Extended dielectric constant step from −80 °C to 336 °C in the BaTiO₃–BiYO₃–Ba(Fe_0.5Nb_0.5)O₃ system

Abstract

Polycrystalline 0.8BaTiO₃–(0.2 − x)BiYO₃–xBa(Fe_0.5Nb_0.5)O₃ (0.8BT–(0.2 − x)BY–xBFN) (x = 0–0.04) ceramics were fabricated via a conventional solid-state reaction method. The phase transition, microstructure and dielectric properties were obtained by X-ray diffraction and dielectric measurements as functions of chemical composition and temperature. A progressive decline in ε_rmax with increasing x leads to nearly temperature-stable dielectric properties over a wide temperature range. For x ≥ 0, dielectric measurements indicate relaxor behavior. Increasing the BFN content effectively improves the dielectric temperature stability of BT ceramics. For x = 0.02, ε_r = 590 ± 15% from −48 °C to 326 °C, and tan δ ≤ 0.02 across the temperature range of −13 °C to 156 °C. For x = 0.04, ε_r = 540 ± 15% from −80 °C to 336 °C, and tan δ ≤ 0.02 across the temperature range of −22 °C to 231 °C. The plot of ac conductivity for x = 0.04 shows the NTCR character of the compounds, and the activation energy of the dc conductivity is not far from the electron from second ionization energy of oxygen vacancies.

---

1. Introduction

Increasing demand has triggered stable growth both in the quantity and variety of surface-mount chip components, which in turn has led to continuing advances in integrated circuit packaging technology.¹ Multilayer ceramic capacitors (MLCC) are widely used in all kinds of electronic products. Traditional high-relative-permittivity dielectrics based on barium titanate are specified to upper working temperatures of 125 °C, 150 °C, and 175 °C for X7R, X8R, and X9R capacitors, respectively.² However, even X9R materials do not fulfill the requirements under some harsh conditions such as oil drilling, aerospace and automotive environments.³ For example, the anti-lock brake system sensors on wheels are required to work in the temperature range of 150–250 °C, while the temperature in the cylinder is 200–300 °C. Therefore, it is important that materials scientists work to broaden the working temperature to reach an ultra-broad temperature range (−55 to 300 °C).⁴ This work has been sufficiently researched.

BaTiO₃ (BT) is a ferroelectric compound with perovskite-type structure and is applied in capacitors due to its good dielectric properties. Doping or formatting solid solutions with other metal oxides or compounds is an effective approach to improve the properties of BT. Among these additives, numerous solid solutions of BT with Bi-based perovskite compounds such as BaTiO₃–Bi(Zn_0.5Ti_0.5)O₃,⁵ BaTiO₃–Bi(Mg_0.5Zr_0.5)O₃,⁶ and BaTiO₃–BiZn_0.5Ti_0.5O₃–BiScO₃ (ref. 7) have been reported. Recently, Wang et al. reported that Ba(Fe_0.5Nb_0.5)O₃ (BFN) ceramics generally exhibit high dielectric constant steps over broad temperature and frequency intervals.⁸ In our past studies, the curves of relative permittivity vs. temperature flattened gradually above the maximum permittivity for BiYO₃ (BY). The distinctive properties of classical relaxor dielectrics are broad, frequency-dependent peaks in the plots of relative permittivity versus temperature and strong frequency dispersion in the dielectric loss tangent.^9–13

In this study, BY and BFN are added to BT ceramics to enhance the dielectric properties. Furthermore, the effects of the addition of BY and BFN on the relaxor behavior and dielectric temperature stability are also investigated.

2. Experimental procedure

0.8BaTiO₃–(0.2 − x)BiYO₃–xBa(Fe_0.5Nb_0.5)O₃ (0.8BT–(0.2 − x)BY–xBFN) (x = 0–0.04) ceramics were prepared via a conventional solid-state reaction technique. The starting materials were high-purity (99.9%) BaCO₃, TiO₂, Bi₂O₃, Fe₂O₃, Nb₂O₅, and Y₂O₃. The powders were dried in an oven at 200 °C for 24 h and then weighed according to the stoichiometric ratios. All batches were mixed by ball-milling with zirconia grinding media in distilled water for 4 h. After drying, the mixed powders were granulated with polyvinyl alcohol (PVA, 5 wt%), sieved through a 80–40 mesh nylon sieve and then pressed into disks with diameters of 10 mm and thicknesses of 1.5 mm. The samples were heat-treated at 600 °C for 30 min to eliminate the PVA and then sintered at 1300 °C for 3 h in air.

The phase structures of the ceramics were determined using X-ray diffraction (XRD D/max-2200 PC, RIGAKU, Japan), and the microstructures were studied by scanning electron microscopy (JSM-6700, JEOL Ltd., Tokyo, Japan). To characterize the electrical properties, the specimens were ground and polished to obtain parallel surfaces. In addition, silver paste was applied to opposite parallel faces, and coated pellets were fired at 600 °C for 10 min. The dielectric constant and loss were measured using a Hioki 3532-50, and conduction behavior as a function of temperature (−100 to 200 °C) and frequency (20 Hz to 2 MHz) were determined using an Agilent 4980A.

3. Results and discussion

3.1. X-ray diffraction study

Fig. 1 shows the XRD patterns of BT and 0.8BT–(0.2 − x)BY–xBFN (x = 0–0.04) ceramics at optimized sintering temperatures. In all patterns shown in Fig. 1, there are no traces of secondary peaks within the detection limit of the instrument, suggesting that BFN and BY diffused into the BT lattices to form homogenous perovskite solid solutions. The (002)/(200) diffraction peaks of the pure BT samples and samples with compositions of x ≤ 0.02 display obvious splitting, in accordance with tetragonal symmetry (P4mm space group).¹⁴ The crystal structure transforms to a pseudocubic cubic symmetry when x = 0.04. The (200) peaks shift continuously toward lower 2θ angles, indicating an increase in lattice parameters. Fig. 2 reveals the microstructures of 0.8BT–(0.2 − x)BY–xBFN (x = 0.02 and 0.04) ceramics at their optimized sintering temperatures. As seen in this figure, all samples are well sintered and exhibit a low level of porosity.

Fig. 1 X-ray diffraction patterns of BT and 0.8BT–(0.2 − x)BY–xBFN (x = 0–0.04) ceramics.

Fig. 2 SEM micrographs of (a) x = 0.02 and (b) x = 0.04 ceramics.

Fig. 3 shows the temperature dependence of relative permittivity (ε′) and dielectric loss tangent (tan δ) measured at various frequencies (1 kHz, 10 kHz, 100 kHz, and 1000 kHz) from room temperature to 400 °C. The BT composition exhibits a reasonably sharp Curie peak at ∼120 °C. Frequency dispersion occurs at x = 0, and the system transforms to a typical relaxor at x ≥ 0. With increasing x, the relative permittivities of the ceramics obviously decrease, and the permittivity–temperature plots exhibit near-flat responses for x ≥ 0. The BFN can obviously increase the temperature stability of BT-based ceramics. The maximum relative permittivity ε_rmax is ∼7727 for x = 0 at 1 kHz and decreases to ∼540 at x = 0.04 (Fig. 3). With increasing BFN content, the typical change in T_εrmax (the temperature of maximum permittivity) in 0.8BT–(0.2 − x)BY–xBFN ceramics can be ascribed to the generation of Ti vacancies.¹⁵

Fig. 3 Temperature dependence of relative permittivity (ε′) and loss tangent (tan δ) for (a) BT, (b) x = 0, (c) x = 0.02, and (d) x = 0.04.

The temperature dependence of the reciprocal of relative permittivity for BT and 0.8BT–(0.2 − x)BY–xBFN (x = 0–0.04) ceramics at 1 kHz is shown in Fig. 4. ΔT_m, which is given by ΔT_m = T_cw − T_m, where T_cw is the temperature at which ε starts to follow the Curie–Weiss law, and T_m is the temperature at which ε is maximized, is used to define the degree of deviation from the Curie–Weiss law. ΔT_m increases with x when x ≥ 0, which suggests increased dielectric diffuseness and enhanced relaxor behavior.¹⁶ Deviation from the Curie–Weiss law is gradually clear in relaxor ferroelectrics with x ≤ 0.04, as shown in Table 1.

Fig. 4 Temperature dependence of the reciprocal of relative permittivity (1/ε) for (a) BT, (b) x = 0, (c) x = 0.02, and (d) x = 0.04. The symbol represents experimental data points, and the solid straight black line shows the fit to the Curie–Weiss law.

Table 1 T_m, T_c, T_cw, ΔT_m, ε_m, γ, ΔT_dif and ΔT_relax for BT and 0.8BT–(0.2 − x)BY–xBFN (x = 0–0.04)

	Tm	Tc	Td	ΔTm	γ	ΔTdif	ΔTrelax
BT	118.85	112.58	121.79	2.94	1.01	2.00	2.54
x = 0	70.62	106.82	168.99	98.37	1.72	49.08	19.94
x = 0.02	68.04	83.06	175.86	107.82	1.80	60.12	29.9
x = 0.04	56.21	110.32	196.79	140.58	1.90	61.2	30.2

---

Curie–Weiss analysis (eqn (1))¹⁹ confirmed typical ferroelectric behavior for BT and 0.8BT–(0.2 − x)BY–xBFN (x = 0–0.04) ceramics with a linear fit to plots of 1/ε versus T at T > T_c and T_c ∼ T_o (Fig. 4):

where C is the Curie–Weiss constant, and T_o is the Curie–Weiss law.

The diffuse characteristics of the ferroelectric–paraelectric phase transition are known to deviate from the typical Curie–Weiss behavior and can be described by a modified Curie–Weiss relationship:^17–19

where ε_m is the maximum value of dielectric constant, ε is the dielectric constant at temperature T, T_m is the temperature at the peak of the dielectric constant, C is the Curie constant, and γ is an indicator of the degree of diffuseness, which gives information on the character of the phase transition. The value of γ is between 1 and 2; the values γ = 1 and 2 are related to a normal ferroelectric and an ideal relaxor,²⁰ respectively. Thus, the value of γ can also be used to characterize the relaxor behavior.

The plots of ln(1/ε − 1/ε_m) vs. ln(T − T_m) for BT and 0.8BT–(0.2 − x)BY–xBFN (x = 0–0.04) ceramics at 1 kHz are shown in Fig. 5 by fitting eqn (2) to calculate the γ values. The resulting values of γ are 1.72, 1.8, and 1.9 at 1 kHz for x = 0, x = 0.02, and x = 0.04, respectively, implying strong relaxor behavior. The diffuse phase transition and deviation from Curie–Weiss behavior may be attributed to disordering.²¹

Fig. 5 Plots of ln(1/ε − 1/ε_m) vs. ln(T − T_m) for (a) BT, (b) x = 0, (c) x = 0.02, and (d) x = 0.04. The symbol represents experimental data points, and solid lines show the fits of the data.

The diffuseness in the phase transition is described by an empirical parameter ΔT_dif, which is defined as ΔT_{dif1 kHz} = T_{0.9εm1 kHz} − T_{εm1 kHz}, where T_{εm1 kHz} denotes the temperature of the dielectric maximum, and T_{0.9εm1 kHz} represents the higher temperature of 90% of the dielectric maximum at 1 kHz. Yet another parameter, which is used to characterize the degree of relaxation behavior in the frequency range of 1–1000 kHz, is described as ΔT_relax = T_{m1000 kHz} − T_{m1 kHz}.²² The values of ΔT_dif and ΔT_relax for BT and 0.8BT–(0.2 − x)BY–xBFN (x = 0–0.04) ceramics are shown in Table 1. To recapitulate, the above empirical characterization with the Curie–Weiss law (ΔT_m) and parameters (γ, ΔT_dif and ΔT_relax) indicates that 0.8BT–(0.2 − x)BY–xBFN are indeed relaxors with diffuse phase transitions and frequency dispersion.

One of the most desirable properties for MLCC materials is the temperature stability of dielectric permittivity. Fig. 6 shows the temperature coefficients for 0.8BT–(0.2 − x)BY–xBFN (x = 0–0.04) ceramics at their maximum sintered densities. Increasing the BFN content to x = 0.04 produces a broad relaxor dielectric peak. At x = 0, the temperature variation in ε_r lay within ±15% of the value of 830 from −33 °C to 173 °C, and the tan δ values are less than or equal to 0.02 across the temperature range of 27–276 °C. For x = 0.02, ε_r = 590 ± 15% from −48 °C to 326 °C, and tan δ ≤ 0.02 across the temperature range of −13 to 156 °C. Furthermore, the temperature range for tan δ ≤ 0.02 increases to −22 to 231 °C at x = 0.04, while Δε/Δε₂₅ ≤ ±15% in the temperature range of −80 to 336 °C for x = 0.04. This indicates that increasing the BFN content can improve the thermal stability of BT ceramics.

Fig. 6 Temperature coefficient of ε for 0.8BT–(0.2 − x)BY–xBFN (x = 0–0.04) ceramics; ε₂₅ is the dielectric permittivity at 25 °C.

Fig. 7 shows the polarization–electric field response measured at room temperature. BT exhibits a typical ferroelectric hysteresis loop. With increasing x, the polarization–electric loops become narrow and exhibit a low dielectric loss, indicating that the addition of BFN eliminates the ferroelectric properties. These results are in accordance with the XRD results indicating that the paraelectric and cubic phases exist in the ceramics; thus, the forms of the polarization–electric plots become narrow ellipsoids with increasing x.

Fig. 7 Polarization–electric field (P–E) loops for (a) BT and (b) 0.8BT–(0.2 − x)BY–xBFN (x = 0–0.04) ceramics.

The plots of Z′ versus Z′′ (Nyquist or Cole–Cole plots) at different temperatures are shown in Fig. 8. All the semicircles are depressed instead of centered on the real axis, which indicates a non-Debye-type relaxation mechanism. Two semicircles can be observed in the two diagrams, indicating that two distinct dielectric relaxation processes exist in the samples, and that the suppressed semicircular arcs indicate two different contributions from the grain interior and grain boundary.

Fig. 8 Nyquist plots of Z′ vs. Z′′ for 0.8BT–0.16BY–0.04BFN ceramics at different temperatures (inset shows the equivalent circuit model used here in association with the brick layer model).

In order to obtain reliable values for the grain resistance and grain boundary and establish a connection between the microstructure and electrical properties, we employed the equivalent circuit model shown in Fig. 8, which is based on the brick layer model.²³ In Fig. 8, R_s is the resistance of lead used in the equipment, R_g is the resistance associated with the grain, and CPE is a constant-phase element indicating the departure from the ideal Debye-type model. The CPE admittance is Y_CPE = A_o(jω)ⁿ = Aωⁿ + jBωⁿ, with

A = A_o cos(nπ/2), B = A_o sin(nπ/2) (3)

where A_o and n are parameters depending only on temperature, and A_o confines the magnitude of the dispersion.

According to Jonscher, the origin of the frequency dependence of conductivity is the relaxation of ionic atmosphere after the movement of the particle. The variations in ac conductivity as functions of frequency at different temperatures are shown in Fig. 9. Jonscher attempted to explain the behavior of ac conductivity using the following universal power law:²⁴

σ(ω) = σ₀ + Aωⁿ (4)

where σ(ω) is the total conductivity, σ₀ is the frequency independent or dc part, which is related to dc conductivity and is the second term of the CPE, n (0 < n < 1) represents the degree of interaction between mobile ions with the lattices around them, and A determines the strength of polarization.²⁵

Fig. 9 Frequency dependence of ac conductivity at different temperatures for 0.8BT–0.16BY–0.04BFN ceramics.

It is reported that the value of n can be larger than unity, and there is no physical argument to restrict the value of n below 1.²⁶ In some glassy, mixed oxides and single crystals, the reported value of n is greater than 1. We have fitted conductivity data using eqn (4) with origin software using the non-linear curve fitting option, and the obtained values of σ₀, A₀, and n are tabulated in Table 2. It can be seen that with decreasing temperature, the value of n decreases.

Table 2 Ac conductivity parameters

Temperature	σ0	A	n
400 °C	1.03099 × 10^−4	2.84424 × 10^−8	0.66679
375 °C	5.53743 × 10^−5	1.64396 × 10^−8	0.69812
350 °C	3.0749 × 10^−5	1.2738 × 10^−8	0.70998
325 °C	1.58963 × 10^−5	1.09789 × 10^−8	0.71346
300 °C	7.18377 × 10^−6	7.63512 × 10^−9	0.73059

---

Extrapolating these curves at low frequencies gives the dc conductivity (σ_dc). The resulting σ_dc is plotted as a function of reciprocal temperature in Fig. 10, and the plot obeys the Arrhenius relation:

σ = σ₀ exp(−E_cond/K_BT) (5)

where σ₀ is the pre-exponential term, and E_cond is the conduction activation energy. The experimental value of conduction activation (E_cond = 0.8 eV) is close to 1 eV, which is the conduction activated energy of an electron from the second ionization energy of oxygen vacancies.²⁷ Therefore, the conducting species in this temperature range is likely to come from the second ionization of oxygen vacancies V_Ö, as reported by Li et al.²⁸

Fig. 10 Temperature dependence of conductivity for 0.8BT–0.16BY–0.04BFN ceramics.

4. Conclusions

The electrical properties of 0.8BT–(0.2 − x)BY–xBFN (x = 0–0.04) ceramics have been studied. The XRD patterns verify that all samples exhibit a single-phase perovskite structure, and the transformation from tetragonal symmetry to pseudocubic structure occurs at 0.02 ≤ x ≤ 0.04. The composites with 0 ≤ x ≤ 0.04 are relaxor dielectrics. The stability in the values of relative permittivity over a wide temperature range is demonstrated for x = 0.04, and the temperature range of tan δ ≤ 0.02 is expanded to −22 to 231 °C at x = 0.04. Furthermore, Δε/Δε₂₅ ≤ ±15% in the temperature range of −80 to 336 °C. The polarization–electric plots develop narrow, ellipsoidal forms with increasing x. Comparisons of dielectric properties with other materials proposed for high-temperature capacitor applications indicate that such a material system could be suitable for high-temperature stable dielectric applications.

Acknowledgements

This research was supported by the National Natural Science Foundation of China (51372144) and the Key Program of Innovative Research Team of Shaanxi Province (2014KCT-06).

References

1. S. F. Wang, J. H. Li, Y. F. Hau, Y. C. Wu, Y. C. Lai and M. H. Chen, J. Eur. Ceram. Soc., 2013, 33, 1793–1799.
2. A. Zeb and S. J. Milne, J. Am. Ceram. Soc., 2013, 96, 3701–3703.
3. J. B. Lim, S. Zhang, N. Kim and T. R. Shrout, J. Am. Ceram. Soc., 2009, 92, 679–682.
4. L. X. Li, J. X. Chen, D. Guo, N. Zhang, M. J. Wang and Y. R. Liu, Ceram. Int., 2014, 40, 12539–12543.
5. N. Raengthon and D. P. Cannn, J. Am. Ceram. Soc., 2011, 95, 1064–1612.
6. A. Zeb and S. J. Milne, J. Eur. Ceram. Soc., 2014, 34, 3159–3166.
7. N. Raengthon, T. Sebastian, D. Cumming, I. M. Reaney and D. P. Cann, J. Am. Ceram. Soc., 2012, 95, 3554–3561.
8. Z. Wang, M. R. Fang, H. J. Li, Y. F. Wen, C. Wang and Y. P. Pu, Compos. Sci. Technol., 2015, 117, 410–416.
9. A. A. Bokov and Z.-G. Ye, J. Mater. Sci., 2006, 41, 31–52.
10. H. C. Wang and W. A. Schulze, J. Am. Ceram. Soc., 1990, 73, 825–832.
11. V. V. Shvartsman and D. C. Lupascu, J. Am. Ceram. Soc., 2012, 95, 1–26.
12. C. Laulhé, F. Hippert, J. Kreisel, M. Maglione, A. Simon, J. L. Hazemann and V. Nassif, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 74, 014106.
13. T. Badapanda, S. K. Rout, L. S. Cavalcante, J. C. Sczancoski, S. Panigrahi, E. Longo and M. Siu Li, J. Phys. D: Appl. Phys., 2009, 42, 175414.
14. X. L. Chen, J. Chen, D. D. Ma, L. Fang and H. F. Zhou, Ceram. Int., 2015, 41, 2081–2088.
15. S. F. Wang, J. H. Li, Y. F. Hsu, Y. C. Wu, Y. C. Lai and M. H. Chen, J. Eur. Ceram. Soc., 2013, 33, 1793–1799.
16. J. Ravez and A. Simon, J. Solid State Chem., 2001, 162, 260–265.
17. F. D. Morrison, D. C. Sinclain and A. R. West, J. Appl. Phys., 1999, 86, 6355–6366.
18. N. Raengthon and D. P. Cannn, J. Am. Ceram. Soc., 2011, 95, 1064–1612.
19. F. Zhu, T. A. Skidmore, A. J. Bell, T. P. Comyn, C. W. James and M. Ward, et al., Mater. Chem. Phys., 2011, 129, 411–417.
20. Q. Zhang, Z. Li, F. Li and Z. Xu, J. Am. Ceram. Soc., 2011, 94, 4335–4339.
21. S. Mahajanl and O. P. Thakur, J. Phys. D: Appl. Phys., 2009, 42, 065413.
22. D. Liu, Y. P. Pu, J. L. Li and X. Shi, Ceram. Int., 2013, 39, 593–596.
23. Q. Q. Ke, X. J. Lou and J. Wang, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 82, 024102.
24. M. A. L. Nobre and S. Lanfredi, J. Appl. Phys., 2003, 93, 5576.
25. M. Vijayakumar, S. Selvasekarapandian, M. S. Bhuvaneswari, G. Hirankumar, G. Ramprasad, R. Subramanian and P. C. Angelo, Phys. B, 2003, 334, 390–397.
26. D. K. Pradhan, B. Behera and P. R. Das, J. Mater. Sci.: Mater. Electron., 2012, 23, 779–785.
27. Z. Wang, X. M. Chen, L. Ni, Y. Y. Liu and X. Q. Liu, Appl. Phys. Lett., 2007, 90, 102905.
28. G. B. Li, S. X. Liu, F. H. Liao, S. J. Tian, X. P. Jing, J. H. Lin, Y. Uesu, K. Kohn, K. Saitoh, M. Terauchi, N. L. Di and Z. H. Cheng, J. Solid State Chem., 2004, 177, 1695–1703.

---