell decreased from 363.788 (x = 0) to 339.515 Å³ (x = 5) due to the substitution of the small Ca⁺² ion in the unit cell for the larger Sr⁺² ion as shown Fig. 5. The alternative orthorhombic arrangement can be obtained by reorienting the monoclinic unit cell with a _m || −a _o, b _m || c _o and c _m || b _o, where the subscripts ‘m’ and ‘o’ refer to the modified and original unit cell parameters.

Table 1 Microwave dielectric properties of Sr_5−x Ca _x Nb₄TiO₁₇ (x = 0–5)

Fig. 5

The plot of normalized unit-cell volume versus x for Sr_5−x Ca _x Nb₄TiO₁₇ (x = 0–5)

Density measurement

The variation in the apparent density of Sr_5−x Ca _x Nb₄TiO₁₇ compounds (x = 0–5) with increase in sintering temperature is shown in Fig. 6. The density of each composition first increased with increase in the sintering temperature from 1400 to 1500 °C and then decreased with further increase in the sintering temperature to 1550 °C. This indicated that the density of each composition reached a maximum at ~1500 °C. A previous study [15] reported a maximum density of ~4.46 g/cm³ for Sr₅Nb₄TiO₁₇ at ~1450 °C. In this study, a maximum density of ~4.74 g/cm³ was achieved for the composition with x = 2, indicating an increase in the density due to the substitution of Ca for Sr.

Fig. 6

The variation in the apparent density of Sr_5−x Ca _x Nb₄TiO₁₇ (x = 0–5) compounds with sintering temperature, showing the maximum density for the compound with x = 2

Microstructural analysis

The secondary electron images (SEI) from thermally etched Sr_5−x Ca _x Nb₄TiO₁₇ (x = 0–5) samples sintered at ~1500 °C are shown in Fig. 7. The edges of the grains observed in sintered Sr_5−x Ca _x Nb₄TiO₁₇ ceramics with x = 0 and 1 appeared less sharp in comparison to those with x ≥ 2. As evident from Fig. 6, the density of these ceramics (with x ≤ 1) was also observed to be lower than the other ceramics sintered at the same temperature, indicating the effect of calcium substitution upon sintering. In general, the grains were irregular in shape with varying size. The average grain size observed in the microstructure of the compositions with x = 2 and 3 was ~1 × 2–1.5 × 5 μm²; however, a small number of elongated (~2.5 × 23 μm²) grains were also observed. The number of elongated grains increased with increase in x (i.e., x = 4, 5) consistent with the XRD results showing the splitting of the compound into layered n = 5 and 6 perovskites. The grains in the microstructure of the compositions with x = 2 and 3 appeared more compact than the other compositions, consistent with the observed higher density of these compounds in comparison to the others.

Fig. 7

SEM images from thermally etched Sr_5−x Ca _x Nb₄TiO₁₇ (x = 0–5) samples showing irregular grains for compositions with x = 0 and 1

Dielectric properties at low frequencies

In spite of forming multi-phasic ceramics, the dielectric properties of the compositions sintered at 1500 °C were better due to their relatively higher densities than those sintered at 1450 and 1550 °C. The variation in the dielectric constant (ε_r) and loss tangent (tanδ) with temperature measured at 1 kHz–1 MHz for Sr_5−x Ca _x Nb₄TiO₁₇ (with x = 0–5) sintered at 1500 °C/2 h is shown in Figs. 8 and 9. Strong anomalies in ε_r and tanδ for the x = 0 compound, which may be due to the ferroelectric to para electric phase transition at ~531.3 °C, were observed. Similar behavior has been reported in ε_r and tanδ measured for Ba_5−x Sr _x DyTi₃V₇O₃₀ (with x = 0–5) ceramics at 435, 355, 326, 82, and 46 °C [20]. The observed decrease in the dielectric constant with increase in frequency may be due to the vanishing of the contribution from interfacial and ionic polarization. Furthermore, a decrease in ε_r was observed with increase in the Ca content due to the reduced ionic polarizability of Ca⁺² (3.17 Å³) than the Sr⁺² (4.25 Å³) [21]. The observed increase in tanδ with increase in temperature may be due to the increased conductivity of the compounds. [20]. The tanδ was observed to be the lowest for the composition with x = 5.

Fig. 8

Variation of relative permittivity of Sr_5−x Ca _x Nb₄TiO₁₇ (x = 0–5) with temperature at 1 kHz–1 MHz, showing anomalies at 531.3, 495.8, 492.5, and 514.9 °C for x = 0, 3, 4, and 5, respectively

Fig. 9

Variation of loss tangent for the Sr_5−x Ca _x Nb₄TiO₁₇ compositions (x = 0 and 5) with temperature at 10 kHz–1 MHz showing low loss for the ceramics with x = 5

Microwave dielectric properties

The microwave dielectric properties of the Sr_5−x Ca _x Nb₄TiO₁₇ (x = 0−5) ceramics are given in Table 1. The unexpectedly low dielectric constant (48) observed for pure Sr₅Nb₄TiO₁₇ with no substitution may be due to its having the lowest density of all the compositions in this study. It is evident from Table 1 that the substitution of Ca decreased the dielectric constant and temperature coefficient of resonant frequency; however, the Q _u  × f _o was not highly improved under the processing conditions employed in this study. The temperature coefficient of resonant frequency, τ _f , for the composition with x = 1 was positive (+12.9 ppm/°C) when sintered at 1400 °C while negative (−2.5 and −14.5 ppm/°C) when sintered at 1450 and 1500 °C, respectively. The nature of the negative τ _f is not known. The substitution of Ca for Sr increased the Q _u  × f _o value from 447 (x = 0) to 3087 GHz (x = 5), decreased the dielectric constant from 48 (x = 0) to 37.6 (x = 5). The substitution of Ca for Sr is known to cause tilting of octahedra [1] which may be the possible reason for the observed decrease in τ _f from 162.4 ppm/°C for Sr₅Nb₄TiO₁₇ to −132.5 ppm/°C for Ca₅Nb₄TiO₁₇.

Conclusions

The effect of Ca substitution for Sr on the phase, microstructure and dielectric properties of Sr₅Nb₄TiO₁₇ was investigated. XRD revealed the formation of single-phase ceramics for the compositions with x = 1–4. Compositions with x = 0 and 5 showed the formation of secondary Sr₂Nb₂O₇ (x = 0), and CaNbO₃ and CaNb₂O₆ (x = 5) phases along with the parent Sr₅Nb₄TiO₁₇ and Ca₅Nb₄TiO₁₇ phases, respectively, upon calcination. XRD of the sintered samples with x = 0–5 showed the decomposition into n = 5 and 6 layered perovskites with the general formula (Sr_1−x Ca _x ) _n (Nb,Ti) _n O_3n+2 at 1500 °C. Such decomposition was most evident in the compositions with x = 4 and 5. The layered perovskite with n = 6 in the compositions with x = 0 and 5 were Sr₆Nb₄Ti₂O₂₀ and Ca₆Nb₄Ti₂O₂₀, respectively; however, the nature of the layered perovskites with n = 6 in the compositions with x = 1–4 could not be identified. The normalized cell volume of the unit cell decreased with increase in the Ca content. The Ca substitution decreased the dielectric constant and τ_f and increased the Q _u  × f _o (GHz) value.