Giant room-temperature electrostrictive coefficients in lead-free relaxor ferroelectric ceramics by compositional tuning

Electrostriction, where electric-field-induced strain (S₃₃ in the out-of-plane direction) is proportional to the square of out-of-plane polarization (P₃) (i.e. S₃₃ = Q₃₃ · P₃², Q₃₃ stands for an electrostriction coefficient), is of a practical interest in aspect of device applications to micro-electromechanical systems (MEMS).¹ Unlike piezoelectricity with a linear electromechanical coefficient, an electrostrictive effect has been known to possess a large variety of technological benefits/advantages for realizing novel MEMS devices, such as little hysteretic characteristics in the electromechanical responses, good thermal stability, a short operational time, and no need of a poling process.^1–4 In conventional relaxor ferroelectrics (FEs) belonging to the family of lead-based complex oxides, high S₃₃ of 0.1% and a large Q₃₃ value of 0.02 m⁴/C² have been realized at room temperature.^4–7 However, due to strict regulations of lead-based compounds for global environment,³ a comparable electrostrictive response should be achieved in lead-free compounds for their eco-friendly and sustainable utilization. Furthermore, for potential applications to actual devices, it may be also accessible at room temperature.

Bi_0.5Na_0.5TiO₃ (BNT)-based relaxor FEs are an excellent class of lead-free oxide materials for realizing high electrostrictive responses.^4,8–16 We notice that an ultrahigh electrostrictive coefficient is attainable in the vicinity of a thermotropic phase boundary which is a thermally-induced boundary between two competing phases [i.e., ergodic relaxor (ER) and non-ergodic relaxor (NR) phases in BNT-based materials].^16,17 In an ergodic phase above a transition temperature (T_F-R), a relaxor FE state, with macroscopic paraelectricity, where the crystallographic symmetry is metrically cubic, is reversibly converted to an FE state by the application of an external electric field. In contrast, for a non-ergodic phase below T_F-R, the FE state induced by a sufficiently large electric field is irreversible.^16,17 Near the phase boundary between the ER and NR phases, large net strain and relatively small remnant polarization values under an external electric field are feasible with little polarization-electric field hysteresis, which enables a nearly pure electrostrictive response with a large Q₃₃ value (∼S₃₃/P₃²) (see Fig. S1 of the supplementary material).⁸ Since the T_F-R of this NR-to-ER phase transition is commonly higher than room temperature [e.g., the T_F-R of Bi_0.5(Na_0.78K_0.22)_0.5TiO₃ (BNKT) ceramics is about 150 °C],¹⁸ tuning the T_F-R towards room temperature is essential for us to achieve a large room-temperature electrostrictive coefficient in BNT-based relaxor FEs.⁹ 

Recently, it has been expected that the transition temperature of T_F-R in BNT-based ceramics can be reduced by modifying the chemical composition of the parent compound.^9,13,18–20 Besides, the T_F–R can be tuned by the external electric field.²¹ Note that bulk BNKT undergoes a NR-to-ER phase transition around ∼150 °C and the degree of FE order is also susceptible to slight changes in the cation content. It is, therefore, possible that the transition temperature in BNKT ceramics can be manipulated, when a room-temperature tetragonal phase is converted to a metrically cubic phase by stoichiometry control.^13,18–20 It is further interesting that an exceptionally high electrostrictive coefficient could appear at room temperature, if the BNKT composite undergoes a NR-to-ER phase transition nearby room temperature.⁹ 

In this work, compositional tuning of T_F-R is implemented to design a new BNT-based material with an ultrahigh room-temperature electrostrictive coefficient. To assess our synthetic approach, Bi-based Bi_0.5(Na_0.78K_0.22)_0.5TiO₃ (BNKT) and Bi(Ni_0.5Ti_0.5)O₃ (BNiT) are selected as a base compound and a chemical modifier, respectively.^21–23 Note that multi-charge valency is accessible in these transition-metal Ni and Ti cations and also, their ionic radiuses are sensitive to the oxidation states.²⁴ By substituting the BNiT for the BNKT, the initial tetragonal structure would change with a pseudocubic (metrically cubic) distortion, probably due to the different ionic radius between competing B-site cations.^18,21,23,24 Thus, the transition temperature T_F-R between non-ergodic and ergodic phases can be shifted towards room temperature.²⁵ By measuring temperature-dependent dielectric responses in the BNKT-BNiT composite, a shift of the T_F-R by compositional modification is clearly demonstrated. A giant room-temperature electrostrictive coefficient is also recorded in the BNKT-BNiT ceramic with a particular BNiT content. The reduction of the T_F-R to room temperature allows large strain available, albeit polarization is relatively small, leading to an enhancement in the room-temperature electrostrictive coefficient.

High-quality (1 − x)BNKT-xBNiT ceramics, where x varies from 0.0 to 0.100, were synthesized using solid state reaction method (see the supplementary material for detail). All (1 − x)BNKT-xBNiT samples were densely synthesized with no impurity/pore on the sample surface implying that the chemical stoichiometry is spatially homogeneous and the incorporated elements are well coalesced (see Fig. S2 of the supplementary material).

To check the crystallinity of our (1 − x)BNKT-xBNiT compounds, we performed x-ray diffraction (XRD) θ-2θ scans of our samples in the 2θ range of 20°-80° (see Fig. S3 of the supplementary material), it was identified that the ceramic samples are a perovskite single phase without any secondary phase. Figure 1(a) of the XRD θ-2θ scans in the narrow range of 35°-50° shows that the crystallographic structure evolves, as the mole concentration of BNiT (x) increases. In the pure BNKT (x = 0.0) with a tetragonal symmetry, the Bragg (002) and (200) peaks around 46.5° are split due to the different out-of-plane d-spacings of c and a domains, whereas the multi-domain state of c + a is indistinguishable in the Bragg (111) peak around 40.0°. It is interesting that the observed splitting in the (002)/(200) peak disappears with the increasing x and finally, the peak becomes single when the x value is 0.040. In contrast, the (111) peak remains roughly unchanged with respect to all x values. Note that the spectral shapes/patterns of these (002)/(200) and (111) peaks are strongly dependent on the crystallographic symmetry of the BNT-based materials (see Fig. S4 of the supplementary material). When the tetragonal symmetry approaches a metrically cubic, the (002)/(200) and (111) peaks get closer to single without a peak separation. It follows that our (1 − x)BNKT-xBNiT ceramics undergo a tetragonal-to-metrically cubic structural transition with the increasing concentration of BNiT inducing changes in the spectra of the x-ray Bragg peaks. In (1 − x)PbTiO₃-xBNiT ceramics with divalent Ni²⁺ cation which disfavors the hybridization with an oxygen atom and thus, promotes the stability of a pseudocubic phase, a similar tetragonal-to-pseudocubic transition has been reported.^26,27 To confirm the oxidation state of Ni, x-ray photoelectron spectroscopy (XPS) analysis of (1 − x)BNKT-xBNiT (x = 0, 0.050, and 0.100) compounds was performed (see Fig. S5 of the supplementary material). The XPS spectra at Ni 2p_3/2 peak shows that divalent Ni²⁺ and metallic Ni states coexist at the binding energy of ∼855.08 and 852.58 eV, respectively. Thus, Ni cation with a charge valence of +2 (i.e., Ni²⁺) is solely substituted for quadrivalent Ti⁴⁺ in the BNKT compound.^28–31 

Temperature-dependent dielectric measurements in our (1 − x)BNKT-xBNiT ceramics reveal that the transition temperature T_F-R of a NR-to-ER transition is shifted towards room temperature, due to the tetragonal-to-pseudocubic phase transition. The temperature dependence of dielectric permittivity and loss (tan δ) was measured under the various ac frequencies of 1, 10, and 100 kHz [Fig. 1(b)]. It is evident that the transition temperature T_F-R (indicated by red dotted lines), where anomalies in both dielectric permittivity and loss appear, changes from 150 to 104 °C, as the x value increases from 0.0 to 0.015, respectively. With a further increase (≥0.040) in the BNiT concentration x, the T_F-R vanishes eventually. Since our dielectric measurements have been done in the temperature range of 30–400 °C, it is highly likely that the T_F-R is shifted below room temperature. Considering the fact that the structural symmetry of our (1 − x)BNKT-xBNiT ceramics is transformed from tetragonal to metrically cubic above x = 0.040, the movement of the T_F-R across room temperature would be closely related with this structural transition. Such shift of the T_F-R in dielectric dispersion behaviors by structural changes has been previously reported in other BNT-based relaxor FEs.^{11,18,20,32–34}

To get a further insight into the compositionally-induced shift of T_F-R, room-temperature polarization-electric field (P₃-E) and bipolar strain-electric field (S₃₃-E) hysteresis loops of our (1 − x)BNKT-xBNiT ceramics were measured as shown in Figs. 2(a) and 2(b), respectively. As the BNiT content x increases, three types of hysteretic behaviors are observed. For a pure BNKT (x = 0.0) which is structurally tetragonal and non-ergodic relaxor at room temperature, it exhibits typical FE hysteresis loops (square-like and butterfly-shaped for P₃-E and S₃₃-E curves, respectively). With the increase of x value (x = 0.040 and 0.050), pinched and sprout-shaped hysteresis loops emerge in the P₃-E and S₃₃-E characteristics due to the development of ergodicity, respectively. Note that the observed intermediate/non-equilibrium state in these hysteretic characteristics is the hallmark of the coexistence of ergodic and non-ergodic phases.^{11,17,20,34–36} In BNT-based relaxor FEs, it should also be noted that the emergence of ergodicity arises from a metrically cubic phase transition.^{11,18,20,32–34} It follows that the (1 − x)BNKT-xBNiT ceramics with the particular BNiT contents of 0.040 and 0.050 are in the vicinity of the NR-to-ER phase transition at room temperature. Thus, it is highly plausible that the transition temperature T_F-R is quite close to room temperature. With a more increment in the BNiT content (x ≥ 0.050), the hysteretic nature becomes much weaker resulting in slimmer and parabolic-like in the P₃-E and S₃₃-E curves, respectively. Namely, the ceramics with these BNiT compositions lies on a metrically cubic and typical ergodic relaxor phase. In unipolar S₃₃-E measurements, a similar tendency in the hysteretic shape is observed, too (see Fig. S6 of the supplementary material).

To gain a room-temperature electrostrictive coefficient Q₃₃ in our (1 − x)BNKT-xBNiT ceramics, strain S₃₃ is plotted as a function of polarization P₃ [Fig. 3(a)]. For this aim, the out-of-plane S₃₃ and P₃ were simultaneously measured at room temperature under a triangular electric-field wave with the frequency of 1 Hz using the aixACCT Ceramic Multilayer Actuator (aixCMA) test bench.¹¹ It should be emphasized that the S₃₃ is linearly proportional to the P₃² with no hysteretic property for the only two samples with the BNiT concentrations x = 0.040 and 0.050 in the proximity of a NR-to-ER phase transition. For other ceramics with different BNiT concentrations (x = 0.0, 0.015, and 0.030), the hysteretic nature still remains in the S₃₃-P₃² curves, which means that the eletrostrictive responses are not pure [Fig. 3(a)]. Considering the fact that Q₃₃ = S₃₃/P₃², for the only two samples with the BNiT concentrations of x = 0.040 and 0.050, the high electrostrictive coefficients of 0.025 and 0.030 m⁴/C² are attributed to the large electromechanical strain of 0.36% and 0.30% and the relatively small polarization of 10 and 8 μC/cm², respectively (see Fig. S7 of the supplementary material). For the ceramics with higher BNiT contents (0.060, 0.075, and 0.100), the Q₃₃ would become smaller due to the small strain values (0.1% ∼ 0.2%) (see Fig. S6 of the supplementary material). It follows that the sizeable increase of the Q₃₃ in the (1 − x)BNKT-xBNiT ceramics (x = 0.040 and 0.050) originates from the enhancement of the S₃₃ and the reduction of the P₃.

BNKT-BNiT compounds could be a promising candidate offering the possibility for new electromechanical devices due to the giant room-temperature electrostrictive coefficient Q₃₃. A comparison of the room-temperature Q₃₃ values with those in other lead-based and lead-free BNT-based electrostrictive materials shows that the highest value is achieved in our (1 − x)BNKT-xBNiT ceramics [Fig. 3(b)].^{4,8,10–14,34,37–42} For the (1 − x)BNKT-xBNiT sample with x = 0.050, the Q₃₃ value increases up to 0.030 m⁴/C². It is worthy of noting the reported Q₃₃ values of lead-based materials such as PMN (∼0.023 m⁴/C²), PZT (∼0.021 m⁴/C²), PLZT (∼0.015 m⁴/C²), PMN crystal (∼0.025 m⁴/C²), and PZN crystal (∼0.023 m⁴/C²).^37–42 We also found that the attained Q₃₃ values are insensitive to temperature variations (i.e., stable for thermal fluctuations) (Fig. 4). In the temperature range between 30 and 150 °C, the S₃₃-P₃² curves were measured in two (1 − x)BNKT-xBNiT ceramics (x = 0.040 and 0.050). It is evident that there is no hysteretic features in the S₃₃-P₃² curves except for that measured at 150 °C [Figs. 4(a) and 4(b)] and further, the derived electrostrictive coefficients Q₃₃ are independent of the temperature with finite values close to 0.025 and 0.030 m⁴/C² (x = 0.040 and 0.050, respectively) [Fig. 4(c)]. Thus, it is highly likely that our BNKT-BNiT compounds are used for the potential applications as micro-devices (e.g., actuator) where high electrostrictive coefficient and robust thermal durability are required around room temperature.

In summary, a large room-temperature electrostrictive coefficient of 0.030 m⁴/C² is attained in BNT-based relaxor ferroelectrics, as the direct product of the compositionally-driven shift of the thermotropic phase boundary between non-ergodic and ergodic phases. With the giant electrostrictive responses, we firmly believe that the lead-free BNKT-BNiT compounds are applicable for the bio-compatible utilizations, enabling the realization of novel MEMS devices with a high performance at room temperature.

See supplementary material for the additional detail regarding experimental procedure, schematic diagram of electrostrictive coefficient, microstructural analysis, XRD study, schematic representation of crystallographic phases, XPS study, unipolar S-E loop, and summary of phase diagram.

This work was supported by Brain Pool Program through the Korean Federation of Science and Technology Societies (KOFST) grant funded by the Ministry of Science, ICT and Future Planning (No. 172S-1-3-1900). The authors would also like to acknowledge the National Research Foundation of Korea (NRF) grants funded by the Korea government (Ministry of Education) (No. 2017R1D1A1B03028614) and (No. 2017R1D1A1B03036032) and Priority Research Centers Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (No. 2009-0093818).