Enhanced Thermal Stability in Dielectric Properties of NaNbO 3 –Modiﬁed BaTiO 3 –BiMg 1/2 Ti 1/2 O 3 Ceramics for X9R-MLCC Applications

: 0.5BaTiO 3 –(0.5 − x )BiMg 1/2 Ti 1/2 O 3 – x NaNbO 3 ( x = 0.10–0.30) ceramics were processed via a conventional solid state sintering route. X-ray diffraction analysis and Raman spectroscopy showed the formation of a cubic perovskite structure. Microstructural analysis of the samples revealed densely packed grains. The addition of NaNbO 3 resulted in the enhancement in dielectric properties as a function of temperature. Relative permittivity decreased from 850 to 564 (at room temperature) with an increase in x ; however, the stability in dielectric properties was improved with an increase in NaNbO 3 concentration. At x = 0.25, relative permittivity ( ε r ) was ~630 ± 15% in a temperature range of − 70–220 ◦ C with low dielectric loss (tan δ ) < 0.025 ( − 57 to 350 ◦ C) and high recoverable energy density ~0.55 J/cm 3 which meet the criterion for X9R MLCC applications.


Introduction
The electronic industry is growing rapidly and demands new materials with improved performance for various applications. Among the electronic components, the capacitor is one of the most widely used components for both low temperature, as well as harsh environment applications [1][2][3][4]. The present market of ceramic capacitors is dominated by the multi-layer ceramic capacitor (MLCC), having the advantages of both high volumetric efficiency and small size [5][6][7]. Trillions of pieces are fabricated every year which makes it one of the most widely used components used in electronic circuits [8]. The major characteristics required for capacitor applications are temperature stable high relative permittivity (ε r ), high breakdown strength (BDS) and low dielectric loss (tan δ) [9]. The Electrical industries association designated the upper and lower working temperature limit for the ceramic capacitor as X7R, X8R and X9R, where 'X' represents the lower working temperature limit which is −55 • C, the mid digit stands for 125, 150 and 200 • C temperature for 7, 8 and 9, respectively and 'R' represents minimum variation in the capacitance value (i.e., ±15%) [10,11]. On the other hand, for high temperature electronics (HTE),

Materials and Methods
0.5BaTiO 3 -(0.5 − x)BiMg 1/2 Ti 1/2 O 3 -xNaNbO 3 (BT-BMT-NN) ceramics with x = 0.10-0.30 were processed through a solid-state sintering route. Reagent grade (purity > 99%, Sigma Aldrich) raw chemicals BaCO 3 , Bi 2 O 3 , TiO 2 , MgO, Na 2 CO 3 and Nb 2 O 5 were dried to remove moisture and hydroxides and then weighted according to the molar ratios of batches. Powders were mixed/milled using a planetary ball mill (Fritsch, pulverisette 7600 rpm, Germany) in isopropanol for 6 h. The mixed powders were calcined at 900 • C for 4 h and then re-milled to dissociate agglomerates. The calcined powders were pressed into cylindrical shape pellets using a 10 mm die, at a pressure of 100 MPa and sintered in the temperature range 1050-1125 • C for 2 h.
The density of sintered samples was measured by the Archimedes principle. Phase analysis was carried out at room temperature using a PANalytical X'pert Pro X-ray diffractometer (United Kingdom), using CuKα radiations. Raman spectra of the samples were collected at room temperature using a Renishaw In Via Reflex microspectrometer (United Kingdom), using a 514 nm Ar laser at a power between 30 and 300 mW. The samples were thermally etched at a temperature 10% lower than the sintering temperature (990 • C for 15 min). The microstructure of the polished and thermally etched samples was analyzed using a JEOL (JSM-6460LV) scanning electron microscope (Japan). For electrical measurements, pellets were coated with silver on both sides. Dielectric properties as a function of temperature were measured using an Agilent 4284A LCR meter (United States). Capacitance (C p ) and tan δ were measured at 1 kHz, 10 kHz, 100 kHz and 1 MHz in the temperature range from −70 to 500 • C. Low-temperature data were collected in liquid nitrogen using a homemade system. Polarization-electric field (P-E) loops were measured at a frequency of 10 Hz, using a modified Sawyer-Tower circuit at room temperature.

Results
X-ray diffraction (XRD) patterns of BT-BMT-NN, sintered at 1100 • C for 2 h are shown in Figure 1a. The diffraction patterns matched JCPDS # 131-0174, having a cubic perovskite structure and the patterns were indexed accordingly. No evidence of secondary peaks was observed within the detection limit of the in-house XRD facility, suggesting that Na and Nb cations were completely soluble in the BT-BMT solid solution. The peaks slightly shifted to high 2θ values (higher d-spacings), indicating a decrease in unit cell volume, which may be due to the replacement of slightly larger Bi ions (r XII = 1.38 Å) by Na ions (r XII = 1.34 Å) at the A-site, and Mg ions (r VI = 0.72 Å) by Nb ions (r VI = 0.64 Å) at the B-site of the host lattice [45]. Figure 1b shows the enlarged view of (200) reflection which shows no splitting, indicating the cubic-like structure.

Results
X-ray diffraction (XRD) patterns of BT-BMT-NN, sintered at 1100 °C for 2 h are shown in Figure 1a. The diffraction patterns matched JCPDS # 131-0174, having a cubic perovskite structure and the patterns were indexed accordingly. No evidence of secondary peaks was observed within the detection limit of the in-house XRD facility, suggesting that Na and Nb cations were completely soluble in the BT-BMT solid solution. The peaks slightly shifted to high 2θ values (higher d-spacings), indicating a decrease in unit cell volume, which may be due to the replacement of slightly larger Bi ions (rXII = 1.38 Å ) by Na ions (rXII = 1.34 Å ) at the A-site, and Mg ions (rVI = 0.72 Å ) by Nb ions (rVI = 0.64 Å ) at the B-site of the host lattice [45]. Figure 1b shows the enlarged view of (200) reflection which shows no splitting, indicating the cubic-like structure. For a better understanding of the local structure, room temperature Raman spectra of BT-BMT-NN ceramics were recorded as shown in Figure 2, which is consistent with the data previously reported for BT-BMT [34]. The Raman bands were overlapped and became broader with the increase of NN content, indicating the disorder in the lattice induced by the multiple ions at the same site [46]. For a better illustration, the Raman spectrum was fitted using a simple Lorentzian function, as shown in Figure 2. The first sharp peak at 117 cm −1 and the second band near 180 cm −1 was related to vibrations of A-site cations and displacement [47]. The band near 180 cm −1 shifted to lower wavenumber, probably due to the incorporation of Na + for Bi 3+ with different ionic sizes [45]. In pure BT, a sharp peak appears near 305 cm −1 along with a dip near 180 cm −1 which is indicative of long-range ferroelectric order. In the present case, a relatively broader band at 335 cm −1 and another band near 281 cm −1 appeared (polar BO6 octahedral vibrations), and starts merging with increasing x-value [48]. This behavior may be associated with the destruction of ferroelectric order and the broadening may be related to the formation of polarnano regions due to multiple cations at the same site of the lattice. The Raman bands in the range 400 to 650 cm −1 are often related to the vibrations of oxygen octahedrons [46]. In the present case, two different bands at 497 and 575 cm −1 were observed which merged with an increase in x. The possible reason for this behavior may be the stretching symmetric vibrations of TiO6 and MgO6 octahedra because ionic radii difference is large between Ti 4+ and Mg 2+ . The substitution of Nb 5+ resulted in a broad band which may be due to decreasing amount of Mg 2+ because the ionic radii difference between Nb 5+ and Ti 4+ is small. The modes at 724 and 773 cm −1 merged to form a broad band. This band is known as A1g mode which is associated with the breathing of BO6 octahedra [49]. A1g mode is symmetric and Raman active in A-site doped BT but the splitting comes from the different For a better understanding of the local structure, room temperature Raman spectra of BT-BMT-NN ceramics were recorded as shown in Figure 2, which is consistent with the data previously reported for BT-BMT [34]. The Raman bands were overlapped and became broader with the increase of NN content, indicating the disorder in the lattice induced by the multiple ions at the same site [46]. For a better illustration, the Raman spectrum was fitted using a simple Lorentzian function, as shown in Figure 2. The first sharp peak at 117 cm −1 and the second band near 180 cm −1 was related to vibrations of A-site cations and displacement [47]. The band near 180 cm −1 shifted to lower wavenumber, probably due to the incorporation of Na + for Bi 3+ with different ionic sizes [45]. In pure BT, a sharp peak appears near 305 cm −1 along with a dip near 180 cm −1 which is indicative of long-range ferroelectric order. In the present case, a relatively broader band at 335 cm −1 and another band near 281 cm −1 appeared (polar BO 6 octahedral vibrations), and starts merging with increasing x-value [48]. This behavior may be associated with the destruction of ferroelectric order and the broadening may be related to the formation of polar-nano regions due to multiple cations at the same site of the lattice. The Raman bands in the range 400 to 650 cm −1 are often related to the vibrations of oxygen octahedrons [46]. In the present case, two different bands at 497 and 575 cm −1 were observed which merged with an increase in x. The possible reason for this behavior may be the stretching symmetric vibrations of TiO 6 and MgO 6 octahedra because ionic radii difference is large between Ti 4+ and Mg 2+ . The substitution of Nb 5+ resulted in a broad band which may be due to decreasing amount of Mg 2+ because the ionic radii difference between Nb 5+ and Ti 4+ is small. The modes at 724 and 773 cm −1 merged to form a broad band. This band is known as A1g mode which is associated with the breathing of BO 6 octahedra [49]. A1g mode is symmetric and Raman active in A-site doped BT but the splitting comes from the different octahedra because the frequency of this mode changes with change in ionic radius which creates asymmetry [50]. octahedra because the frequency of this mode changes with change in ionic radius which creates asymmetry [50]. The samples were sintered in the temperature range 1050-1125 °C for 2 h (Figure 3a). For all samples, the bulk density increased with an increase in temperature from 1050 °C . For a sample with x = 0.10, maximum bulk density was observed at 1075 °C which decreased with a further increase in temperature. For samples with x > 0.10, the highest bulk density was observed at a sintering temperature of 1100 °C . The decrease in bulk density above optimal sintering temperature may be due to the volatile nature of bismuth or abnormal grain growth. However, for a better comparison, the samples sintered at 1100 °C (optimal sintering temperature) were selected for investigation because bismuth is volatile and a slight temperature change may affect the properties of the sintered ceramics. Figure  3b shows a variation in bulk density versus x (NaNbO3 concentration). The bulk density decreased with an increase in x which may be due to the decreasing amount of bismuth as the atomic weight of bismuth is higher than sodium.   Figure S1), which is, technologically, of great importance for the fabrication of MLCCs [51]. The average grain size and relative density of all the samples sintered at 1100 °C are given in Table 1. For a sample with x = 0.25, both the relative density and grain size were larger.  The samples were sintered in the temperature range 1050-1125 • C for 2 h (Figure 3a). For all samples, the bulk density increased with an increase in temperature from 1050 • C. For a sample with x = 0.10, maximum bulk density was observed at 1075 • C which decreased with a further increase in temperature. For samples with x > 0.10, the highest bulk density was observed at a sintering temperature of 1100 • C. The decrease in bulk density above optimal sintering temperature may be due to the volatile nature of bismuth or abnormal grain growth. However, for a better comparison, the samples sintered at 1100 • C (optimal sintering temperature) were selected for investigation because bismuth is volatile and a slight temperature change may affect the properties of the sintered ceramics. Figure 3b shows a variation in bulk density versus x (NaNbO 3 concentration). The bulk density decreased with an increase in x which may be due to the decreasing amount of bismuth as the atomic weight of bismuth is higher than sodium.
Crystals 2022, 12, 141 4 of 10 octahedra because the frequency of this mode changes with change in ionic radius which creates asymmetry [50]. The samples were sintered in the temperature range 1050-1125 °C for 2 h (Figure 3a). For all samples, the bulk density increased with an increase in temperature from 1050 °C . For a sample with x = 0.10, maximum bulk density was observed at 1075 °C which decreased with a further increase in temperature. For samples with x > 0.10, the highest bulk density was observed at a sintering temperature of 1100 °C . The decrease in bulk density above optimal sintering temperature may be due to the volatile nature of bismuth or abnormal grain growth. However, for a better comparison, the samples sintered at 1100 °C (optimal sintering temperature) were selected for investigation because bismuth is volatile and a slight temperature change may affect the properties of the sintered ceramics. Figure  3b shows a variation in bulk density versus x (NaNbO3 concentration). The bulk density decreased with an increase in x which may be due to the decreasing amount of bismuth as the atomic weight of bismuth is higher than sodium.   Figure S1), which is, technologically, of great importance for the fabrication of MLCCs [51]. The average grain size and relative density of all the samples sintered at 1100 °C are given in Table 1. For a sample with x = 0.25, both the relative density and grain size were larger.    Figure S1), which is, technologically, of great importance for the fabrication of MLCCs [51]. The average grain size and relative density of all the samples sintered at 1100 • C are given in Table 1. For a sample with x = 0.25, both the relative density and grain size were larger.  The εr and tanδ as a function of temperature for BT-BMT-NN ceramics measured at different frequencies from 1 kHz-1 MHz in a temperature range of −70 to 500 °C is shown in Figure 5a-e. The temperature of maximum εr (Tm) drastically decreased from 91 to −40 °C , with an increase in x from 0.10 to 0.30. A similar effect of decreasing Tm was reported for NaNbO3-modified BaTiO3-Bi(Zn0.5Ti0.5)O3 solid solution [52]. εr linearly decreased with an increase in Na + and Nb 5+ concentration which encouraged the short-range ferroelectric behavior. A similar trend was observed for (1 − x)NaNbO3−xBaTiO3 ceramics [53]. Another reason for the decrease in εr may be the smaller polarizability of Nb 5+ than Ti 4+ . As evident from the P-E loops and Raman data, the crystal structure is cubic but still, εr is higher than centrosymmetric structures, such as CaTiO3 which may be attributed to the formation of polar nanoregions (PNRs) due to the occupancy of multiple cations at the same site [31].   The ε r and tan δ as a function of temperature for BT-BMT-NN ceramics measured at different frequencies from 1 kHz-1 MHz in a temperature range of −70 to 500 • C is shown in Figure 5a-e. The temperature of maximum ε r (T m ) drastically decreased from 91 to −40 • C, with an increase in x from 0.10 to 0.30. A similar effect of decreasing T m was reported for NaNbO 3 -modified BaTiO 3 -Bi(Zn 0.5 Ti 0.5 )O 3 solid solution [52]. ε r linearly decreased with an increase in Na + and Nb 5+ concentration which encouraged the shortrange ferroelectric behavior. A similar trend was observed for (1 − x)NaNbO 3 -xBaTiO 3 ceramics [53]. Another reason for the decrease in ε r may be the smaller polarizability of Nb 5+ than Ti 4+ . As evident from the P-E loops and Raman data, the crystal structure is cubic but still, ε r is higher than centrosymmetric structures, such as CaTiO 3 which may be attributed to the formation of polar nanoregions (PNRs) due to the occupancy of multiple cations at the same site [31].   The εr and tanδ as a function of temperature for BT-BMT-NN ceramics measured at different frequencies from 1 kHz-1 MHz in a temperature range of −70 to 500 °C is shown in Figure 5a-e. The temperature of maximum εr (Tm) drastically decreased from 91 to −40 °C , with an increase in x from 0.10 to 0.30. A similar effect of decreasing Tm was reported for NaNbO3-modified BaTiO3-Bi(Zn0.5Ti0.5)O3 solid solution [52]. εr linearly decreased with an increase in Na + and Nb 5+ concentration which encouraged the short-range ferroelectric behavior. A similar trend was observed for (1 − x)NaNbO3−xBaTiO3 ceramics [53]. Another reason for the decrease in εr may be the smaller polarizability of Nb 5+ than Ti 4+ . As evident from the P-E loops and Raman data, the crystal structure is cubic but still, εr is higher than centrosymmetric structures, such as CaTiO3 which may be attributed to the formation of polar nanoregions (PNRs) due to the occupancy of multiple cations at the same site [31].  T m shifted to a lower temperature and the thermal stability of E r was enhanced with the increase of NN content. Dielectric properties of BT-BMT-NN ceramics are listed in Table 2 while the variation in ε r as a function of temperature is shown in Figure 5f. The sample with x = 0.10 possesses E r = 850 ± 15% across −8 to 450 • C and tan δ < 0.025 (25-412 • C). Upon further increase in x, lower temperature stability was enhanced below room temperature, but high temperature stability degraded. An optimum set of dielectric properties were achieved for x = 0.25, i.e., E r = 630 ± 15% stable over the temperature range −70 • C to 220 • C, and the dielectric loss was <0.25 over the operating temperature range −57-350 • C which satisfy the requirements of the X9R type capacitor. Table 2. Dielectric properties of BT-BMT-NN ceramics. A dispersion below T m was observed which is indicative of the "relaxor behavior". The relaxor behavior of ferroelectric materials can be effectively described with the help of modified Curie-Weiss law [56,57].
Here, 'ε m ' represents maximum ε r , 'γ' and 'C' are constants. The value γ varies from 1-2 for normal to ideal relaxor ferroelectrics. γ is obtained from the slope of log(1/E − 1/E m ) versus log(T − T m ) as plotted in Figure 6. The value γ ranges from 1.32 to 1.54, indicating relaxor-like behavior. It has been reported [48,58] that cation disorder at the A-and/or B-site is responsible for relaxor behavior, in agreement with the Raman studies ( Figure 2). Tm shifted to a lower temperature and the thermal stability of ɛr was enhanced with the increase of NN content. Dielectric properties of BT-BMT-NN ceramics are listed in Table 2 while the variation in εr as a function of temperature is shown in Figure 5f. The sample with x = 0.10 possesses ɛr = 850 ± 15% across −8 to 450 °C and tanδ < 0.025 (25-412 °C ). Upon further increase in x, lower temperature stability was enhanced below room temperature, but high temperature stability degraded. An optimum set of dielectric properties were achieved for x = 0.25, i.e., ɛr = 630 ± 15% stable over the temperature range −70 °C to 220 °C , and the dielectric loss was <0.25 over the operating temperature range −57-350 °C which satisfy the requirements of the X9R type capacitor. A dispersion below Tm was observed which is indicative of the "relaxor behavior". The relaxor behavior of ferroelectric materials can be effectively described with the help of modified Curie-Weiss law [56,57].
Here, 'εm' represents maximum εr, 'γ' and 'C' are constants. The value γ varies from 1-2 for normal to ideal relaxor ferroelectrics. γ is obtained from the slope of log(1/ɛ − 1/ɛm) versus log(T − Tm) as plotted in Figure 6. The value γ ranges from 1.32 to 1.54, indicating relaxor-like behavior. It has been reported [48,58] that cation disorder at the A-and/or Bsite is responsible for relaxor behavior, in agreement with the Raman studies ( Figure 2).   Figure 7 shows P-E loops for BT-BMT-NN samples measured at 75 kV/cm. The samples with x = 0.30 exhibited the lowest maximum polarization, P m = 3.43 µC/cm 2 while the maximum P m value of 6.4 µC/cm 2 was obtained for samples with x = 0.10. Similar behavior was observed in Na + and Nb 5+ doped Bi 0.5 Na 0.5 TiO 3 -BT ceramics [59]. The decrease in polarization may be due to the decrease in polarizability of the constituents. However, a non-linear trend was observed in P m values for samples with x = 0.2 and 0.25. For both these two samples, an opening in the P-E loops was also observed which shows a weakly nonlinear dielectric behavior. The same phenomenon was observed for (1 − x)BT-xBMT (x = 0.4) ceramics which was attributed to the increase in the conductive tan δ at room temperature [33]. For BT-BMT-NN with x = 0.20 and 0.25, tan δ is very low (<0.025) at room temperature; therefore, the origin for the observed weak non-linear behavior needs further investigation. The recoverable energy density (W rec ) calculated for the sample with x = 0.10 was 0.5 J/cm 3 at an applied electric field of 110 kV/cm. A maximum W rec of 0.55 J/cm 3 and an efficiency of 82% were observed for the sample with x = 0.25 at 150 kV/cm.  Figure 7 shows P-E loops for BT-BMT-NN samples measured at 75 kV/cm. The samples with x = 0.30 exhibited the lowest maximum polarization, Pm = 3.43 μC/cm 2 while the maximum Pm value of 6.4 μC/cm 2 was obtained for samples with x = 0.10. Similar behavior was observed in Na + and Nb 5+ doped Bi0.5Na0.5TiO3-BT ceramics [59]. The decrease in polarization may be due to the decrease in polarizability of the constituents. However, a nonlinear trend was observed in Pm values for samples with x = 0.2 and 0.25. For both these two samples, an opening in the P-E loops was also observed which shows a weakly nonlinear dielectric behavior. The same phenomenon was observed for (1-x)BT-xBMT (x = 0.4) ceramics which was attributed to the increase in the conductive tanδ at room temperature [33]. For BT-BMT-NN with x = 0.20 and 0.25, tanδ is very low (< 0.025) at room temperature; therefore, the origin for the observed weak non-linear behavior needs further investigation. The recoverable energy density (Wrec) calculated for the sample with x = 0.10 was 0.5 J/cm 3 at an applied electric field of 110 kV/cm. A maximum Wrec of 0.55 J/cm 3 and an efficiency of 82% were observed for the sample with x = 0.25 at 150 kV/cm.

Conclusions
0.5BaTiO3-(0.5 − x)BiMg1/2Ti1/2O3−xNaNbO3 (x = 0.10-0.30) lead-free ceramics were prepared which form a single phase cubic perovskite structure, indicating that Na + and Nb 5+ are soluble in the host lattice. Although the samples have a paraelectric phase, the high ɛr was due to the formation of nano-polar regions, confirmed from the Raman spectra of the samples. Microstructural analysis shows that dense ceramics of the solid solution could be easily prepared by a solid state route. The samples exhibited a flat temperature dependent response. Optimum dielectric properties were observed for the sample with x = 0.25 i.e., ɛr = 630 ± 15% over the temperature range from −70-220 °C and tanδ < 2.5% over −57 to 350 °C , suggesting that it as a promising candidate material for the X9R capacitor.

Conclusions
0.5BaTiO 3 -(0.5 − x)BiMg 1/2 Ti 1/2 O 3 -xNaNbO 3 (x = 0.10-0.30) lead-free ceramics were prepared which form a single phase cubic perovskite structure, indicating that Na + and Nb 5+ are soluble in the host lattice. Although the samples have a paraelectric phase, the high E r was due to the formation of nano-polar regions, confirmed from the Raman spectra of the samples. Microstructural analysis shows that dense ceramics of the solid solution could be easily prepared by a solid state route. The samples exhibited a flat temperature dependent response. Optimum dielectric properties were observed for the sample with x = 0.25 i.e., E r = 630 ± 15% over the temperature range from −70-220 • C and tan δ < 2.5% over −57 to 350 • C, suggesting that it as a promising candidate material for the X9R capacitor.