Fabrication of 〈100〉-oriented Bi(Mg_0.5Ti_0.5)O₃-modified BiFeO₃-BaTiO₃ ceramics and their piezoelectric properties

The chemical compositions, 0.90{0.33BaTiO₃-0.67BiFeO₃}0.10Bi(Mg_0.5Ti_0.5)O₃ (33BTBF-10BMT), were prepared using BaCO₃, Fe₂O₃, Bi₂O₃, MgO, and TiO₂ powders. The starting powders were first mixed through ball milling for 24 h and then calcined at 800 °C for 6 h. Next, the calcined powder and 0.1 wt% of MnO₂ were performed through second ball milling to reduce leakage current and create a large size difference between the matrix powder and templates, aiding the TGG process. ^3,25) The average size of the matrix powder was less than 500 nm.

The BaTiO₃ (BT) templates were synthesized using the three-step molten salt method. A bismuth-based layered structure of Bi₄Ti₃O₁₂ (BiT) precursor, which formed an aurivillius-type layered structure, was prepared through the molten salt method. Initially, Bi₂O₃, TiO₂, NaCl, and KCl were used as raw powders and mixed by hand milling. The mixture was then heated at 1100 °C for 1 h, and the resulting product was washed several times with hot deionized water. In the second step, the synthesized BiT precursor was mixed with BaCO₃, TiO₂ powders in a molar ratio of 1:1.1:1.1 by magnetic stirring. An equal amount of NaCl and KCl was added, and the mixture was heated at 1080 °C for 1 h. The powder was washed with hot deionized water several times to remove the remaining flux, resulting in the formation of BaBi₄Ti₄O₁₅ (BBiT). Lastly, the synthesized BBiT precursor was mixed with BaCO₃ powder in a molar ratio of 1:4 by magnetic stirring. Equal amounts of NaCl and KCl were also added, and the mixture was heated at 950 °C for 3 h. The resulting powder was washed several times with hot deionized water and then with dilute HNO₃ to obtain high-quality BT templates without remaining Bi₂O₃. The average size of the BT templates was a length of 10–20 μm and a thickness of 0.5 μm.

For the tape-casting process, the slurry was prepared by mixture with the calcined 33BTBF-10BMT with 0.1 wt% of MnO₂ matrix powder, 7 vol% BT templates, binder, and solvent. The mixing process was conducted by ball milling for 20 h at a speed of 40 rpm. The slurry was tape casted with a doctor-blade thickness of 200 μm. After drying, the green sheets were cut as square sheets of 20 × 20 mm², stacked, and uniaxial pressed under 60 MPa at 80 °C for 10 min. The green compact after pressing was cut into a size of 6 × 6 mm², and then the binder was burnt out at 600 °C 5 h. At last, the textured ceramics were obtained by sintering process at 1000 °C–1100 °C for 4 h. The nontextured ceramics were sintered at 1000 °C for 4 h using a conventional solid-state synthesis. ²⁾ Relative density was measured by the Archimedes method.

The crystal structure was investigated using an X-ray diffractometer (XRD; Ultima IV, Rigaku) with Cu Kα radiation. For microstructure observation through scanning electron microscopy (SEM; JSM-6510, JEOL), the ceramics with cross-sectional areas were mirror polished and chemically etched using a 1 N HCl solution for 10 min. The average grain size for the textured ceramics was estimated using the linear intercept method. The degree of orientation was determined by the Lotgering method, and the degree of orientation along the 〈100〉 direction, F₁₀₀, was calculated by the following formula:

$$\begin{eqnarray*}&&{F}_{h00}=\displaystyle \frac{P-{P}_{0}}{1-{P}_{0}},P=\displaystyle \frac{\displaystyle \sum {I}_{\left(h00\right)}}{\displaystyle \sum {I}_{\left(hkl\right)}},{P}_{0}=\displaystyle \frac{\displaystyle \sum {I}_{0\left(h00\right)}}{\displaystyle \sum {I}_{\left(0hkl\right)}},\end{eqnarray*}$$

where I and I₀ represent the peak intensities of textured and nontextured ceramics, respectively.

The sintered pellets were polished and cut into dimensions of 2 mm (length) × 2 mm (width) × 0.4 mm³ (thickness). The prepared samples were annealed at 800 °C for 1 h and then quenched in water. The continuous thermal annealing and water-quenching processes enhanced the ferroelectric and piezoelectric properties by recovering surface-damaged layers and domain wall depinning. ^26–28) Gold electrodes were coated on the sample surfaces and heated at 300 °C for 10 min. The temperature dependence of the dielectric constant and loss was measured using an LCR analyzer (6440B, Wayne Kerr Electronics) in the temperature range of 20 °C–500 °C. Bipolar polarization­ electric field loops and unipolar strain­ electric (S­E) field curves were determined using a ferroelectric and strain measuring system (JP005-SE, Kitamoto Denshi) with a displacement meter (Millitron 1202IC, Mahr). The measurements were conducted at 0.1 Hz at 22 °C.

A nonpolar 〈100〉 direction is a target direction for engineered domain configuration owing to the rhombohedral symmetry of the 33BTBF-10BMT matrix. ^2,5–7) The degree of orientation (F₁₀₀) of 33BTBF-10BMT textured ceramics with 7 vol% BT templates was examined as a function of sintering temperature (T_s). Since BFBT-based solid solutions typically achieve dense ceramics when sintered around 1000 °C, the evaluation of F₁₀₀ began from a T_s of 1000 °C, as shown in Fig. 1. Interestingly, textured 33BTBF-10BMT ceramics using BT templates exhibited the orientation along the 〈100〉 direction even at 1000 °C, with an F₁₀₀ of 44%. The similar lattice parameter between the BT template and the 33BTBF-10BMT matrix may contribute to a less lattice mismatch between them, resulting in epitaxial growth of oriented grains by a coherent interface between the template and matrix. ^21,23,24) Similar F₁₀₀ results were achieved at 1020 °C and 1040 °C. Beyond 1060 °C, higher F₁₀₀ results were attained, reaching a maximum orientation of 80% at 1080 °C. However, the F₁₀₀ gradually decreased after 1080 °C, and the degree of orientation at 1100 °C was similar to that at 1060 °C. To comprehend these results, the microstructural characteristics of the samples within the T_s range from 1000 °C–1100 °C were carried out, as shown in Fig. 2. The microstructure profiles were obtained using SEM from the cross-sectional surfaces of the samples, which were prepared through a mirror-polishing and a chemical etching process. From 1000 °C–1040 °C, small grains of approximately 2 μm and the presence of templates were clearly observed. On the other hand, at 1060 °C, templates inside brick-like grains and larger grains around 5 μm were observed. It should be noted that some templates were not properly aligned along the casting direction, indicating the need for further optimization by processing techniques such as BT template contents, doctor-blade thickness, casting speeds, and others. ^24,29) At 1080 °C, even larger grains of approximately 10 μm were observed, and a higher level of grain homogeneity was observed compared to T_s below 1060 °C. The different F₁₀₀ results obtained from XRD, comparing relatively low T_s (1000 °C, 1020 °C, and 1040 °C) with high T_s (over 1060 °C), may be closely related to the volume of large oriented grains, as shown in Fig. 3. It is known that the growth of oriented grains contributes to an enhanced degree of orientation. ²¹⁾ This result indicates that larger oriented grains at relatively high T_s are important for achieving a high degree of orientation. At 1100 °C, however, the lower F₁₀₀ appears to be correlated with the low relative density, although large oriented grains were observed around 17 μm. The distinct phenomenon such as round-shaped grains and large pores were observed, suggesting decomposition caused by excessively high T_s for the matrix composition. It should be noted here that additional analysis, such as electron backscattered diffraction, is required to confirm the exact volume of oriented grains across a wide range of samples. The tendencies of F₁₀₀, relative density, and grain size as a function of T_s are shown in Fig. 3. The relative density exhibited above 90% from 1000 °C–1060 °C, 90% at 1080 °C, and 80% at 1100 °C. The relative density started to decrease from 1080 °C, while large oriented grains increased from 1060 °C. The result of the degree of orientation as a function of T_s indicates a trade-off relationship between the volume of large oriented grains and relative density. The low relative density at high T_s of 1100 °C may be related to the presence of large pores due to the decomposition issue in the matrix. This result is consistent with the expectation that it may be due to the large gap of T_s compared to typical nontextured BFBT-based piezoelectric ceramics, which are commonly sintered at around 1000 °C. Thus, selecting an optimized T_s is critical for achieving a significant number of oriented grains without pores caused by decomposition.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

It should be noted that SEM images revealed the presence of non-reacted BT templates after the sintering process. This indicates that BT templates do not react with the 33BTBF-10BMT matrix, particularly similar to the case of Pb-based textured ceramics produced by BT or SrTiO₃ (ST) templates. ²¹⁾ The diffusion issue and interaction between templates and matrix are important factors in the TGG process. Thus, additional analysis, such as energy dispersive spectroscopy (EDS), is required to confirm the diffusion issue across a wide range of samples.

A T_s of 1060 °C and 1080 °C represent a promising candidate for achieving high degrees of orientation (67% and 80%, respectively) as well as dense ceramics (relative densities over 90%), as supported by the Lotgering factor and microstructure analysis. Thus, two textured ceramics were compared to their nontextured counterpart. It should be noted that the results of other textured ceramics with different T_s can be seen in Table I. Figure 4 and Table I show the temperature dependence of the dielectric constant and loss of nontextured and textured ceramics at 1 kHz. Dielectric constant peaks were observed at 425 °C for nontextured ceramics, while relatively low dielectric constant peaks were observed for both textured ceramics at T_s of 1060 °C and 1080 °C. Furthermore, the dielectric constant peak significantly declined at higher T_s, especially at 1100 °C, as shown in Table I. The decrease in the maximum temperature (T_max) might be associated with two reasons in this research. One is the decreased relative density when increasing the T_s. The other reason is the partial diffusion of Ba ions into the interface between the BT template and matrix 33BTBF-10BMT regions at higher T_s. On the other hand, the dielectric constant peak was not significantly changed up to a T_s of 1040 °C as shown in Fig. 4. This means that BT templates may be related to less chemical reaction with the 33BTBF-10BMT matrix at relatively low T_s. Once again, the analysis of EDS is required to investigate the diffusion issue. The dielectric constant of textured ceramics shows a high dielectric constant compared to their nontextured counterparts at 23 °C with 1 kHz, except for textured ceramics sintered at 1100 °C due to the low relative density. The increased dielectric constant in textured ceramics might be related to the presence of BT templates, owing to their larger dielectric constant compared to any BMT-modified BFBT composition. ³⁰⁾ The dielectric loss (loss tangent) for all nontextured and textured ceramics does not drastically increase until 150 °C, while a dramatic increase in large dielectric loss at high temperature may be associated with the relaxation of defect dipoles. ^2,3)

Zoom In Zoom Out Reset image size

Figure 5 shows the bipolar P–E hysteresis loops of the nontextured and textured ceramics at 22 °C with 0.1 Hz. All measurements were conducted after the poling cycles. The remanent polarization (P_r) of textured ceramics with 80% orientation is 24.1 μC cm⁻², which is clearly lower than that of nontextured ceramics (31.2 μC cm⁻²). The decrease of P_r in textured ceramics might be attributed to the grain orientation along the nonpolar direction and the presence of BT templates. For the former case, a gradual decrease of P_r is due to the increase of orientation degree along the nonpolar direction. For rhombohedral crystals, a lower P_r is observed when an electric field is applied along the 〈001〉 direction due to the 〈111〉 spontaneous polarization direction of the perovskite rhombohedral crystal. ¹³⁾ Most of the grains are aligned along the nonpolar 〈100〉 direction through the grain orientation process in the rhombohedral BMT-modified BFBT system. Hence, P_r gradually decreases when the degree of orientation is increased, as shown in Fig. 5 and Table I. For the latter case, the presence of non-reacted BT templates acts as a composite in the BMT-modified BFBT system, resulting in relatively low P_r in the textured ceramics. The lower P_r in the textured ceramics was attributed to the lower P_r and coercive field (E_c) of the BT ceramics. ³⁰⁾ It should be noted that the textured ceramics sintered at 1100 °C show the lowest P_r despite the degree of orientation not being maximum, which may be related to the lowest relative density.

Zoom In Zoom Out Reset image size

The unipolar S–E curves for the nontextured and textured ceramics are shown in Fig. 6. Enhanced maximum strain (S_max) was observed in textured ceramics compared to nontextured ceramics. The highest S_max is observed in textured ceramics with 80% orientation, measuring 0.29%, while S_max of nontextured ceramics is 0.19%. The d₃₃ ^* (calculated by S_max/E_max) exhibits trends analogous to the results of S_max. The highest d₃₃ ^* (410 pm V⁻¹) was achieved by textured ceramics with 80% orientation, while d₃₃ ^* for nontextured ceramics is 270 pm V⁻¹. Furthermore, the textured ceramics with 67% orientation achieved a comparable value of d₃₃ ^* (405 pm V⁻¹) to textured ceramics with 80% orientation. These values represent an approximately 1.5 times higher value compared to nontextured ceramics. A similar enhancement of piezoelectric strain response has been observed in other Pb-free piezoelectric textured ceramics. BT-based textured ceramics, for instance, show approximately 1.4 times higher d₃₃ ^* than their nontextured ceramics. ³¹⁾ Over d₃₃ ^* of 600 pm V⁻¹ has been attained, but its application is limited due to the lower Curie temperature, around 90 °C. Similar tendencies have been observed in other Pb-free (K,Na)NbO₃ (KNN) and (Bi_0.5Na_0.5)TiO₃ (BNT)-based piezoelectric textured ceramics, which show high d₃₃ ^* but lower Curie temperature, typically below 300 °C. ^32,33) The textured ceramics in this study sintered up to 1060 °C exhibit a T_max of 400 °C with d₃₃ ^* of 300 pm V⁻¹. The only textured ceramic with 67% orientation shows a d₃₃ ^* of over 400 pm V⁻¹ and a Curie temperature of 400 °C, as shown in Fig. 7. The highest d₃₃ ^* (410 pm V⁻¹) was achieved by textured ceramics with 80% orientation, but it shows a T_max of 358 °C. Despite the relatively low T_max in textured ceramics with 80% orientation, it is still higher than other Pb-free piezoelectric ceramics. ^31–34) It should be mentioned here that the significant enhancement of piezoelectric strain response in the textured ceramics is attributed to the high orientation factor rather than the presence of BT templates. The decrement of positive strain in BT/BMT-modified BFBT composite was reported. ³⁰⁾ In unipolar S–E curves, open hysteresis can be observed in textured ceramics. The strain hysteresis of nontextured ceramics is 10%, while it shows 15% for most textured ceramics, as shown in Table I and Fig. 6. Ideally, hysteresis-free strain should be expected in highly 〈100〉-oriented rhombohedral perovskite ceramics. In [001]-poled rhombohedral crystals, there are only four ferroelectric domains along the [111], $[1\bar{1}1],$ $[\bar{1}11],$ $[\bar{1}\bar{1}1]$ directions since 〈111〉 is the polar direction, as observed in PZN, PZN-PT, and PMN-PT single crystals. ¹³⁾ Thus, hysteresis-free strain can be easily expected due to the absence of domain wall movement caused by energetically equivalent domains under the electric field along the [001] direction. In the case of 〈100〉-oriented BMT-modified BFBT piezoelectric ceramics, however, the oriented grains are not exactly aligned along the nonpolar direction, resulting in open hysteresis. Another plausible reason for open hysteresis in textured ceramics along the nonpolar direction might be associated with internal stresses from the presented BT templates embedded within the oriented grains. Internal local stresses of up to 8 MPa were observed in the vicinity between ST templates and the PMN-PT matrix, and this may be related to the poor stability of the domain state due to the possibility of ferroelastic switching of domains. ^21,35) Hence, this could account for open hysteresis observed in textured ceramics even along nonpolar directions. Alternatively, it could also be due to the low relative density in this research caused by large porosity at higher T_s. Changes in the movement of domain walls can occur due to any of these reasons under the applied electric field and mechanical stresses.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

For further enhancement of the piezoelectric strain response, there are possible strategies for the future, such as exploring other BMT-modified BFBT systems and increasing the degree of orientation and relative density. For the former case, the 40BTBF-BMT systems exhibit the highest d₃₃ ^*, relatively low strain hysteresis, and a comparable T_max compared to other 25, 30, and 33BTBF-BMT systems reported in previous literature. ²⁾ Thus, it is expected that further enhancement of the piezoelectric strain response, low hysteresis, and maintaining a high T_max can be achieved by applying grain orientation along the 〈100〉 direction. On the other hand, for the latter case, achieving over 80% orientation and high relative density simultaneously remains a challenge. The use of sintering aids, longer T_s time, the need for further optimization of BT template contents, doctor-blade thickness, casting speeds, and other processing techniques, are required to promote TGG and densification at relatively lower T_s. ^24,29)