1 Introduction

The magnetoelectric (ME) composites has gained great deal of attention over to single phase multiferroics due to its strain mediated ME coupling between piezoelectric and magnetostrictive phase [1, 2, 3]. Such ME composites with high coupling have paved the way for practical device application in meRAMs, sensors, and low frequency energy harvesters etc. [4, 5, 6, 7, 8]. The resultant ME coupling depends upon the type of magnetic and piezoelectric phase, their relative fraction, and type of geometry [9, 10]. Among the choices, Pb-based piezoelectric materials showed highest ME coupling due to its high piezoelectric coefficient (d33), high remanent polarization (Pr) and low coercivity (Ec) [11, 12]. Several reports on Pb-based ME composites as energy harvesters evidences its enormous ME response [13, 14, 15]. However, the hazardous effect of Pb, inclined researchers to explore the environment friendly alternate piezoelectric phase with slightly compromised properties [16, 17, 18, 19]. One of the most investigated piezoelectric phase is Bi0.5Na0.5TiO3 (BNT) which possess good Pr ~ 38 μC cm−2, moderate d33 ~ 58–95 pC/N and high temperature sustainability upto 320 °C [20, 21, 22]. On the other hand, spinel NiFe2O4 (NFO) is suitable magnetic phase as it exhibit strong piezomagnetic coefficient (dλ/dH ~ 251 ppm/T), good saturation magnetization (Ms ~ 55 emu/g) and low coercivity (Hc ~ 180 Oe) [23, 24, 25]. Further, the partial substitution of Ni2+ by Zn2+ enhances the MS as well dλ/dH [26, 27, 28].

Several studies on BNT based composites with different spinel ferrites were carried out and demonstrated the presence of ME coupling (4.0–7.5 mV/cm.Oe) [29, 30, 31, 32]. However, the applied frequency and dc magnetic field was 1 kHz and 3–5 kOe respectively, which is relatively high for device realization. In this work, composites of BNT and Ni0.5Zn0.5Fe2O4 (NZFO) were prepared by microwave sintering and their structural, dielectric and ME properties were investigated. The ME coupling of 4.3 mV/cm.Oe at very low magnetic field (800 Oe) and frequency (10 Hz) is observed, which indicate this ME composite could be a potential material, where low frequency is prerequisite requirement.

2 Experimental

High purity Bi(NO3)3 5H2O, CH3COONa, TiC12H28O4, Fe(NO3)3 9H2O, Ni(NO3)2 5H2O, and Zn(NO3)2 6H2O, were used to synthesized BNT and NZFO powders. The respective phase precursors were weighed in stoichiometry ratio and dissolved in acetic acid, and 2-methoxy ethanol for preparing solution of BNT and in deionized water, and citric acid for synthesizing solution of NZFO. The pH for solution of NZFO was maintained at 7 by adding ammonia into it. These solutions were continuously stirred and set at 120 °C till the gel formation. The obtained gels were dried at 180 °C and a white powder was obtained for BNT whereas the combustion has taken place for NZFO. The as-synthesized powders of BNT and NZFO were calcined at 600 °C and 900 °C respectively for 3 h. To prepare (1−x)BNT–xNZFO composites (x = 0.05–0.30; Δx = 0.05), the appropriate weight ratio of calcined powders were wet mixed using planetary ball mill for 3 h. The rpm and charge to ball ratio were fixed to 250 and 1:5 respectively. After mixing, powders were dried and uniaxially pressed into cylindrical pellets at a pressure of 50 MPa. The sintering of as-pressed pellets were carried out in microwave furnace at 1050 °C for 1 h. The phase identification of sintered pellets were carried out by X-ray diffraction (XRD) (X'PERT Pro, PANalytical) pattern, using Cu–Kα radiation. Raman spectroscopy was carried out by Micro-Raman Spectrometer (Labram HR Confocal, Horiba, France) (instrumental resolution ± 1 cm−1) equipped with a 532 nm diode pumped solid state laser at 25 mW power. A field emission gun-scanning electron microscopy (FEG–SEM) (Sigma 500, Carl-Zeiss, Germany) was used to study the microstructure of the samples. Prior to electrical and ME measurements, silver paint was used on both surfaces of the pellets. The frequency dependent dielectric measurements (100 Hz–1 MHz) were carried out using impedance analyzer (Solarton I-1260, UK) at 40–400 °C. The composite specimens were poled for ferroelectric measurements at 10 kV with 2 cm tungsten needle to specimen distance using Corona poling unit (Millman thin films PVT. LTD. Pune, India). Keithley (6517B, USA) electrometer was used to measure the leakage current density (J) with varying dc electric field (E). The d33 measurements were carried out using d33 meter (Sinocera, YE2730A, China). Ferromagnetic studies were done by vibrating sample magnetometer (VSM) (Lakeshore-7404, USA). Further, the Precision multiferroic-II system (Radiant Technology, USA) was used for ferroelectric and ME voltage coefficient (αME) measurements. The ME measurements were carried out in an applied ac magnetic field of 3 Oe at 10 Hz using Helmholtz coil (Lakeshore MH-6, USA). The dc magnetic field was varied using electromagnets (GMW 5480, USA). In this charge (q) and capacitance (C) was measured with Hdc at fixed Hac. The calculated voltage (Vout = q/C) was used to determine αME in terms of thickness (t) and Hdc as

$$\alpha_{ME} = \frac{{V_{out} }}{{t.H_{dc} }}$$
(1)

3 Results and Discussions

Figure 1 shows the representative XRD pattern of sintered 0.80BNT–0.20NZFO composite refined with rhombohedral (R3c) and cubic (Fd3m) phases. Both BNT and NZFO phases coexists without any impurity, which suggests no intermediate reaction has taken place among the phases within the used sintering conditions. The refined parameters, Rexp, Rwp were closed to 20 and χ2 is nearly 1 that suggests good agreement between obtained and fitted patterns. No changes in the lattice parameters and characteristic peak positions were observed for both the phases with obvious reasons. The obtained phase fraction from the refinement is comparable with the relative weight fraction of individual phases that used to prepare composites.

Fig. 1
figure 1

Refined XRD pattern of 0.80BNT–0.20NZFO composite sintered at 1050 °C for 1 h

Full size image

Figure 2 depicts the Raman spectra of BNT, NZFO and 0.80BNT–0.20NZFO specimens ranging from 100 to 750 cm−1. As suggested by group theory all 3 Raman active modes (A, B and C) are observed for BNT and in close agreement with previous studies [15, 17]. The bands A, B and C demonstrate the vibrations of Bi/Na–O, Ti–O, and Ti–O6 octahedra respectively. Further, these bands were deconvoluted with eight peaks that expressed the stretching and bending of the metal–oxygen bonds. The spectra of NZFO showed three Raman bands denoted by M, N and O. The M and N are related to symmetric stretching and anti-symmetric bending of metal–oxygen bonds at octahedral site respectively, whereas O corresponds to stretching at tetrahedral site [33]. The deconvoluted Raman peak positions of pure BNT and NZFO is tabulated in Table 1. In composites, the peaks of both the phases have been observed. However, low intensity of NZFO peaks ascribed to its low volume fraction. It is to be noted that similar to XRD, no change in peak position is observed, which denied the presence of micro-strain towards bond compression/stretching along interphase boundaries.

Fig. 2
figure 2

Raman spectra of BNT, NZFO, and 0.80BNT–0.20NZFO ceramics

Full size image
Table 1 Deconvoluted peak position of BNT and NZFO Raman spectra
Full size table

Figure 3 shows the representative backscattered electron images along with elemental mapping of 0.80BNT–0.20NZFO sintered composite. A dense microstructure with well distinguished sharp interphase boundaries of BNT (bright) and NZFO (dark) is clearly visible in the composite samples. Both phases have equiaxed grains with similar size distribution of 1–3 μm. However, the BNT shows larger fraction of coarse grains due to its higher phase fraction, whereas grain growth of NZFO may be hindered by the major BNT phase. The NZFO phase is found to be agglomerated as a consequence of mechanical mixing. The elemental mapping (Fig. 3b) along with individual elements present confirms BNT and NZFO grains as bright and dark contrast respectively in the samples. Mapping of Na element is beyond the detection limit of the equipment.

Fig. 3
figure 3

Microstructure a backscattered image, b elemental color mapping of 0.80BNT–0.20NZFO ceramics

Full size image

Figure 4 depicts the temperature (T) dependent dielectric constant (εr) and loss tangent (tanδ) at 1 MHz for (1−x)BNT–xNZFO specimens. For pure BNT and 0.90BNT–0.10NZFO specimens, εr found to increase gradually with temperature upto 220 °C and then a sharp increase has been observed. Previously, temperature dependent XRD and neutron diffraction studies suggested ferroelectric to anti-ferroelectric phase transition (referred as depolarization temperature, Td), with a corresponding change in the crystal structure i.e., rhombohedral to tetragonal respectively [34, 35]. The enlarged view of tanδ for BNT (inset Fig. 4b) also confirmed the transition at 220 °C. The anomalous increase in εr upto 320 °C suggests that transition is not sharp and persists till 400 °C [36]. This increase in εr despite the occurrence of antiferroelectric phase, may be ascribed to the existence of interphase boundaries which contributes to the polarization. The decrease in εr above 320 °C (Tm) is due to antiferroelectric to paraelectric phase transition as supported by sharp increase in tanδ after 320 °C [36] The composite with higher NZFO (x > 0.10) content showed diffuse phase transition behavior, which is usually observed in ME composites [37, 38].

Fig. 4
figure 4

Temperature dependent a εr and b tanδ of (1−x)BNT–xNZFO specimens at 1 MHz. Inset represent the enlarged view of tanδ for pure BNT

Full size image

Figure 5 showed the frequency dependent conductivity (σac) plots for BNT-NZFO specimens at different temperatures. The σac is calculated by the formula [39, 40],

$$\sigma_{ac} = \varepsilon_{0} \varepsilon^{\prime \prime } \omega$$
(2)

here ε0 is absolute permittivity in free space, ε″ is imaginary permittivity of specimen and ω = 2пf is angular frequency. The plot appears to be the combination of plateau and inclined conductivity regions. According to the Jonscher’s power law, the plateau indicates the dc conductivity (σdc) and inclined region represent frequency dependent conductivity (Aωn). Therefore, σac can be written as [41]

$$\sigma_{ac} = \sigma_{dc} + A\omega^{{\text{n}}}$$
(3)
Fig. 5
figure 5

Frequency dependent ac conductivity of (1−x) BNT–xNZFO specimens at different temperatures

Full size image

The increase in conductivity with frequency suggests that the conduction is governed by hopping of charge carriers between the localized state in accordance to Jump relaxation model (JRM) [42, 43]. On increasing temperature, thermally activated charge carriers contribute towards conduction. At sufficiently high temperatures, the contribution of frequency dependent conductivity is relatively small within the studied frequency range. Further Arrhenius plot of dc conductivity (σdc) for (1−x)BNT–xNZFO specimens are shown in Fig. 6. The activation energy (Eg) is calculated by

$$\sigma_{dc} = \sigma_{0} {\text{exp}}\left( { - E_{{\text{g}}} /KT} \right)$$
(4)

where K is the Boltzmann constant.

Fig. 6
figure 6

Temperature dependent conductivity of (1−x)BNT–xNZFO specimens at 1 MHz

Full size image

The conductivity increases with temperature for each specimen that represents their semiconducting behavior. Two slopes in low and high temperature regime are observed and suggests the different types of carriers are responsible for the conduction. At low temperatures, the conduction is governed by the hopping of electrons, while at high temperature the diffusion of oxygen vacancy contributes [44]. The increase in interphase boundaries fraction with NZFO content, restricts the movement of oxygen vacancies and consequently requires higher Eg as observed.

The ferroelectric behavior of (1−x)BNT–xNZFO specimens is confirmed by their RT P–E loops at 10 Hz as shown in Fig. 7a.

Fig. 7
figure 7

a P–E loops, b d33*, c J–E measurements, and d M–H loops of (1−x)BNT–xNZFO specimens

Full size image

A well saturated hysteresis has been observed for x = 0.00 and x = 0.05. Further increase of low resistive NZFO phase enhances the conduction losses, result in low field sustainability and unsaturated loops of composites. The obvious decrease in Pr is found with non-ferroelectric NZFO phase induction as shown in Table 2. Figure 7b depicts the E dependent bipolar strain plot of BNT–NZFO specimens. A normalized strain of 105.2 pm/V has been observed in pure BNT, which is decreased in composites with NZFO content. Further, an obvious declined trend in d33 is observed with NZFO content as tabulated in Table 2. The enhancement of conduction losses with NZFO content in composite specimens is confirmed by J–E plots as shown in Fig. 7c. The sharp increase in J till 0.5 kV/cm attributes to the space charge conduction. Above 0.5 kV/cm, gradual increase of J indicates the contribution of grain boundaries, and Poole-Frankel emission [45, 46]. The M–H loops for (1−x)BNT–xNZFO specimens are shown in Fig. 7d. The saturation magnetization (Ms) increases in composites due to high magnetic phase (NZFO) content.

Table 2 Pr, d33, normalized d33*, J, and Ms of (1−x)BNT–xNZFO specimens
Full size table

The ME coupling is a product tensor of both ferroelectric and ferromagnetic characteristics. As composite has exhibited both ferroelectric and ferromagnetic properties, a large ME response in such specimens is expected. The ME coefficient (αME) for all composites as a function of dc magnetic field (Hdc) with an ac field (Hac) of 3 Oe at 10 Hz has shown in Fig. 8a. The αME increases with Hdc till 800 Oe and decreases afterward that suggest the maximum strain mediated coupling occurred at 800 Oe. As NZFO content increases αME also increases upto x = 0.20 content and thereafter decreases (Fig. 8b). The maximum obtained value of αME is 4.33 mV/cm.Oe for 0.80BNT–0.20NZFO composite. The low αME for composite below x = 0.20 is due to the small fraction of magnetostrictive phase. For x > 0.20, the lower value of αME is due to the excess amount of NZFO content that have high J and limits the poling effect in composite [47]. Further, the effect of f on αME has been investigated as shown in Fig. 8c. No noticeable changes have been observed with f that suggest the linear behavior of αME at off-resonance f condition [48, 49], which is further supported by identical behavior of αME for 0.8 BNT–0.2 NZFO with Hdc at different f as shown in Fig. 8d. The high value of αME at low frequency is not reported so far, which indicates the potential of material. A comparative of BNT based ME composites is shown in Table 3.

Fig. 8
figure 8

Variation of αME with a Hdc at 10 Hz, b NZFO content and c f at 800 Oe for BNT–NZFO composites. d Variation in αME with Hdc at different f for 0.80BNT–0.20NZFO composite

Full size image
Table 3 Comparison of ME coefficient at different frequency for BNT based composites
Full size table

4 Conclusion

Lead-free ME particulate composite of BNT–NZFO were successfully synthesized. The coexistence of both phases was confirmed by XRD and FEG-SEM. The Raman spectroscopy suggested absence of interfacial micro-strains between BNT and NZFO phase. The temperature dependent dielectric study displayed the Td (~ 220 °C) and Tm (~ 320 °C) for pristine BNT and 0.90BNT-0.10NZFO specimens. However, such transition temperatures were obscured in 0.80BNT–0.20NZFO and 0.70BNT–0.30NZFO. The frequency dependent σac plot at different temperature followed the Jump relaxation model. The value of Eg was found to be increased with NZFO content that suggest the interphase boundaries restrict the movement of charge carriers. The NZFO content enhanced the ferroelectric losses and leakage current density in composites. An obvious increase in Ms with NZFO content were observed in composites. All samples showed good ME coupling and highest value of 4.33 mV/cm.Oe at 800 Oe was obtained for 0.80BNT–0.20NZFO at 10 Hz.