3.2. X-ray Diffraction Studies
As an example, an X-ray diffraction pattern of 67BFBT ceramics sintered at 970 °C is shown in
Figure 2. The search-match process was performed using Match! software (Crystal Impact, Bonn, Germany) [
26]. Phase analysis was performed, and it was found that the reference pattern of Ba
0.3Bi
0.7FeO
2.85 [
27] matches all peaks shown within the measuring range 2
Θ with Figure-of-Merit (
FoM) parameter
FoM = 0.96.
It is worth noting that the reference pattern (source of entry: Crystallography Open Database; COD ID 4341652) exhibited a tetragonal structure (space group P4mm) with the following unit cell parameters: a = 3.9963 Å and c = 4.0032 Å. Based on Archimedes’ method and calculations of theoretical density, it was found that 67BFBT ceramics reached a relative density of 94.5%.
Line profile analysis was performed, and the resulting Williamson–Hall plot is shown in
Figure 3.
One can see from the Williamson–Hall plot that the average size of the crystallites is <D> = 801 Å. The average strain (<ε>), which is a measure of micro-deformations, is <ε> = 0.01%.
3.3. Data Validation of Impedance Measurements
It is commonly known (e.g., see [
16,
17,
18,
19]) that impedance spectroscopy is extremely susceptible to random disturbances. Therefore, knowing the quality of the measured impedance data is extremely important to facilitate correct analysis. Kramers–Kronig (K-K) relations are very helpful for data validation [
17,
19]. The Kramers–Kronig rule states that the imaginary part of a dispersion is fully determined by the form of the real part of dispersion over the frequency range ∞ ≥
ν ≥0. Similarly, the real part of dispersion is determined by the form of the imaginary part of dispersion [
17]. In the present study, an analysis based on the K-K relations was performed with the use of the computer programme created by Boukamp [
17,
19].
Kramers–Kronig transform test results of impedance data measured for 67BFBT ceramics at room temperature (
RT) within the frequency range Δ
ν = (10
−1–10
5) Hz are shown in
Figure 4a and
Figure 5a. A complex impedance diagram combined with K-K transform test results measured at a signal voltage of
U = 0.1 V
rms is shown in
Figure 4b. An Impedance diagram with K-K transform test results measured at a signal voltage of
U = 2.0 V
rms is shown in
Figure 5b. One can see in
Figure 4a and
Figure 5a that the data recorded at RT exhibit a small deviation from K-K behaviour (residuals are less than 0.4%). Upon inspection of the results given in
Figure 4b and
Figure 5b, it can be determined that there is very good agreement between the measurements (blue circles) and K-K calculations (red crosses).
The quality parameter “chi-squared” reached a value of χ2 = 3.3 × 10−7–5.5 × 10−7 for room temperature measurements taken at U = 0.1 Vrms (E = 166 V/m) and U = 2.0 Vrms (E = 3333 V/m), respectively. The above-mentioned results proved the high quality of the measurements and fully justified further analysis of the impedance data.
An alternative representation of the impedance data showing the influence increasing electric field intensity on the spectroscopic dependence of the reactance (reactance times pulsation: −
Z”
ω) of the piezoelectric ceramic sample 67BFTO at room temperature is shown in
Figure 6a. It can be seen that the spectroscopic plots shift towards higher frequencies as the electric field strength increases. Additionally, the dependence of the real part of complex impedance (
Z’) on the imaginary part of complex impedance times pulsation (−
Z”
ω) (
Figure 6b) shows the substantial dependence on the electric field strength.
As shown in the next paragraph of this paper (
Section 3.4), an increase in the electric field strength causes an increase in the circumferential magnetic field strength. In turn, the dependence of electric (and dielectric) parameters on magnetic field strength (and vice versa) is a key feature of multiferroic materials, especially lead-free piezoelectric ceramics such as 67BFTO.
3.4. Piezoelectric Ceramics Characterisation with the Resonant Method
It is commonly known (e.g., see [
21,
24]) that values of the piezoelectric properties of a material can be derived from the resonance behaviour of suitably shaped specimens subjected to a sinusoidally varying electric field. Therefore, the impedance measurements were performed for 67BFBT ceramics within the following frequency range:
ν = 100 kHz–10 MHz, which corresponded to the frequency ranges of resonant spectra of the radial mode and the thickness extension mode for thin disk sample [
21,
22].
To reveal the influence of a weak circumferential magnetic field on the resonance behaviour of multiferroic lead-free material, the measuring voltage
U = 0.1 V
rms and
U = 2.0 V
rms was taken. Taking into account all the impedances in the circuit, the measuring sinusoidal signal leads to electric currents (
iac) through the sample of 67BFBT ceramics in the range following range:
iac = 1.2 mA (at
Uac = 0.1 V
rms,
ν = 10 MHz)–11.7 mA (at
U = 2.0 V
rms,
ν = 10 MHz). The ac field amplitude (rms) generated by these currents in a radial point
r on the sample can be calculated as:
where
r is the radial point considered on the sample cross-section, and
a is its total radius. The rms range of ac-measuring fields leads to circumferential magnetic fields (i.e., on the sample edges—the lateral surface of the disk-shaped sample—is where it is highest) between 0.05 and 0.49 A/m (at
U = 0.1 V
rms and
U = 2.0 V
rms,
ν = 10 MHz, respectively) [
28]. The results of the calculated radial magnetic field intensity
Hac are given in
Figure 7. It is worth noting that the resonance behaviour of the 67BFBT ceramic sample is also reflected in
Figure 7. Positions of the resonances are closely related to the piezoelectric properties of the material.
Figure 8 shows the frequency response of a 0.67Bi
1.02FeO
3–0.33BaTiO
3 ceramic thin disk that was 7.52 mm in diameter and 0.6 mm thick. Electrodes were deposited onto both faces of the disk, and then the disk was poled in the direction perpendicular to the faces of the disk. The measurements were taken within a temperature range from −20 °C to 50 °C. It should be noted that the frequency peaks visible at about 3–4 × 10
5 Hz (
Figure 8) are radial resonances, whereas the frequency peaks visible at about 3–5 × 10
6 Hz (also in
Figure 8) are related to the thickness mode resonance.
Spectroscopic plots of modulus of complex impedance |
Z| exhibit the characteristic frequencies, namely
νmin when the impedance |
Z| is at its minimum (|
Z|
min) and
νmax when the impedance is at its maximum (|
Z|
max) (
Figure 8a,b). At the same time, the phase angle (Θ) given in
Figure 8c,d tends to have a value of “zero.”
3.5. Modelling of Impedance—Frequency Characteristics of the Piezoelectric Equivalent Circuit
Let us first explain the notation used. In the adopted notation, square or box brackets [ ] denote that elements are in series-connected, whereas round brackets or parentheses ( ) denote the parallel connection of electric elements. According to the adopted notation, (RC) is a parallel circuit, while [RC] is a series connection of the elements R and C.
To accurately approximate the behaviour of the piezoelectric specimen close to its fundamental resonance, it can be represented by the electric equivalent circuit ([L
1R
1C
1]C
0) consisting of a “mechanical arm” (
L,
C, and
R connected in series) and
C0 (which corresponds to the electrical capacitance of the specimen) connected in parallel [
24]. In this connection, it is worth remembering that the impedance of the parallel circuit can be represented by the equivalent series circuit consisting of equivalent in series-connected resistance and reactance values.
In the case of our simulation and fitting, the modified electric equivalent circuit [RsCPE
1([L
1R
1C
1]C
0)] including resistance Rs and constant phase element CPE
1 connected in series with the “piezoelectric” equivalent circuit was used.
Figure 9 shows the electric equivalent circuit used for the simulation and fitting of the impedance response of the ceramic specimen vibrating close to its fundamental resonance.
It should be noted that the resonance frequency
νr and antiresonance frequency
νa correspond to the “zero” value of reactance for the electric equivalent circuit (
Figure 9). The reactance of the “mechanical arm” is zero at the series resonant frequency
νs when:
where
ω is the angular frequency (
ω = 2
πν).
The reactance of the parallel circuit is zero at the parallel resonant frequency (
νp). The parallel resonance
νp occurs when the currents flowing in the two arms are in antiphase, which is when:
In this connection, it must be pointed out that the relation between the above-mentioned characteristic frequencies of the equivalent circuit is as follows:
νmin <
νs <
νr. However, the difference between them is very small (
νmin~
νs~
νr) [
24]. Similar relation exists between antiresonance, parallel resonance, and |
Z|
max frequencies:
νa <
νp <
νmax (
νa~
νp~
νmax). What is important is that values of
νmin an
νmax can be readily measured using an impedance analyser (Alpha-AN High Performance Frequency Analyzer).
An example of the fitting results obtained for impedance characteristics measured under the influence of a weak circumferential magnetic field is shown in
Figure 10.
Figure 10a shows the spectra measured at
Hac = 0.05 A/m, (
E = 166 V/m;
U = 0.1 V
rms) at 20 °C, whereas
Figure 10b shows the results obtained for spectrum measured under the following conditions:
Hac = 0.49 A/m, (
E = 1333 V/m;
U = 2 V
rms) at 20 °C.
Experimental data were fitted to the electric equivalent circuit using the ZView programme (Scribner Associates, Inc. Southern Pines, NC, USA). Complex non-linear least squares method (CNLS) was employed for the analysis of the impedance/frequency data of the electroceramics [
17]. The fitting procedure was limited to the frequency range of the radial resonances ∆
ν = (2–6) × 10
5 Hz. The quality of the fitting procedure was estimated according to the following parameters: “
chi-squared” (
χ2) and weighted sum of squares (
WSS) [
29]. In the case of the fitting results shown in
Figure 10, the parameters were as follows: “
chi-squared” was
χ2 = 2.78 × 10
−3 and
χ2 = 3.23 × 10
−3 for
U = 0.1 V
rms and
U = 2.0 V
rms, respectively. The weighted sum of squares was
WSS = 0.314 and WSS = 0.177 for
U = 0.1 V
rms and
U = 2.0 V
rms, respectively.
Figure 11 shows the dependence of the fitting quality parameters, namely “
chi-squared” (
χ2) and weighted sum of squares (
WSS), on the temperatures at which the impedance/frequency characteristics of 67BFBT ceramics were recorded.
One can see in
Figure 11 that both
χ2 (“
chi-squared”) fitting quality parameter (
Figure 11a) and WSS parameter (
Figure 11b) change within one order of magnitude. Upon visual inspection, it can be seen that the linear approximations used for the
χ2 and WSS data approximation show that a higher electric field strength of the measuring signal (and, at the same time, a higher value of the circumferential magnetic field strength) improves the quality of the fitting procedure. This also means that the data is less susceptible to external interference. A comparison of the statistical characteristics of the obtained “
chi-squared” fitting quality parameters showed that the standard deviation was SD = 0.00192 and SD = 0.00212 for the low value of the electric field strength
E = 166 V/m (
U = 0.1 V
rms) and high electric field strength
E = 3333 V/m (
U = 2.0 V
rms), respectively. The
WSS parameter showed that the standard deviation was SD = 0.21743 and SD = 0.11077 for
E = 166 V/m and
E = 3333 V/m, respectively. An increase in the intensity of the measuring field led to a substantial improvement in the quality of further data simulation.
3.6. Calculation of Piezoelectric Parameters of 67BFTO Ceramics
The entry parameters used to calculate the piezoelectric parameters of BFBT ceramics, namely resonant frequencies
νr, anti-resonant frequencies
νa, impedance, and free capacitance
CT on 1 kHz were measured by using the aforementioned impedance analyser. As shown in
Figure 8, a thorough analysis of the experimental data (i.e., measured impedance spectra) was also performed. The procedure for the calculation of single coefficients is described in detail in classical textbooks (e.g., see [
21,
24,
25]), scientific papers (e.g., see [
22,
23]), or standards (e.g., see [
30]). The results of the calculations are given in
Figure 12 and
Figure 13.
Figure 12a shows dependence of the planar coupling factor
kp for the vibration along the radial direction in a circle-shaped disk of 67BFTO ceramics on temperature. The results of the calculations of the mechanical quality factor
Qm are given in
Figure 12b.
One can see from
Figure 12a that the coupling factor for both magnetic circumferential fields has a value of about
kp = 31%. Local extremes visible on the plots at −10 °C and −30 °C differ from the average value by about 1%. Therefore, it is reasonable to use linear regression fit for data analysis. The difference between the linear fit plots is rather small but can easily be discerned. Moreover, the linear regression fit for a higher magnetic field (
Hac = 0.49 A/m) shows better stability with temperature within the measured temperature range.
One can see from
Figure 12b that the mechanical quality factor
Qm exhibits non-monotonic behaviour with increasing temperature for both of the used values of magnetic circumferential field intensity. The local extremes are clearly visible. Within the temperature range ∆
T = (−30–+20) °C, the courses of the
Qm curves are almost identical. The influence of magnetic field intensity on
Qm becomes noticeable at about room temperature (20 °C). The difference between the linear fit plots can easily be discerned in
Figure 12b. One can see that the linear regression fit for higher magnetic field intensity (
Hac = 0.49 A/m; red line in
Figure 12b) shows a negative slope with temperature within the measured temperature range, whereas the linear regression fit for
Qm behaviour at a smaller magnetic circumferential field exhibits a positive one (
Hac = 0.05 A/m; blue squares; blue line in
Figure 12b).
Figure 13a shows the dependence of piezoelectric charge coefficient
d31 on the radial vibration mode of a thin disk (excited through the piezoelectric effect across the thickness of the disk). One can see from
Figure 13a that the piezoelectric charge coefficient
d31 increases from
d31 ≈ 30 pC/N to
d31 ≈ 45 pC/N at the following temperature range: ∆
T = (−20–+40) °C. The courses of the
d31 curves calculated for data measured at radial magnetic field
Hac = 0.05 A/m (blue squares; solid line;
U = 0.1 V
rms) and
Hac = 0.49 A/m (
U = 2.0 V
rms; red stars; dashed line) are almost identical. Additionally, the linear regression fits for the experimental data cannot easily be discerned in
Figure 13a. Linear regression (
Y =
A +
B ×
X) for the data measured at
Hac = 0.05 A/m (
U = 0.1 V
rms) exhibited the following parameters:
A = 3.44221 × 10
−11;
B = 2.57759 × 10
−13; and
R = 0.99097. Alternatively, linear regression for the data measured at
Hac = 0.49 A/m (
U = 2.0 V
rms) showed the following values:
A = 3.4412 × 10
−11;
B = 2.57385 × 10
−13; and
R = 0.99102. Therefore, to show the possible influence of the circumferential magnetic field on the piezoelectric charge coefficient
d31, the relative change in the
d31 piezoelectric modulus was introduced, and the results are plotted in
Figure 13b.
One can see from
Figure 13b that the plot of the relative change in the
d31 piezoelectric modulus exhibits non-monotonic behaviour with increasing temperature. Within the measuring temperature range, one local maximum and one local minimum are clearly visible in the plot. One can see from
Figure 13b that, according to linear fit, the relative change in the piezoelectric modulus has a positive sign within the whole measuring temperature range. This means that an increase in the intensity of the circumferential magnetic field generated by electric currents through the sample causes the suppression of the piezoelectric response of the multiferroic ceramic sample under study.
The above-mentioned influence of circumferential magnetic field on the piezoelectric charge coefficient
d31 can explain the dielectric properties of multiferroic ceramics. The point is that the magnetic field influences both real (
ε’) and imaginary (
ε”) parts of the complex dielectric permittivity. It was found that the higher the radial magnetic field, the lower the dielectric permittivity (at a given temperature). As an example, the dependence of the real and imaginary parts of the complex dielectric permittivity on frequency (below resonance) for radial magnetic field
Hac = 0.023 A/m (
U = 2.0 V
rms; ν = 10
5 Hz) and
Hac = 0.012 A/m (
U = 1.0 V
rms; ν = 10
5 Hz) at 20 °C are shown in
Figure 14.
The influence of the circumferential magnetic field on the dielectric properties of 67BFBT ceramics was also revealed when the difference of the modulus of dielectric permittivity was plotted against the difference of circumferential magnetic field caused by the electric currents (
iac) flowing through the ceramic sample (
Figure 15).
One can see in
Figure 15a that the difference in circumferential magnetic fields generated by the electric currents flowing through the sample linearly depends (in log-log scale) on the frequency of the measuring sinusoidal signal. On the other hand,
Figure 15b shows that the change in magnetic field strength causes a change in the modulus of the complex dielectric permittivity. Thus, the possibility of adjusting the dielectric permittivity (and therefore capacitance value) via changes in magnetic field intensity was obtained for 67BFBT ceramics. It is worth noting that changing the capacitance of a capacitor in an electric circuit has a predictable effect on the complex impedance and phase angle in the circuit. These parameter changes can be exploited to yield tuneable impedance-matching networks, tuneable filters, phase shifters, and other functional multiferroic devices.