Phase-boundary engineering

In pursuit of a high piezoelectric constant d ₃₃, a number of recent research activities have focused on modifying phase-transition temperatures to construct a phase boundary at room temperature. It is known that KNN-based materials undergo a sequential phase transition from rhombohedral (R) to orthorhombic (O) to tetragonal (T) to cubic (C) phases with increasing temperature. For pure KNN ceramics, a low piezoelectric coefficient (d ₃₃ ∼ 80 pC/N) was reported due to the absence of phase coexistence (O-T, R-O, or R-T). To improve the piezoelectric response, it is essential to shift the phase-transition temperatures (rhombohedral to orthorhombic phase-transition temperature T _R-O or orthorhombic to tetragonal phase-transition temperature T _O-T) to room temperature to construct desired phase structures.^{Reference Li, Wang, Zhu, Cheng and Yao6} A large amount of research shows that substituting with elements such as Li, Ag, Zr, Hf, Sb, and Ta, can decrease T _O-T to room temperature, resulting in the formation of an O-T phase boundary and relatively good piezoelectricity (d ₃₃ = 200∼416 pC/N).^{Reference Wu, Xiao and Zhu8} On the other hand, an R-O phase boundary can be shifted to room temperature, but is accompanied by poor piezoelectricity (d ₃₃ ≤ 250 pC/N).^{Reference Wu, Xiao and Zhu8} It is suggested that by selecting the appropriate substitutes, both T _O-T and T _R-O can be simultaneously shifted to room temperature, omitting the intermediate transition to the O phase and forming an R-T phase boundary instead.^{Reference Wu, Xiao and Zhu64} As a result, ultrahigh piezoelectricity (d ₃₃ = 490∼570 pC/N) can be achieved,^{Reference Xu, Li, Lv, Wu, Zhang, Xiao and Zhu20,Reference Wang, Bando, Katsumata, Inaguma, Taniguchi and Itoh52,Reference Zushi, Ariizumi, Kojima, Wang and Bando53} as shown in Figure 1a. It should be mentioned that the reported high value of 700 pC/N is not included here because it was achieved for textured KNN ceramics,^{Reference Li, Zhai, Shen, Zhang, Li, Zhu and Zhang61} while the values shown in Figure 1 were measured using non-textured samples.

Figure 1a presents the historical (2004–2017) evolution of d ₃₃ in KNN-based ceramics with different phase boundaries, where the highest d ₃₃ values reported in each year are presented. The ceramics with the O-T phase boundary were widely investigated at first. In 2004, Saito et al. reported a large piezoelectric constant (d ₃₃ ∼ 416 pC/N) in textured KNN-based ceramics by shifting the O-T phase boundary.^{Reference Saito, Takao, Tani, Nonoyama, Takatori, Homma, Nagaya and Nakamura3} Since then, the O-T phase boundary constructed by composition modification, such as with ion substitutions (Li, Ta, and Sb) or the addition of second compounds (LiNbO₃, LiTaO₃, and LiSbO₃), is considered to be an effective method to enhance piezoelectricity.^{Reference Wu, Xiao and Zhu8} Wu et al. successfully prepared a series of high-performance KNN-based ceramics by developing the R-T phase boundary. In 2013, a large d ₃₃ of 425 pC/N was obtained in (K,Na)NbO₃-LiSbO₃-BaZrO₃ (KNN-LS-BZ) ceramics with the R-T phase boundary, which is the first reported value comparable to the KNN-based textured ceramics designed by Saito et al. Since 2014, successive improvements in piezoelectricity (d ₃₃ = 460∼570 pC/N) have been achieved in non-textured KNN with the R-T phase boundary.^{Reference Xu, Li, Lv, Wu, Zhang, Xiao and Zhu20,Reference Qin, Zhang, Yao, Lu and Zhang26,Reference Wang, Bando, Katsumata, Inaguma, Taniguchi and Itoh52,Reference Zushi, Ariizumi, Kojima, Wang and Bando53} Figure 1b summarizes data for d ₃₃ as a function of Curie temperature, T _C, for KNN-based ceramics with different phase boundaries. The construction of the R-T phase boundary was anticipated to improve the piezoelectricity of KNN-based ceramics.

As previously discussed, the phase structure is of paramount importance for KNN-based materials, while the coexistence of O-T or R-T phases is of greatest concern. Lab-scale x-ray diffraction can be used to distinguish the O from the T phase, which makes it easy to identify the O-T phase boundary.^{Reference Wang and Li65} However, detailed structural analysis of the phase constitution around the R-T phase boundary remains to be achieved, resulting in controversy over the existence of the R-T phase boundary.

Temperature stability

KNN-based piezoceramics have shown electromechanical properties equivalent to those of PZT in the intermediate and low-temperature regime (25°C–200°C).^{Reference Xu, Li, Lv, Wu, Zhang, Xiao and Zhu20,Reference Li, Zhai, Shen, Zhang, Li, Zhu and Zhang61} Temperature-dependent properties of piezoelectrics are important as well because there are circumstances where electromechanical coupling at a high temperature is demanded (e.g., piezoelectric fuel engine injectors, ultrasonic gas velocity sensor applications, and high-power ultrasonic transducers). However, unlike the temperature-insensitive morphotropic phase boundary (MPB) in PZT, phase boundaries in KNN solid solution are tilted, known as polymorphic phase boundaries^{Reference Dai, Zhang and Zhou66} and are temperature sensitive. High piezoelectric activities are always combined with the phase coexistence region to increase the likelihood of obtaining temperature-insensitive performance in a single phase. For many years, KNN materials were associated with a tradeoff between electromechanical performance and thermal stability.

Only recently has the issue of thermal stability been partly resolved by phase structure engineering (e.g., constructing a diffuse phase transition). Wang et al. reported a temperature-insensitive field-induced strain, which varies less than 10% when the temperature varies from room temperature to 175°C, demonstrating the great potential of lead-free piezoceramics for temperature-insensitive piezoelectric actuators in KNN piezoceramics.^{Reference Wang, Yao, Jo, Gobeljic, Shvartsman, Lupascu, Li and Rödel67} Yao et al. revealed the occurrence of distinct O-T phase coexistence over a wide temperature range (0°C–100°C) under an electric field stimulus.^{Reference Yao, Wang, Jo, Webber, Comyn, Ding, Xu, Cheng, Zheng, Hou and Li68} In situ synchrotron x-ray diffraction experiments combined with dielectric constant measurements revealed that the phase transition between O and T phases is extended, constituting a diffused phase transition. Wang et al. constructed an R-T phase boundary, where d ₃₃ remained above 200 pm/V up to 200°C by codoping titanate and zirconate into KNN.^{Reference Wang, Wang, Yao, Li, Schader, Webber, Jo and Rödel55} Zhang et al. showed how the field-induced strain could be further enhanced while maintaining temperature insensitivity by doping manganese into the material.^{Reference Zhang, Wang, Du, Dai, Sun, Li, Hu, Thong, Zhao, Xi, Yue and Li56,Reference Zhang, Wang, Zhou, Zhou, Chu, Lv, Wu and Li69}

Requirement and opportunity

The fundamental requirements for applications of KNN include not only the piezoelectric coefficient d ₃₃, but also other properties (e.g., electromechanical coupling factor [k] and mechanical quality factor [Q _m]). Similar to PZT, these properties can be modified by using either donor or acceptor doping in KNN to achieve a “softening” or “hardening” effect, respectively.^{Reference Hagh, Jadidian, Ashbahian and Safari70,Reference Vendrell, Garcia, Bril, Ochoa, Mestres and Dezanneau71} The tunable properties have turned the requirements into opportunities. For example, the properties can be purposely fitted for applications in the high-frequency mode, such as ultrasonic cleaning, and medical transducers, where the figure of merit is the product of k ^{Reference Egerton and Dillon2} and Q _m.^{Reference Li, Kakemoto, Hoshina and Tsurumi72}

Another important issue concerning the application of KNN materials is cost. It has long been argued that Ta is expensive, thus KNN materials containing Ta have less potential for commercialization; however, this problem has currently been solved since compositions without Ta still show properties comparable to PZT.^{Reference Li, Wang, Zhu, Cheng and Yao6,Reference Matsuoka, Kozuka, Kitamura, Yamada, Kurahashi, Yamazaki and Ohbayashi73} Nevertheless, the production cost of KNN may still not be as inexpensive as that of PZT, mainly due to the use of Nb.^{Reference Koruza, Bell, Frömling, Webber, Wang and Rödel5}

Interestingly, the cost burden of KNN can be compensated during device manufacture by the multilayer ceramic structure. By stacking more ceramic layers with smaller thicknesses, multilayer piezoelectric devices enable the enhancement of overall displacement and an effective reduction in driving voltage.^{Reference Gao, Guo, Zhang and Randall74} Expensive precious metals as the inner electrode materials in PZT multilayer structures are the only choice since they can withstand the cofiring process with ceramics at high temperatures.^{Reference Koruza, Bell, Frömling, Webber, Wang and Rödel5,Reference Gao, Guo, Zhang and Randall74} Encouragingly, KNN and the main doping elements are thermodynamically compatible with base metals such as Ni and Cu. Kobayashi et al. proved the possibility of cofiring Ni inner electrodes in KNN-LiF ceramics.^{Reference Kobayashi, Doshida, Mizuno and Randall75} Hayashi et al. confirmed the reliability of multilayer piezoelectric ceramics with Ni inner electrodes under many conditions (e.g., stability with respect to high and low temperatures, high humidity, thermal shock).^{Reference Hayashi, Kawada, Kimura, Nakai, Tabata, Shiratsuyu, Nada and Takagi76} Gao et al. demonstrated the capability of cofiring inner Cu electrodes with KNN piezoceramics through precisely controlling experimental conditions.^{Reference Gao, Ko, Guo, Hennig and Randall77} Thus, the thermodynamic compatibility of KNN ceramics with base metal electrodes provides a cost advantage over PZT for multilayer structures, and is therefore one of the incentives to push lead-free materials toward applications.

Remaining challenges

Several challenges still remain. KNN-based ceramics are notorious for their poor sinterability and the inability to control the microstructure ever since their discovery in the late 1950s.^{Reference Egerton and Dillon2} Processing-related problems have been recognized as the main reasons preventing their wide application.^{Reference Zhang, Xia and Shrout9}

Generally, sintering of KNN ceramics is characterized by a narrow sintering range of only a few tens of degrees centigrade close to the solidus temperature of 1140°C, as shown in Figure 2.^{Reference Malič, Koruza, Hreščak, Bernard, Wang, Fisher and Benčan7,Reference Acker, Kungl and Hoffmann78,Reference Li, Zhen, Zhang, Zhang and Wang79} Also, it is difficult to obtain high-density KNN piezoceramics without additives at low-sintering temperatures. The addition of LiNbO₃, LiTaO₃, or LiSbO₃ can lower the sintering temperature of KNN, but their influence on the grain-size distribution differs.^{Reference Saito, Takao, Tani, Nonoyama, Takatori, Homma, Nagaya and Nakamura3} Inferring from the literature, it seems that such solid solutions require less strict control of the sintering process to reach high densities as compared to KNN. In contrast to the clear role of PbO in PZT, the benefits of using excess alkalis in KNN are controversial; for example, some researchers suggest that excess alkalis can promote densification,^{Reference Bomlai, Wichianrat, Muensit and Milne80} while the others believe that it will hinder densification.^{Reference Acker, Kungl and Hoffmann78} Liquid-phase sintering aids such as Cu compounds,^{Reference Matsubara, Yamaguchi, Kikuta and Hirano81} (K_5.4Cu_1.3Ta₁₀O₂₉) (KCT),^{Reference Matsubara, Yamaguchi, Kikuta and Hirano82,Reference Park, Jang, Tran, Tai and Lee83} and germanate-based^{Reference Bernard, Benčan, Rojac, Holc, Malič and Kosec84} glassy phases have been effectively used to promote densification and, in some cases, lower the sintering temperature.

Figure 2.

Sintering curves (shrinkage versus temperature, T) for a lead-free piezoelectric (K_0.5Na_0.5)NbO₃ (KNN) and Pb(Zr_0.53Ti_0.47)_0.98Nb_0.02O₃ (PZT), which is added for comparison (dashed line). The samples were heated at 5 K/min in air. While pure KNN is characterized by an extremely narrow shrinkage interval just below the solidus temperature (T _solidus), PZT exhibits a much broader shrinkage interval. Note: l ₀, diameter of pellet; Δl, change in diameter of pellet.^{Reference Malič, Koruza, Hreščak, Bernard, Wang, Fisher and Benčan7}

Altering the sintering atmosphere is a feasible approach to promote densification and modulate the microstructure. KNN ceramics sintered in reducing atmospheres (75% N₂–25% H₂ or H₂) reached higher relative densities than their counterparts sintered in air or oxygen, and have uniform and fine microstructures, which is related to an increase in the oxygen vacancy concentration and changes in the grain-boundary structure.^{Reference Fisher and Kang85} Feasibility of sintering in reducing atmospheres enabled realizing KNN-based multilayers with base-metal electrodes, as previously mentioned.^{Reference Kawada, Kimura, Higuchi and Takagi62,Reference Kobayashi, Doshida, Mizuno and Randall75,Reference Gao, Ko, Guo, Hennig and Randall77}

It has been shown that the vapor pressures of alkalines over their respective niobates are a few orders of magnitude lower than the vapor pressures of PbO or Bi over respective PZT or alkali bismuth titanate materials,^{Reference Malič, Koruza, Hreščak, Bernard, Wang, Fisher and Benčan7,Reference Popovič, Bencze, Koruza and Malič86} meaning that packing powder is not needed to preserve the stoichiometry if the sintering time is not extended. On the other hand, losses of alkalines may occur in earlier stages of processing, including during the manipulation of humidity-sensitive alkaline reagents and solid-state synthesis.^{Reference Hagh, Jadidian and Safari87}

The quest for functional properties of lead-free piezoelectric ceramics, which should be comparable to those of lead-based complex perovskites, has been oriented toward compositions on phase boundaries, which typically include a number of cations on the same crystallographic sites. In the case of KNN, the choice of niobium oxide reagent, orthorhombic or monoclinic, contributed to a nonhomogeneous solid solution.^{Reference Hreščak, Benc̃an, Rojac and Malic̃88} Similarly, different diffusion rates of sodium and potassium ions may contribute to local heterogeneities.^{Reference Malič, Jenko, Holc, Hrovat and Kosec89} Nonhomogeneous distribution of Nb and Ta in KNN-based prepared by conventional solid-state reaction contributed to irreproducible functional properties. This problem could be overcome by prereacting B-site ions (e.g., Nb₂O₅ and Ta₂O₅ first form a solid solution)^{Reference Wang, Damjanovic, Klein, Hollenstein and Setter90} or utilizing individual binary compounds,^{Reference Hagh, Jadidian and Safari87} or alternatively, by mechanochemical activation of the reagent mixture.^{Reference Rojac, Benčan, Uršič, Malič and Kosec91} In other words, each new lead-free material formulation with reliable and reproducible functional properties is also a challenge from the processing point of view.

Summary

In this article, we have reviewed milestone achievements in the development of KNN-based materials as promising lead-free piezoceramics. Research on KNN has been proven successful, and the exploration is entering a new application-oriented stage. The first batch of commercial devices using KNN materials is expected on the market soon, while extension of these materials to broad applications still requires unremitting efforts.