Structural, dielectric, and piezoelectric properties of lead-free (1 − x)K1/2Na1/2NbO3 − xCa(Zn1/3Ta2/3)O3 perovskite solid solution

Polycrystalline ceramics in (1 − x)K_1/2Na_1/2NbO₃ − xCaZn_1/3Ta_2/3O₃ (abbreviated as KNN-CZT) solid solution have been fabricated by using conventional solid-state synthesis route and the effect of CZT addition on the crystal structure, dielectric, and piezoelectric behaviour of KNN has been studied. Rietveld refinement of the room temperature X-ray diffraction data confirmed the crystal structure to be a pure perovskite phase for the compositions in the range x = 0 − 0.10. Further, the crystal structure gradually changed from orthorhombic to cubic via the formation of rhombohedral and tetragonal mixed phases with the increase in x. Raman spectroscopy suggested damping of phonon modes and a strong anharmonicity in the crystal arising due to the increased disorder in the structure as a consequence of multivalent cations occupying the A and B sites in CZT substituted compositions. Dielectric behaviour indicated the shifting of both tetragonal-cubic and orthorhombic–tetragonal phase transitions towards room temperature. The temperature dependent dielectric constant was modelled by Lorentz quadratic law, and the fitted value of diffuseness parameters confirmed an increase in diffuseness of phase transition with increasing substitution. Sample with the composition corresponding to x = 0.02 showed an improved piezoelectric coefficient d₃₃ ~ 125 pC/N and electromechanical coupling coefficient k_p ~ 30% at room temperature and d₃₃ ~ 61 pC/N and k_p ~ 24% at 300 °C which make this material a potential candidate for high-temperature piezoelectric applications. Variation of voltage coefficient (g₃₃) with the change in compositions are also reported. Improvement in the piezoelectric properties is attributed to the reduced oxygen vacancies.

Electronic supplementary material The online version of this article (10.1007/s10854-019-01881-1) contains supplementary material, which is available to authorized users.

The continued development of piezoelectric materials has led to an extensive application range from regular use to more specialized devices such as sensors, actuators, medical devices, etc. [[1]]. Lead-based materials are prominent in the piezoelectric industry for many decades due to their ability to produce significant piezoelectric properties as well as high Curie temperature. Mostly, the optimum piezoelectric properties are reported in compositions at the morphotropic phase boundary (MPB), which is the boundary separating the ferroelectric rhombohedral phase (high-temperature R3m or low-temperature R3c) from the ferroelectric tetragonal phase (P4mm) with respect to the composition in the composition—temperature phase diagram and is also independent of temperature [[3]]. Nevertheless, the toxic nature of lead limits their utilization and processing due to environmental concerns and restriction on the lead by many regulatory agencies [[5]–[7]]. Extensive work is in progress to develop suitable alternative materials with piezoelectric properties comparable to that of the lead-based system. Currently, the main focus is on three groups of perovskite-structured lead-free piezoelectric ceramics, i.e. BaTiO3 based, Na0.5Bi0.5TiO3 (NBT) based and K1/2Na1/2NbO3 (KNN) based materials [[8]–[11]]. KNN-based materials are promising candidates to replace lead owing to its good piezoelectric properties as well as high Curie temperature [[12]–[14]]. KNN is a solid-solution of ferroelectric compound potassium niobate (KNbO3) with space group Cm2m and antiferroelectric compound sodium niobate (NaNbO3) with space group Pbma at room temperature. The composition near x = 0.5 is of great interest because of the superior ferroelectric and piezoelectric properties [[15]–[18]]. This composition shows a rather high piezoelectric longitudinal response, a transverse coupling coefficient between those in modified lead titanate and hard PZT (d33 reported to be as high as 90 pC/N for (K0.5Na0.5)NbO3, and a relatively low dielectric constant (around 400 at room temperature). During heating, KNN shows three phase transitions: Rhombohedral to Orthorhombic (TR–O ~ − 110 °C), Orthorhombic to Tetragonal (TO–T ~ − 200 °C) and Tetragonal to Cubic (TC ~ − 400 °C) [[14]]. Compositions having polymorphic phase boundary near room temperature are reported to exhibit enhanced piezoelectric and ferroelectric properties due to the coexistence of two phases. Several solid-solutions based on KNN have been studied extensively in an effort to improve piezoelectric properties as well as to gain insights into the mechanisms responsible for such improvement. For instance, Guo et al. developed lithium niobate and lithium tantalate substituted KNN via solid-state synthesis route and reported an improved d33 around 206 pC/N [[19]]. A high d33 value ~ 320–324 pC/N, as well as better temperature stability, was achieved by the Li's group [[20]]. Zheang et al. and Rubio et al. have found the d33 value ~ 400 pC/N by modification of Rhombohedral–Tetragonal (R–T) phase boundary in KNN-based materials [[21]]. Satio et al. developed 〈001〉 textured (K0.44Na0.52Li0.04)(Nb0.86Ta0.10Sb0.04)O3 ceramics by reactive template grain growth (RTGG) method and achieved a d33 value of 416 pC/N [[23]]. A breakthrough in KNN-based piezoceramics was reported by Wu et al. in 2014 where excellent d33 (up to ≈ 420–490 pC/N) was achieved in 0.96(K0.44Na0.56)(Nb0.95Sb0.05)O3–0.04Bi0.5(Na0.18K0.82)0.5ZrO3 ceramics synthesized using the conventional solid-state method [[24]]. The same group in 2016 found highest d33 value around 570 ± 10 pC/N in a tertiary system (1 – x – y) K1−wNawNb1 – zSbzO3 − yBaZrO3 − x Bi0.5K0.5HfO3 [[25]]. Apart from improving d33, significant efforts have been devoted to improve the other piezoelectric properties of KNN based system like piezoelectric voltage coefficient (g33), planar electromechanical coupling coefficient (kp) and piezoelectric strain coefficient $\left(d_{33}^{\ast }\right)$ etc. For example, a high electromechanical thickness coupling coefficient was calculated around 82% for Li modified KNN based single crystal [[26]]. Some authors have reported improved electromechanical coupling in KNN based ceramics by exploring low temperature synthesis techniques to achieve better sinterability with controlled stoichiometry. Ohbayashi et al. developed (K, Na)NbO3-based lead-free piezoelectric ceramic with a KTiNbO5 system, (K1−xNax)0.86Ca0.04Li0.02Nb0.85O3−δ − KTiNbO5–BaZrO3–Co3O4–Fe2O3–ZnO with higher kp value of 51% [[12]]. The major issue with KNN based materials is the instability of piezoelectric properties with temperature, i.e., a slight change in temperature can decrease the property drastically. Several research groups tried to produce morphotropic phase boundary in KNN based system as in PZT to enhance the thermal stability of piezoelectric properties. Wang et al. developed a morphotropic phase boundary (MPB) in KNN-BNZ system with composition 0.035 < x < 0.045 [[27]]. Recently Pan et al. investigated MPB in (K, Na)NbO3–(Ca, Bi, Na)ZrO3 solid solution with optimum piezoelectric properties [[28]]. Another effective way to solve this issue is to make the polymorphic phase transition (PPT) more diffuse so that two phases coexist for a wide range of temperature and thereby achieving stable piezoelectric behavior in a certain temperature range [[29]–[31]]. However, improvement in piezoelectric properties is usually accompanied by a decrease in TC which limits the operation temperature window. In 2018, Quan et al. reported large piezoelectric response $\left(d_{33}^{\ast }\right)$ ~ 505 pm/V in the textured KNLN − BZ − BNT but with low TC ~ 247 °C which restricts their applications at high temperature [[32]].

One of the issues with KNN based ceramics is obtaining high relative density, which is partly due to volatilization of K+ and Na+ during sintering [[33]]. One of the primary objectives of this work was to develop dense KNN based materials with the improved thermal stability of piezoelectric properties by reducing the defects and/or by making phase transition(s) diffused. The perovskite ABO3 type additive can promote the formation of orthorhombic-tetragonal phase boundary at room temperature. In the present work, the solid solution of (1 − x)K1/2Na1/2NbO3 − xCaZn1/3Ta2/3O3 [(1 − x)KNN − xCZT] has been developed via the solid-state synthesis route, and the effects of CZT doping on the structural, dielectric, and piezoelectric properties of this system have been systematically investigated.

Samples in the (1 − x)K1/2Na1/2NbO3 − xCaZn1/3Ta2/3O3 solid-solution (x = 0, 0.005, 0.01, 0.02, 0.05, 0.07, 0.10) were prepared by the conventional solid-state synthesis technique. The raw materials: potassium carbonate (K2CO3)(purity > 99.9% Sigma Aldrich), sodium carbonate (Na2CO3) (purity > 99.9%, Sigma Aldrich), calcium nitrate tetrahydrate (Ca(NO3)·4H2O) (purity > 99%, Sigma Aldrich), zinc nitrate hexahydrate (Zn(NO3)2·6H2O) (purity > 99%, Sigma Aldrich), niobium pentoxide (Nb2O5) (purity > 99%, Alfa Aesar) and tantalum pentaoxide (Ta2O5) (purity > 99.5%, Alfa Aesar) (purity > 99.5%) were weighed in the stoichiometric ratio, mixed, and ground using mortar and pestle. K2CO3 and Na2CO3 were taken 10% extra due to the volatility of K+ and Na+ during high-temperature sintering and dried in an oven overnight before weighing due to the hygroscopic nature of these precursors. The ground powder was heated at 500 °C for the decomposition of nitrates and carbonates and was again heated at 700 °C for 12 h in air with intermediate grinding. The calcined powders were ground and mixed with small amount of PVA which was used as a binder and uniaxially pressed into the discs of about 1–2 mm in thickness and 10 mm in diameter. These pressed pellets were first heated to 600 °C for 5 h to burn out the PVA and subsequently were sintered at 1050–1080 °C for 12 h. The optimum sintering temperature was increased with increase in CZT content. The experimental densities of the sintered pellets were measured by the Archimedes method using Xylene (room temperature density ~ 0.86 g/cm3) as the liquid medium.

X-ray diffraction analysis was performed by Bruker D2 Phaser X-ray Diffractometer using Cu–Kα radiation to confirm the phase purity and crystal structure of the sintered samples. X-ray diffraction data were collected from 10° to 70° with a step size of 0.02°. Unpolarised micro-Raman measurements in the spectral range of 10 to 2000 cm−1 are performed in backscattering geometry via Labram HR-Evolution Raman spectrometer. To avoid the local heating only about 2mW laser power was used. The spectral excitation was probed using three different laser sources: 532 nm, 633 nm, and 785 nm. The laser beam was focused on the sample surface through 50 × LWD objective. The scattered signal collected with Peltier cooled charged coupled device detector. For the electrical measurements, the electrodes were prepared by applying the silver paste on either side of the pellets and curing at 400 °C for 20 min. The temperature dependent dielectric measurements in 30 °C to 550 °C temperature range and at various frequencies in 1 kHz–1 MHz range were carried out by using ZM 2376 LCR meter (NF Corporation). For piezoelectric measurements, samples were poled by applying a DC field ~ 20-30 kV/cm in a heated silicon oil bath. The room temperature piezoelectric charge coefficient was measured by using YE2730A d33 Meter under an applied force of 0.25 N and at a frequency of 110 Hz. Electromechanical coupling factor (kp) was calculated by the resonance–antiresonance method for all samples.

The crystallographic structural analysis for (1 − x)KNN − xCZT ceramics with (x = 0, 0.005, 0.01, 0.02, 0.05, 0.07, 0.10) was examined by the X-ray diffraction (XRD) studies and the room temperature XRD patterns are shown in Fig. 1. The XRD patterns plotted in semi-logarithm scale are presented in the Supplementary Information as Fig. S1. No secondary phases were detected (within the sensitivity limit of X-ray diffractometer used in this experiment) in samples with x < 0.07. A small amount of unidentified impurity phase is indicated by a low-intensity peak ≈ 37° (indicated by the symbol *) in the diffraction patterns for samples x = 0.07 and 0.1. Downward arrows in Fig. 1 denote the diffraction peaks for (001)C and (011)C plane corresponding to Cu–Kβ (λ = 1.39 Å) in the incident X-ray beam. 1. The diffraction patterns around the 2θ position (44–48°) of the peaks corresponding to the Bragg reflections (200)O and (022)O (subscript O indicates that the indexing is done on the basis of an orthorhombic unit cell) were enlarged to emphasise the variation in the peak position and relative peak intensity with CZT doping. It must be noted that for a pure orthorhombic KNN, the intensity ratio of (002)PC/(200)PC (subscript PC indicates that the indexing is done based on the primitive cubic unit cell) peaks at 2θ position ≈ 46° is expected to be about 2 [[34]]. The calculated intensity ratio of (022)/(200) peak decreases with increase in CZT content and (022) and (200) peaks merge into a single broad and slightly asymmetric peak for compositions with x ≥ 0.05. Similar behaviour has been observed in various KNN-based solid-solution and is an indication of modification in crystal structure upon doping [[34]]. It should be noted that the shifting of the peaks in X-ray diffraction patterns, shown in Fig. 1, has two contributions: first from the change in lattice parameters, and second due to the sample displacement errors. The vertical bars shown in the right side panel of Fig. 1 represent the sample-displacement-error corrected 2θ position of (002)PC peak and demonstrate the reduction in unit cell volume with the increase in CZT doping concentration. These peaks suggest that the orthorhombic phase (Amm2) for pure KNN is transforming to mixed phase tetragonal and orthorhombic phase with space group P4mm + Amm2 for x = 0.05 and then into pseudocubic with further increase in CZT content. Pauling rules for the coordination of cations in ionic crystals is dependent on the cation to anion radius ratio [[35]]. The ionic radii of K+ (coordination number, C. N. 12), Na+ (C. N. 12), Nb5+ (C. N. 6), Ca2+ (C. N. 12), Zn2+ (C. N. 6), and Ta5+ (C. N. 6), are 1.64 Å, 1.39 Å, 0.64 Å, 1.34 Å, 0.74 Å, and 0.64 Å, respectively [[36]]. Accordingly, Ca2+ is expected to occupy the 12 coordinated A-site whereas Zn2+, and Ta5+ are expected to occupy the octahedral coordinated B-site of the perovskite unit cell. The difference in ionic radii develops a lattice mismatch which results in structural fluctuations [[37]]. The phase stability depends on the tolerance factor of the material (which reflects the structural distortion) given by Eq. 1:

Graph

where RA and RB are the ionic radii of cations at A- and B-site and RO is the ionic radius of oxygen anion. The tolerance factor of the system decreases from 0.996 for pure KNN to 0.989 for the composition with x = 0.10 with the increase in CZT content. Lowering of tolerance factor (for t ≤ 1) in perovskites is related to the lowering of crystal symmetry [[38]]. Accordingly, an increase in the structural distortion with increasing CZT content is expected. Nevertheless, the mixed cations occupancy at A-site and/or B-site in perovskites results in the localized chemical and structural in homogeneities which affect the crystal structure. To further study the compositional dependence of the crystal structure changes Rietveld refinement of room temperature X-ray diffraction data was performed using TOPAS 3.2 software.

Graph: Fig. 1Room temperature powder X-ray diffraction patterns for (1 − x)KNN − x CZT with x = 0, 0.005, 0.01, 0.02, 0.05, 0.07, and 0.10. Indexing of the planes is on the basis of relevant unit cells (subscripts O: orthorhombic, T: tetragonal, and C: cubic)

Representative calculated patterns along with the observed patterns for selected compositions (x = 0, 0.02, 0.05, and 0.10) are shown in Fig. 2 Rietveld refinement of the room temperature X-ray diffraction data for compositions with x = 0.005, 0.01, and 0.07 are shown in Supplementary Information as Fig. S2. and the fitted parameters are listed in Table 1. The open symbols are the experimental data, solid red lines are the calculated/fitted patterns, the solid line at the bottom is the difference between the observed and calculated pattern, and vertical bars represent the 2θ positions of all possible Bragg reflections for the corresponding space group (Fig. 2). During the Rietveld refinement, peak shape was modelled using the Pseudo-Voigt with an axial divergence asymmetry function, and the background was modelled using a 6-order polynomial function for all the compositions. In the initial step of the refinement, parameters such as sample displacement error, unit cell parameters, and background parameters were varied. Fractional coordinates for the space groups Amm2, P4mm, R3m, and Pm $\overline{3}$ m, as reported in the literature, were used as the starting model and the occupancies of Wyckoff positions sites were fixed according to the nominal stoichiometry for each composition [[39]].

Graph: Fig. 2Rietveld refinement of the room temperature X-ray diffraction data for the selected compositions: a x = 0, b x = 0.02, c x = 0.05, and d x = 0.10

Crystallographic data (space group, lattice parameters) and structure refinement parameters of various compositions in (1 − x)K1/2Na1/2NbO3 – (x)CaZn1/3Ta2/3O3 system

Temperature-dependent dielectric studies (see next section) carried out on the sample with the composition corresponding to x = 0.02 indicated that the diffused orthorhombic to tetragonal phase transition ( $T_{O-T}$ ) is centered around 150 °C. It suggests that symmetry of crystal structure for this sample is lower than tetragonal, i.e. orthorhombic or orthorhombic + rhombohedral mixed phase) at room temperature. For samples with x = 0, 0.005, 0.01 and 0.02, refinement using the single orthorhombic Amm2 space group resulted in a good match between calculated and observed patterns and acceptable reliability R-factors. The overall trend of (002)PC/(200)PC in the X-ray diffraction patterns (Fig. 1 panel on the right) is suggestive of a decrease in the orthorhombic phase content with the increase in CZT doping. For x ≥ 0.05, a broad asymmetric peak with no apparent splitting is observed. Accordingly, a combination of two phases [orthorhombic Amm2 + tetragonal P4mm for x = 0.05, tetragonal P4mm + rhombohedral R3m for x = 0.07, and tetragonal P4mm + cubic Pm $\overline{3}$ m for x = 0.10] was used during refinement. Good agreement between the observed and calculated patterns, low values of R factor settle the suitability of assigned structure models. The calculated lattice parameters (a, b, c), volume (V), and the values of the goodness of fit (GOF), Rp, and Rwp for various compositions are provided in Table 1. On CZT substitution, the effective ionic radius of B-site shows a small increase (+ 0.1 Å on complete B-site substitution) whereas the effective ionic radius of A-site shows a relatively larger decrease (0.17 Å on complete B-site substitution). Thus, the unit cell volume decreases with increase in CZT substitution.

Raman spectroscopic technique is used to probe the dynamics of phonons in the system as a function of doping. As confirmed by the Rietveld refinement of the room temperature X-ray diffraction data, KNN possesses an orthorhombic structure with Amm2 ( $C_{2\nu }^{14}$ ) space group. According to the group theory, orthorhombic Amm2 phase consists of 4A1 + 4B1 + 3B2 + A2 Raman active optical modes. All these modes are also infrared active except mode A2. The observed lattice vibrations can be separated into translational modes of the isolated cations and internal modes of coordination polyhedron [[40]]. The number of Raman active modes for a given material depends on the crystal structure. However, it is important to note that the disorders result in mode overdamping as well as a break in symmetry and, consequently, in modes overlapping and apparent disappearance of some of the low-intensity modes [[41]]. CZT doping induced positional and structural cationic disorders are reflected by the decrease in the intensity of modes in doped samples. With the increase in the CZT doping in KNN, there is a change in the crystal symmetry as confirmed by the Rietveld analysis of the room temperature X-ray diffraction data and same is also reflected by the change in intensity and FWHM of these modes. No Raman active modes are expected for cubic perovskite structure (i.e. for space group Pm $\overline{3}$ m). Modes observed in the Raman spectrum for the composition corresponding to x = 0.1 is because of the presence of a mixed tetragonal (86.1 ± 1%) and cubic (13.8 ± 1%) phase. The NbO6 octahedron consists of A1g(v1) + Eg(v2) + 2F1u (v3, v4) + F2g(v5) + F2u(v6) modes of vibration. Among these modes A1g(v1), Eg(v2) and F1u(v3) are stretching modes whereas F1u(v4), F2g(v5) and F2u(v6) are bending modes [[43]]. Figure 3 shows the room temperature Raman spectrum of pure KNN revealing the vibration modes of NbO6 octahedron in the spectral range of 100–1000 cm−1. Modes with the frequency lower than 200 cm−1 can be assigned to the translational modes of Na+/K+ cations and rotations of the NbO6 octahedra. The F2u(v6) mode which is both Raman and IR active appears in this region, but it is too weak to be distinguishable. A relatively strong Raman active mode appears on account of O–Nb–O bending vibrations of NbO6 octahedra around 260 cm−1 (F2g(v5)) which corresponds to a nearly perfect equilateral octahedral symmetry, while another strong Raman active mode appears at 615 cm−1 (A1g(v1)) attributed to a symmetric Nb–O stretching vibration. The other two distinguishable Raman active modes located at 550 and 860 cm−1, belong to Eg(v2) and [A1g(v1) + F2g(v5)] vibrational modes, respectively.

Graph: Fig. 3Room temperature Raman spectrum of K 1/2 Na 1/2 NbO 3 using 532 nm laser

The change in the crystal structure of the KNN with CZT content is also reflected in the shift in Raman modes, corresponding intensities and the line widths [[45]]. Figure 4a–c shows the observed Raman spectra of the (1 − x)KNN − xCZT ceramics at room temperature with x ranging from 0.005 to 0.10 using 532 nm, 633 nm and 785 nm laser source, respectively. The observed results reveal that the intensities of the bands originated from the vibrations of the NbO6 octahedra decrease with increasing CZT content. The observed Raman band intensity depends on sample structure and phase orientations. Our observation of decreasing intensity with increasing doping suggests such local reorientations/phase modifications. Moreover, the addition of the CZT contents to the pure sample affect the packing of molecules in the crystal which influence the prominent factors such as vibrational frequency and phonon lifetime. Hence, the variation of the mode frequency and full width at half maximum (FWHM)/damping coefficient for the prominent phonon modes, i.e. P2, P4, P5 was taken into consideration. Details of the frequency for all the modes as a function of doping is given in Table 2 for three different excitation wavelengths. Inset of Fig. 4a and c shows the normalized intensity of the modes P2, P4, and P5, respectively, as a function of different incident photon energy. Our observation suggests a resonance in the visible range of spectra (i.e. ~ 2.3 eV).

Graph: Fig. 4Evolution of Raman spectra of (1 − x)KNN − x CZT as a function of doping (x) for a 532 nm, b 633 nm and c 785 nm laser source respectively. Inset in a and c shows the normalized intensity of the prominent modes as a function of different incident photon energy

List of observed modes frequency for different concentrations (x) of CZT for three different incident photon wavelengths

Units are in cm−1

Figure 5a shows the evolution of modes frequency and linewidth as a function of doping for the excitation wavelength of 532 nm. It is observed that with increasing CZT content, modes P2, P4, and P5 shift towards lower wave number. We note that peak frequency, as well as linewidth for all these modes, shows nominal change with doping till x ≤ 0.02 suggesting no change in the phase. However, with further doping, for x = 0.05 and 0.1, change in the frequency and linewidth is drastic, clearly reflecting the change in symmetry of the sample. Also, we note that all the observed modes in the Raman spectra are associated only with the NbO6 octahedral. Conventionally, with a change in symmetry in different phases, one would expect emergence of new phonon modes or loss of existing modes according to the changed symmetry. However, in the present case with different phases at higher doping, the number of modes remains same; this can be attributed to the fact that all the observed modes are associated with the vibrations of oxygen atoms in BO6 octahedral. Further, the Raman modes at room temperature for KNN based compositions are broad and show substantial overlapping. Octahedron does change its orientation, and effective distance between the oxygen and B-site cation also vary in different phases as also reflected in the renormalization of the mode frequency and linewidth at higher doping. The peaks shift to the lower frequency at higher doping, suggesting a decrease in the binding strength of B-site with oxygen atoms and this may result from the reorientation of the octahedral in different phases at higher doping.

Graph: Fig. 5Dopant concentration (x) dependence of mode frequency and FWHM for modes P2, P4, P5 of (1 − x)KNN − x CZT a for 532 nm, and b for 633 nm. Solid lines are the guide to the eyes

Further, the changes in structural distribution due to the addition of CZT content reflect the effect on the lifetime of the excited phonons and the damping coefficient (FWHM) of modes P2, P4, and P5 increased with the increasing x thus suggesting that disorders are increasing with the substitution. Also, when the sample was exposed to 633 nm (see Fig. 5b) and 785 nm (not shown here) laser source, all the modes shows similar behaviour as that in case of 532 nm laser source. Details of the evolution of the frequency of these phonon modes as a function of doping are given in Table 2.

We notice that with increasing doping, the change in the mode frequency is very high, ranging from ~ 5 to 15 cm−1. Adopting a simple cubic anharmonicity model for phonon, considering the decay of an optical phonon into two phonons of equal energy, the change in the mode frequency is given as $\mathrm{\Delta }\omega \left(T\right)=A\left[1+2n\left(\frac{\omega }{2}\right)\right]$ , where, ω is the mode frequency, $n\left(\omega \right)=1/\left(exp\left(\frac{ħ\omega }{k_{B}T}\right)-1\right)$ is the Bose–Einstein factor and A is the self-energy parameters which is related to the phonon–phonon interaction. For the modes under considerations, we made a rough estimation that if one start with the pure KNN sample than to observe a similar change in the mode frequencies temperature should be in the range of 400–500 K. It further suggests that the effect of doping on the phonons can be mapped to the corresponding thermal effect. Our observations suggest strong anharmonicity in the crystal arising due to the atomic displacements on addition of CZT to the pure KNN.

FESEM imaging was performed to study the microstructural evolution of KNN-CZT ceramics with increasing CZT content. Figure 6 shows the micrographs of the fractured surfaces of selected compositions (x = 0, 0.02, 0.05, and 0.10) in (1 − x)KNN − xCZT system Note that scale bars are different for different micrographs. The relative density for all compositions was calculated by using the experimental density measured by the Archimedes method and the theoretical density of the sample calculated by using the unit cell volume and weight. The relative density was found to be in the 91–93% range for all samples. The average grain size was found to decrease from 6 µm (for x = 0 sample) to 0.7 µm (x = 0.10 sample) with increase in CZT doping. Similar behaviours are also reported for solid solutions of KNN with various ABO3 type compounds [[46]].

Graph: Fig. 6SEM micrographs of the fractured surface of (1 − x)KNN − x CZT ceramics for x = 0, 0.02, 0.05, 0.10

The frequency dependence of room temperature dielectric constant and dielectric loss is shown in Fig. 7 for all samples (x = 0, 0.005, 0.01, 0.02, 0.05, 0.07, and 0.10). The typical characteristics of dielectric constant and dielectric loss are seen for all samples, i.e. dielectric constant and dielectric loss both decreased with increase in frequency. This behaviour can be attributed to Maxwell–Wagner type interfacial space charge polarization explained by Koop's theory. According to this theory, in polycrystalline ceramics microstructure consists of grains separated by the grain boundaries. Mobile charges in the system including defects and oxygen vacancies create a space charge polarization across grain boundary and sample-electrode interface which dominates the overall polarization in lower frequency regime (especially at high temperatures) and diminishes at higher frequencies [[48]]. The dielectric constant and dielectric loss also varies with the composition; dielectric constant increases with x while the dielectric loss is decreasing with increase in CZT content (x). The highest room temperature dielectric constant ~ 1164 (at 1 MHz) and a low dielectric loss ~ 0.04 (at 1 MHz) was achieved for the composition corresponding to x = 0.07. The decrease in dielectric loss of the sample is due to a reduction in oxygen ion vacancy of the system. Oxygen vacancies in this system are created due to volatilization of K and Na, which is suppressed with the addition of nonvolatile Ca2+ at A- site (as K/Na amount in the sample becomes lesser) [[50]–[52]]. At higher doping dielectric loss is increasing due to the poor electrical insulation of the samples which can be attributed to the increase in the concentration of defect dipoles $\left[3{\text{Ca}}_A^{·}-{{\text{Zn}}^{″\prime }}_{\mathit{Nb}}\right]$ ; here ${\text{Ca}}_A^{·}$ represents the point defect with +1 effective electronic charge formed due to divalent calcium ion replacing the A-site of the perovskite structure and ${{\text{Zn}}^{″\prime }}_{\mathit{Nb}}$ represents the point defect with − 3 effective electronic charge formed as a consequence of divalent zinc ion substituting pentavalent niobium ion at B-site.

Graph: Fig. 7Variation of room temperature dielectric constant (a) and dielectric loss (b) with frequency for various compositions

The polymorphic phase transition behaviour of this system confirms the structure determined by XRD. Temperature dependence of dielectric constant and dielectric loss for various compositions at four different frequencies 1 kHz, 10 kHz, 100 kHz and 1 MHz is shown in Fig. 8. Two phase transitions for the pure KNN corresponding to TO–T ~ 200 °C and TC ~ 400 °C can be seen in Fig. 8a and with the increase in doping content up to x = 0.05, both TO–T and TT–C shift towards lower temperature. With further increase in CZT content, the peak corresponding to TO–T is not apparent in Fig. 8e–g which is possibly due to the increase in diffusivity of the phase transition. It also can be noticed from Fig. 8e–g that all CZT doped samples have lower Tm as compared to that of undoped KNN and it shows an almost linearly decreasing relation with x as shown in Fig. 9a. Phase coexistence plays a crucial role in the enhancement of piezoelectric properties of the materials due to the increase in the number of possible polarization directions [[53]–[55]]. It was reported that Ta5+ substitution at Nb5+ site had been used to decrease the TO–T and increase TR-O with a controlled rate; hence, TO–T or TR–O can be tuned to near room temperature [[56]]. Analysis of X-ray diffraction data suggests that the rhombohedral phase coexists with the tetragonal phase for the composition x = 0.07. This would happen only if the rhombohedral to orthorhombic phase transition TR-O is shifting towards higher temperatures with an increase in x.

Graph: Fig. 8Temperature dependent behavior of dielectric constant (solid lines) and dielectric loss (dashed lines) at various frequencies for a x = 0, b x = 0.005, c x = 0.01, d x = 0.02, e x = 0.05, f x =0.07, and g x =0.10

Graph: Fig. 9a The plot for Tm (temperature of dielectric maximum ) and TO–T versus x and b variation in diffuseness of transition temperature with increase in x

The sharp phase transition observed in pure KNN sample becomes progressively more diffused with the increase in CZT substitution, but no significant frequency dispersion in transition temperature appears in this system. The broadening of peaks represents the overlapping of several ferroelectric and non-ferroelectric regions. The degree of diffuseness has been calculated by using modified Curie Wiess law,

2 $$\frac{1}{{\epsilon }_r}-\frac{1}{{\epsilon }_m}=C^{-1}{\left(T-T_m\right)}^{\gamma }$$

Graph

where C is Curie–Weiss constant and $\gamma \left(1\le \gamma \le 2\right)$ is a frequency dependent parameter which, reflects the degree of diffuseness of the phase transition at higher frequency [[57]]. Figure 10a-g show the $ln\left(\frac{1}{{\epsilon }_r}-\frac{1}{{\epsilon }_m}\right)$ versus $ln\left(T-T_m\right)$ curve for all compositions where the symbols represent the experimental data, and solid lines represent the linear fitting of Eq. (2). A good fitting was obtained for all compositions in the temperature range starting, from the temperature above the transition temperature (Tm). The calculated fitting parameters C and γ by using Curie–Weiss law are shown in Table 3. The degree of diffuseness (γ) increases from 1.13 to 1.63 with an increase in CZT doping, which suggests that the structural changes occurring with doping induce the diffusivity of the transition. The effect of substitution at A-site and B-site of KNN on the diffuseness of ε(T) maximum can also be observed by Fig. 9b. The reason behind the diffuse phase transition behaviour of the system can be explained by various theoretical predictions such as super paraelectricity, combinational fluctuation theory, and polar micro-regions (PMR) etc. [[58]]. The structural disorder and compositional variations are commonly responsible for these models. The broadening of the peak and slight frequency dispersion in KNN-CZT system is assigned to the random distribution of atoms, i.e. Ca2+ is occupying the A-site and Zn2+ and Ta5+ replacing the Nb5+ at B-site of ABO3 perovskite structure of KNN-CZT which results in a structural disorder of the system due to the differences in ionic radii and charge imbalance. The increased diffuseness of the phase transitions indicates that the mixed phase is achieved over a relatively larger temperature range, which would help obtain thermally stable piezoelectric and ferroelectric properties.

Graph: Fig. 10a–g Fitting of dielectric constant versus temperature curve of various compositions by using modified Curie–Weiss law at 1 MHz. Symbols represent experimental data while solid lines represent the fitted curve

Values of various dielectric and piezoelectric parameters for different compositions of (1 − x)K1/2Na1/2NbO3 − (x)CaZn1/3Ta2/3O3 system

Piezoelectric properties of the KNN-CZT were measured for all compositions after poling in the silicon oil bath at 30–80 °C by applying a dc electric field of 2–3 kV/mm. The samples used for the piezoelectric measurements were ~ 1 mm in thickness and ~ 10 mm in diameter. The piezoelectric charge coefficient was measured by using a d33 m based on the Berlincourt method [[60]]. The piezoelectric charge coefficient (d33) value for the pure KNN was found to be ~ 88 pC/N which is comparable to the typical values reported for this composition. With the increase in CZT content, the value of d33 increases to a maximum of 125 pC/N for the composition corresponding to x =0.02 and then start decrease upon further increase in CZT amount reaching to 20 pC/N (for x = 0.10). The planar electromechanical coupling factor is a key parameter for transducer applications. The electromechanical coupling factor (kp) was evaluated by using resonance and antiresonance frequencies corresponding to the minimum and maximum frequencies in an impedance frequency spectrum. The planar electromechanical coupling coefficient for all samples was calculated by using Eq. 3.

3 $$k_p=\frac{1}{\sqrt{0.395\frac{f_r}{f_a-f_r}+0.574}}$$

Graph

where fr and fa are resonant and antiresonant frequencies in frequency dependent impedance spectrum [[61]–[65]]. Calculated values of kp for various compositions are provided in Table 3. The highest value of kp was found to be ~ 30% for the x = 0.02 sample.

The improvement in the room temperature piezoelectric properties in many KNN and NBT-based systems has been attributed to the coexistence of tetragonal and rhombohedral phases. The shifting of polymorphic phase transition near the room temperature by appropriate substitution has been used widely to achieve this albeit at the expense of useful temperature range. It should be noted that in the present system, the sample with composition x = 0.02 shows highest d33 and kp though its crystal structure is orthorhombic with TO-T above 150 °C (i.e., not a mixed phase composition). This indicates that the enhancement in piezoelectric properties for the x = 0.02 sample is because of oxygen vacancies suppression [[66]]. Oxygen vacancy in oxide systems can move easily due to its high mobility and accumulate at the places with low free energies such as grain boundaries or domain walls. Further increase in doping concentration (x > 0.02) decreases the piezoelectric properties which can be attributed to the drastic decrease in the average grain size. Piezoelectric properties are strongly related to the grain (and domain) size. The smaller grain size results in a larger fraction of grain boundaries in the sample which may impede the long-range cooperative phenomena like piezoelectricity and ferroelectricity. Furthermore, accumulation of defects at the grain boundaries can act as an obstacle to the domain wall motion and results in domain pinning effect [[56]].

The piezoelectric voltage coefficient for all samples was calculated using the piezoelectric charge coefficient (d33) and dielectric permittivity (ε33) measured at 110 Hz.

4 $$g_{33}=\frac{d_{33}}{{\epsilon }_{33}}$$

Graph

As shown in Table 3, the value of g33 varies with doping and the maximum value (31.01 × 10−3 Vm/N) is for 2% CZT doped sample. The temperature stability of piezoelectric properties of the sample with the composition corresponding to x = 0.02 to is shown in Fig. 11. It is observed that the value of electromechanical coupling factor kp remains almost constant with increase in temperature up to 200 °C and starts decreasing rapidly on further increase in temperature. The piezoelectric coefficient d33 also shows good thermal stability, and the decrease in its value is from 125 pC/N at room temperature (27 °C) to 105 pC/N at 150 °C. Above 150 °C, d33 decreases at a relatively faster rate with increasing temperature. This behaviour can be explained in terms of mobility of ferroelectric domains. It is to be noted that the temperature of orthorhombic to tetragonal phase transition as evidenced by the dielectric response is ~ 150 °C. With the increase in temperature up to 150 °C the domain walls are moving insignificantly so that sample remains poled; but above 150 °C, domain wall movement becomes more facile, so depolarization takes place and consequently, piezoelectric properties of the sample start to degrade. The improved piezoelectric properties of the x = 0.02 sample maintain good values even at higher temperatures (d33~ 60 pC/N, kp ~ 24% at 300 °C) demonstrating this composition as a potential lead-free candidate for piezoelectric applications at higher temperatures.

Graph: Fig. 11Variation of piezoelectric parameters with the temperature for the ceramic sample with x = 0.02: a electromechanical coupling factor (kp) versus temperature and b piezoelectric charge coefficient (d33) versus temperature

In summary, ceramics of various compositions (1 − x)KNN − xCZT system were fabricated via the conventional solid-state reaction route and were characterized for their structural, dielectric, and piezoelectric properties. The X-ray diffraction analysis showed that the structural change occurring with an increase in CZT content in systems. The structure was found to change from orthorhombic (Amm2) for pure KNN to a mixed phase of rhombohedral (R3m) and tetragonal (P4mm) phase for x = 0.07 sample and then to a mixed tetragonal (P4mm) and cubic (Pm $\overline{3}$ m) phase for x = 0.1 sample. The average grain size was found to decrease with the increase in doping concentration. The room temperature dielectric permittivity and dielectric loss increase with the increase in CZT content. In accordance with the XRD results, temperature dependent dielectric behaviour also confirmed the structural changes by shifting of polymorphic phase transition temperatures and both TO–T and Tm shifted towards room temperature. Further, diffuseness of the phase transitions also increased due to the increased structural disorder in the system with increasing substitution. The value of piezoelectric parameters improved for certain compositions, and the planar electrometrical coupling factor (kp) was calculated to be ~ 30% along with an enhanced piezoelectric coefficient d33 = 125 pC/N for the sample with x = 0.02. The temperature variation of the piezoelectric properties indicates the usefulness of the 2% CZT substituted composition in high-temperature piezoelectric applications. Interestingly, the enhanced piezoelectric properties are obtained for the composition with only 2% and not for the sample which shows the coexistence of rhombohedral and tetragonal phases at room temperature which illustrates that the improvement in piezoelectric properties is defect mediated rather than through the shifting of polymorphic phase transition.

SK thanks Scientific and Engineering Research Board (SERB) for the funding this research through the Early Career Research Award (Grant Number: ECR/2017/000561). SK and PK also gratefully acknowledges the financial support from the Department of Science & Technology, New Delhi under INSPIRE Faculty scheme. PK also thanks AMRC, IIT Mandi for Raman spectroscopy facility.

Graph: Supplementary material 1 (DOCX 330 kb)

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.