Introduction

Research in the domain of lead-free piezoelectric materials has progressively gained momentum due to environmental concerns of toxic lead (Pb). Lithium sodium niobate, Na1−x Li x NbO3 (LNN), a promising lead-free piezoelectric system, has been receiving attention due to its potential in piezoelectric resonator and filter applications, for the compositions near the morphotropic phase boundary (MPB), viz., x = 0.12, commonly known as LNN-12 [1]. Both electromechanical coupling factor, k p , and mechanical quality factor, Q m , the figure of merit for these applications, found to be strongly dependent on the density and microstructure of the sintered sample [2, 3]. Hence, the knowledge of sintering behavior is essential in order to have good understanding of the final microstructure that evolves during sintering [4]. Usually, these alkali niobate-based ceramics are prepared by solid-state reaction (SSR) between alkaline carbonates and niobium pentoxide (Nb2O5) at temperature ~900 °C followed by sintering at temperature ~1250 °C. Synthesis of these niobates using conventional ceramic method leads to many unavoidable issues such as loss of alkali elements at elevated temperature that essentially create problems in achieving desired stoichiometry in the final product, chemical inhomogeneities, and abnormal grain growth. Sometimes, the variability of different decomposition/melting temperatures of starting materials leads to chemical inhomogeneities. The Nb2O5 starting material has a melting point of 1520 °C, whereas Na2CO3 and Li2CO3 have much lower melting points, 851 and 720 °C, respectively. Therefore, Nb2O5 is relatively refractory in nature, whereas Li2CO3 decomposes at a faster rate than Na2CO3 at lower temperature [5]. Another very important issue arises in synthesis process from the hygroscopic behavior of the alkaline carbonates, which are commonly used as raw materials for preparation by SSR [6, 7]. The humidity content in these raw materials headed toward a wrong weight measurements, if special care is not taken, and makes it difficult to maintain the proper stoichiometry in the final product. There are many techniques reported for sodium potassium niobate (KNN) system to minimize the effect of volatilization of alkali elements, for example spark plasma sintering [8] and double crucibles [9]. These methods are complicated, not cost-effective, and therefore undesirable for mass production in industries. Moreover, the use of excess alkali elements to compensate the loss was also conceived as an alternative [10]. However, it has been reported in the literature that excess alkali addition may promote grain growth, which deteriorates electrical properties in the resulting ceramics.

In view of the above, our strategy is to study the microstructure behavior of stoichiometric LNN-12 ceramics and also the effect of introducing excess Li2CO3 into starting mixture in order to compensate possible alkali element losses during sintering at higher temperature. Particle size and morphology are also investigated to better understand the densification behavior. Here, we have investigated the effect of only Li2CO3 addition on microstructure and electrical conductivity as it is well known that the tendency of Li loss is more pronounced than Na [11, 12].

Materials and methods

All the samples in this study were prepared using conventional solid-state reaction followed by sintering. The starting raw materials were reagent-grade Nb2O5, Na2CO3 (both 99.5 % pure, Loba Chemie, India), and Li2CO3 (99.0 % pure, Merck, India) and were kept in electrical oven for overnight at 150 °C to avoid the humidity content in alkali carbonates. These were mixed in the desired stoichiometry of Na1−x Li x NbO3 (x = 0.12) for one sample, and deliberately excess Li2CO3 of 1, 3, 5 mol% was added with stoichiometric Li0.12Na0.88NbO3 to prepare the other samples. The mixtures were ball-milled for 24 h in alcohol to obtain proper mixing and a surface-active fine powder. The finely mixed constituent powders were then solid-state-reacted at 900 °C for 5 h to get phase pure powder. The size distribution of the phase pure powder was measured using laser diffraction particle size analyzer (Beckman Coulter, LS 13 320). The powders were then uniaxially pressed at 150 MPa into pellets. To obtain the optimized sintering condition, the stoichiometric LNN-12 green pellets were sintered at different temperatures (1050–1250 °C) for varied durations (1, 2, 4, and 6 h) in air. The phase purity of the stoichiometric LNN-12 and excess Li2CO3-added samples was checked by powder X-ray diffraction (XRD) at room temperature using X-ray diffractometer (X’Pert, PANalytical) with Cu-Kα radiation. For this, the final sintered pellets were finely crushed and annealed at 500 °C for 12 h to reduce residual stress. The microstructural analysis of polished and thermally etched sintered pellets was carried out using scanning electron microscope (SEM) (JEOL-JSM 7600F). The average grain size, G, was estimated using a linear intercept method for the data collected over 200–300 grains, sampled at different locations of each of the SEM micrographs, using a standard image analysis software (ImageJ, NIH, USA). The dielectric constant and electrical conductivity of the samples were recorded using Impedance Analyzer (Alpha High Resolution, Novocontrol) in the frequency range 0.1–106 Hz over the temperatures 50–500 °C.

Results and discussion

Phase analysis

Figure 1 shows XRD pattern of stoichiometric LNN-12 and excess Li2CO3 (1, 3, 5 mol%)-added samples. Within the detectable limit of XRD, stoichiometric LNN-12 and compositions with 1 and 3 mol% Li2CO3-added samples show a characteristics pure perovskite structure. The peaks in the XRD pattern are indexed to a major orthorhombic phase (JCPDS# 033-1270) and is characterized by splitting of (202)/(080) peaks. However, in addition to pure perovskite phase, for x = 5 mol% Li2CO3-added sample, a minute secondary phase was detected (indicated as * in Fig. 1). This phase could be assigned to Li3NbO4 with a cubic structure (JCPDS# 016-0459) which is possibly a low melting liquid phase and thus contributes to abnormal grain growth. The lattice parameters and the unit cell volumes for all the compositions are too calculated and are listed in Table 1.

Figure 1
figure 1

XRD patterns of stoichiometric LNN-12 and 1, 3, 5 mol% Li2CO3-added LNN-12 samples. Minor secondary phase Li3NbO4 is marked as (asterisks) in excess Li2CO3-added samples

Full size image
Table 1 Variation in lattice parameters and unit cell volume of stoichiometric LNN-12 and 1, 3, 5 mol% Li2CO3-added LNN-12 samples
Full size table

Particle morphology and size distribution

The SEM micrographs of calcined (SSR) powder of stoichiometric LNN-12 and excess Li2CO3 (1, 3, 5 mol%)-added samples are shown in Fig. 2. In order to confirm the particle size distribution (PSD) that obtained from SEM data, as shown in inset of Fig. 2a–d, the particle size analysis of the calcined powder was also carried out using laser diffraction particle size analyzer, and their distributions are shown in Fig. 2e. For stoichiometric LNN-12, a uniform and narrow particle size distribution has been observed with an average particle size ~0.5 μm; however, Li2CO3-added samples show a non-uniform and broadened distribution with slightly higher average particle size (>0.5 μm). The non-uniformity exists in the particle size distribution in Li2CO3-added samples as few coarser particles could be due to secondary crystallization where coarser particles are formed coalescing smaller particles. The observed coarser particles act as nucleation centers for grain growth during sintering [10].

Figure 2
figure 2

a–d SEM micrographs of calcined (SSR) powder, and e particle size distribution of stoichiometric LNN-12 and excess Li2CO3 (1, 3, 5 mol%)-added samples measured using laser diffraction particle size analyzer

Full size image

Microstructural development in stoichiometric sample

The SEM images of polished and thermally etched surfaces of the LNN-12 discs, sintered at different temperatures (1050–1250 °C) with a dwelling duration of 2 h, are shown in Fig. 3. For all the samples, uniform and well-developed grains with no exaggerated grain growth are seen. The average grain size and relative density obtained from these samples are compiled in Table 2. The graphs in Fig. 4a show the sintering temperature-dependent densification and grain growth in the LNN-12 compacts. In the temperature range 1050–1190 °C, the sintered density increases from 90.5 to 95.9 % and the average grain size from 1.1 to 2.3 μm (Fig. 4a). The densification process, as expected, has proceeded through progressive elimination of the connected open porosity, and is primarily via the lattice and the grain boundary diffusion paths [13]. At higher sintering temperatures (T > 1190 °C), though further densification is marginal, an accelerated grain growth is clearly seen (Fig. 4b). Reverting to Fig. 3 and Table 2, as the temperature is progressively increased from 1190 to 1220 °C, and finally to 1250 °C, the grain size has also increased from 2.3 (Fig. 3d) to 3.1 μm (Fig. 3e) to 4.2 μm (Fig. 1f). The relative increase in grain size is substantial, whereas only a nominal increase from 95.9 to 96.8 % (Fig. 2a) is seen in the overall sintered density. A careful observation of the grain size distribution histograms, given as insets in Fig. 3d–f, clearly shows that the smaller grains, in spite of not having many curved boundaries, are progressively merged into the larger grains. This is akin to Oswald ripening. Additionally, the residual dispersed open porosity in the low-temperature (≤1190 °C)-sintered sample indicates insufficient sintering parameters (time and temperature). It is well known that definite activation energy is required for the grain boundary movement, and if either the temperature or time is insufficient, it leads to discrete open pores. At higher sintering temperatures (≥1190 and ≤1250 °C), the open pores collapse resulting in a relative decrease in numbers of pores and an accelerated grain growth [14].

Figure 3
figure 3

SEM photomicrographs of the polished and thermally etched LNN-12 pellets sintered at a 1050 °C/2 h, b 1100 °C/2 h, c 1150 °C/2 h, d 1190 °C/2 h, e 1220 °C/2 h, and f 1250 °C/2 h, showing clean phase and progressively improved densification and grain size

Full size image
Table 2 Variation in the average grain size and relative density (%) of the LNN-12 ceramic samples sintered at different temperatures (1050–1250 °C) for 2 h
Full size table
Figure 4
figure 4

Variation in a grain size and relative density (%) with sintering temperatures (1050–1250 °C), b grain size with relative density (%) that obtained from (a). Variation in c grain size and d relative density (%) with different soaking durations (1–6 h) for LNN-12 ceramic sintered at 1200 and 1250 °C

Full size image

In the stoichiometric LNN-12 ceramic pellets, it is seen that at T > 1190 °C, the grain growth proceeds at a faster rate compared to the rate of densification. In order to distinguish the difference in the rate of grain growth between the earlier and the later stages of sintering, the samples have been systematically subjected to isothermal sintering at two different temperatures, viz., 1200 °C (a little higher than 1190 °C) and 1250 °C, and for dwelling durations of 1, 2, 4, and 6 h. The average grain size variation with soaking duration, for 1200 and 1250 °C, is shown in Fig. 4c and also listed in Table 3. The average grain size in the samples sintered at 1200 °C has increased from 2.3 to 3.8 μm for the samples sintered at 1250 °C increased marginally from 3.9 to 4.5 μm, while the soaking duration has been increased from 1 to 6 h. This indicates that at the early stage of sintering, the rate of grain growth is higher, which is primarily led by grain boundary diffusion. However, this progressively slows down with soaking time and reaches a minimum at the final grain size, at the specified sintering temperature. Figure 4d shows the variation in sintered density with soaking durations for the samples sintered at 1200 and 1250 °C. A highest value of sintered density of 96.9 % was obtained for the samples sintered at 1250 °C for 2 h. A decrease in density for soaking duration higher than 2 h has been observed for all samples which may be attributed to loss of alkali elements at elevated temperature and long soaking durations. An optimized sintering condition therefore has been decided on the basis of highest measured sintered density (96.9 %) and moderate grain size (4.2 μm) and found to be 1250 °C for 2 h.

Table 3 Variation in average grain size with soaking durations for isothermal sintering at 1200 and 1250 °C of LNN-12 ceramic
Full size table

Microstructural development in excess Li2CO3-added samples

Figure 5a–c shows the SEM images of polished and thermally etched surfaces of the stoichiometric LNN-12 and excess Li2CO3 (1, 3 mol%)-added pellets, sintered at the same optimized sintering condition (1250 °C for 2 h) for the sake of comparing. The 5 mol% Li2CO3-added sample did not sustain the sintering temperature of 1250 °C and melted. This has been further substantiated by differential scanning calorimetry (DSC) experiments carried out at a heating rate of 10 °C/min for stoichiometric LNN-12 and excess Li2CO3 (1, 3, 5 mol%)-added powder samples as indicated in Fig. 6. From the inset of Fig. 6, it can be seen that there is no endothermic peak for stoichiometric LNN-12 and 1–3 mol% Li2CO3-added samples, while a weak peak ~1135 °C for 5 mol% Li2CO3-added sample has been observed. This confirms the melting of the 5 mol% Li2CO3-added sample and therefore not considered for further investigation. The effect of excess Li2CO3 addition has a significant effect on the microstructure development as shown in Fig. 5. For stoichiometric LNN-12 sample, a uniform and normal grain growth with an average grain size of 4.2 μm (as discussed in “Particle morphology and size distribution” section) is observed as shown in Fig. 5a. However, for Li2CO3-added sample, grain size distributions are no longer uniform, but consist of a large number of small grains and a small fraction of large size grains as shown in the inset of the grain size distribution histogram in Fig. 5b, c. For 1 mol% excess Li2CO3-added sample, the average grain size remains same (~4.2 μm); however, larger grains reach size up to 15 μm as marked by the arrow in Fig. 5b. Further, in 3 mol% Li2CO3-added sample, the average grain size increases from 4.2 to 4.9 μm, while larger grains of size ~23 μm (inset of Fig. 5c) have been observed. These signify that addition of excess Li2CO3 promotes an abnormal grain growth (AGG). The observation of AGG is believed to originate from the formation of low melting liquid phase at the non-stoichiometric (Li-rich/Na-poor) grain boundaries, which enhances dramatically the atomic mobility and resulting in AGG [15, 16].

Figure 5
figure 5

SEM photomicrographs of the polished and thermally etched pellets sintered at 1250 °C/2 h. a Stoichiometric LNN-12, b 1 mol% Li2CO3-added, and c 3 mol% Li2CO3-added samples

Full size image
Figure 6
figure 6

DSC curves for stoichiometric LNN-12, along with excess Li2CO3 (1, 3, 5 mol%)-added powder samples

Full size image

Electrical properties

Temperature-dependent permittivity in the range 50–500 °C measured at selected frequencies is depicted in Fig. 7. Both the real (ε′) and imaginary (ε″) parts of permittivity show two peaks, representing a ferroelectric to paraelectric transition at T m and ferroelectric to ferroelectric transition at T ff which appears at the lower temperature side of permittivity maxima. Broadening of the permittivity peaks in ε′ (T) are observed near both the transition temperatures (T m , T ff); however, frequency-dependent dispersion in T m has not been observed. This could be related to diffuse ferroelectric phase transition observed in pure LNN system [11] nevertheless of different characteristics of classical relaxor ferroelectrics. The usual ground for the appearance of diffuse phase transition is structural frustration induced due to an equivalent site in the lattice that is occupied by different types of cations. In accession, in the present study, both the transition peaks (T m , T ff) are set up in very close proximity, which also contribute to a diffuse phase transition. In stoichiometric LNN-12 and 1 % Li2CO3-added sample, position of transition peaks is almost at the same temperatures, T m  = 325, 316 °C and T ff = 217, 220 °C, respectively. This is in agreement with the minute change in unit cell volume as listed in Table 1. However, well-separated peaks at T m  = 420 °C and T ff = 195 °C for 3 % Li2CO3-added sample were observed. The frequency dependence (0.01–1 MHz) of real and imaginary parts of dielectric permittivity (ε′ and ε″, respectively) on a log–log scale in the temperature range 50–500 °C (with step of 50 °C) is depicted in Fig. 8. Both the parameters (ε′ and ε″, respectively) show strong low-frequency dispersion in dielectric permittivity (LFDD), characteristic of the high-temperature region due to a dominant DC conduction process. The electrical conductivity in Li2CO3-added non-stoichiometric LNN-12 might be associated with defects and their migration [17]. In stoichiometric LNN-12, the conduction is mainly electronic as the intrinsic lithium-ion (Li+) concentration is small compared to electronic contribution, whereas that in the non-stoichiometric material is ionic [18]. Samples with excess Li2CO3 added, diffusion of Li+ enhances greatly which results in higher electrical conductivity.

Figure 7
figure 7

Frequency dependence (10−1–106 Hz) of real (ε′) and imaginary (ε″) parts of permittivity measured at different temperatures (50–500 °C with step of 50 °C) for stoichiometric LNN-12 (0 mol%) and excess Li2CO3 (1, 3 mol%)-added samples

Full size image
Figure 8
figure 8

Temperature dependence (50–500 °C) of real (ε′) and imaginary (ε″) parts of permittivity measured at different frequencies (102–106 Hz) for stoichiometric LNN-12 (0 mol%) and excess Li2CO3 (1, 3 mol%)-added samples

Full size image

So, to look into the conductivity behavior in Li2CO3-added LNN-12, temperature and frequency dependence of electrical conductivity was studied. The temperature-dependent DC conductivity response of stoichiometric LNN-12 and Li2CO3-added samples is presented in Fig. 9b–d. The data represent the Arrhenius relation σ dc = σ 0 exp(−E a /kT), where σ 0 is a pre-exponential factor and E a , k, and T represent the activation energy for conduction, Boltzmann’s constant, and the absolute temperature, respectively. The activation energy for conduction (E a) was estimated by the slope of the ln σ versus 1000/T plot. The DC conductivity was determined from the total conductivity following Jonscher’s power law: σ total = σ dc + Aω n, where second part of the equation is associated with the frequency-dependent conductivity [19]. The representative graph of frequency-dependent total electrical conductivity at various temperatures for the stoichiometric LNN-12 sample is presented in Fig. 9a. For all the samples, two different regions of conductivity have been observed mentioned as region I and region II associated with ferroelectric and paraelectric phase, respectively, as shown in Fig. 9b–d. In the low-temperature region I (ferroelectric phase), extrinsic conduction resulting from impurity ions present in the lattice dominates, whereas intrinsic ionic conduction resulting from the movement of the component ions occurs in the high-temperature paraelectric phase [20, 21]. In stoichiometric LNN-12 with no excess Li+, the dominating conduction is mainly electronic and shows activation energy of 0.68 eV in the low-temperature region. With the increase in Li+ concentration, a deviation from stoichiometry occurs and conduction mechanism changes to ionic. At higher temperature, the values of activation energy obtained (~1 eV) are normally associated with thermally driven intrinsic ionic conduction for stoichiometric LNN-12 and extrinsic ionic conduction for non-stoichiometric LNN-12 due to mobility of Li+ [17, 22].

Figure 9
figure 9

a Representative graph for frequency-dependent electrical conductivity at various temperatures for stoichiometric LNN-12, b–d plot of ln σ dc versus 1000/T for stoichiometric LNN-12 and 1, 3 mol% Li2CO3-added samples, respectively. Activation energies are obtained from the linear fit

Full size image

In a comparative study, the variation in dielectric constant, loss factor, and electrical conductivity with measuring frequency at room temperature is shown in Fig. 10. The room-temperature parameter values at 1 kHz such as dielectric constant are found to increase from 160 to 475 °C, as Li2CO3 is increased from stoichiometric to 1 % and from 160 to 690 °C for 3 % Li2CO3-added sample (Fig. 10a). The loss factor has increased from 0.02 to 0.03 for stoichiometric to 1 % and finally reaches 0.04 for 3 % Li2CO3-added sample (Fig. 10b). The electrical conductivity value for stoichiometric LNN-12 is found to be ~10−12 (Ω cm)−1 and increases up to two orders of magnitude for 3 % Li2CO3-added sample (Fig. 10c), which is found the more influential effect of excess Li2CO3 addition.

Figure 10
figure 10

Frequency dependence of a dielectric constant, b loss factor, and c electrical conductivity of stoichiometric LNN-12 and 1, 3 mol% Li2CO3-added LNN-12 samples measured at room temperature

Full size image

Conclusions

The microstructural development of stoichiometric LNN-12 has been investigated systematically, and optimized sintering condition has been obtained on the basis of highest achieved density with moderate grain size. Excess Li2CO3 has been added to compensate possible alkali evaporation during sintering at elevated temperature. An abnormal grain growth has been observed for higher Li2CO3-added sample, and it is concluded that addition of excess Li2CO3 in starting material should be restricted up to 3 mol%. The room-temperature electrical conductivity has been found to be two orders of magnitude higher than that of stoichiometric LNN-12 because of the uncompensated excess Li2CO3 in the lattice. The room-temperature dielectric constant increases almost 4.5 times (160–690), whereas a change in loss factor from 0.02 to 0.04 has been observed when excess Li2CO3 is added up to 3 mol%. Therefore, it can be understood from the results that there is a marked effect of Li2CO3 addition on microstructure and dielectric properties. Hence, for a specific application, it is important to maintain proper stoichiometry in ceramics for microstructure and the development of appropriate processing schedule to optimize its electrical properties.