Advances in lead-free high-temperature dielectric materials for ceramic capacitor application

: Ceramic capacitors with upper operating temperatures far beyond 200°C are essential for high-temperature electronics used in deep oil drilling, aviation, automotive industry and so on. Recent advances in existing lead-free dielectrics for potential high-temperature capacitor applications are reviewed and grouped into three categories according to the parent component of the solid solution. Their desirable temperature stabilities were summarised comprehensively. However, there are still some limitations in the current research, such as achieving low loss in a wide temperature range and maintaining stable dielectric properties with different frequencies or at different voltages. Furthermore, the successful implementation of multilayer ceramic capacitors is one of the biggest challenges, which will have far-reaching impacts on the realisation of high-temperature capacitor application in the future.


Introduction
Capacitors, as a kind of indispensable passive component, are widely used in every electronic equipment because they can serve a host of functions, such as snubbing, filtering, direct current (dc) blocking, coupling, decoupling and so on [1][2][3]. Currently, the market for ceramic capacitors is dominated by multilayer ceramic capacitors (MLCCs) [2]. A typical architecture of MLCC, shown in Fig. 1, is comprised of dielectric layers, internal electrodes, and end terminal. Dielectric layers and internal electrodes are alternately stacked and the internal electrodes are connected in parallel, which can enhance the volumetric efficiency of capacitors significantly and meet requirements of miniaturisation and integration of electronic component better [2,4]. Nevertheless, with multifold developments of advanced electronics in the past few decades, the demands on high-performance functional electric components are growing prominently and rapidly. Simultaneously, the requirements on capacitor materials imposed by industries are increasing as well. Nowadays, more and more researchers have been aware of the necessity and importance of materials suitable for high-temperature electronics (HTE), since electronic equipment in many industry fields is required to operate properly at a high temperature beyond 200°C under harsh environmental conditions [5,6].
For example, in the oil and gas industries, electronics and sensors that monitored the down-hole drilling require a higher upper limit of working temperature. If the geothermal gradient is >25°C/km, the higher stable operating temperature of drilling devices, the deeper can be explored [5]. In aviation and aerospace, transforming the traditional centralised architecture of engine monitoring and control system into a distributed control scheme can reduce the complexity of the interconnections and save hundreds of pounds of aircraft weight [5,7], while many electronics have to place nearby the engine where temperature is at 200-300°C. In the automotive industry, the anti-lock brake system sensors on wheels are required to operate at a temperature range of 150-250°C and ambient temperature ranges for cylinder pressure sensing are 200-300°C [5,8]. In addition, with the advancement of silicon-on-insulator technology [9,10] and the development of wide band-gap semiconductor materials [11][12][13], the operating temperature of semiconductors has been extended to 200°C or >300°C. Capacitors used with these circuits in high-temperature electrics are required to withstand the same harsh conditions as well [14]. Therefore, investigations of high-temperature capacitor dielectrics, which can operate stably beyond 200°C or even 300°C, are extremely essential for new technological applications.
To be used reliably in such applications at extremely high temperatures, ceramic capacitor dielectrics must fulfil the following requirements [15,16]: (i) The permittivity should be high with low loss in the applied temperature range, which is synonymous with the high volumetric efficiency of single layer dielectric. (ii) High resistivity and high resistant-capacitor (RC) constant at elevated temperature are needed to reduce the leakage current of charged dielectrics and ensure the performance of capacitor devices. (iii) It is vital to acquire temperature-insensitive dielectric permittivity within the variation of 15% over a wide temperature and frequency range.
There are various types of ceramic materials that can be used to fabricate capacitors, while their dielectric properties are greatly different. In general, commercially available ceramic capacitor dielectrics are basically categorised into three classes [2].
Class I dielectrics are usually considered as temperaturecompensating ceramic materials. They present the least variable temperature, voltage, and frequency properties, while their dielectric permittivity is very low, often on the magnitude order of 10 1 or 10 2 .
Class II dielectrics are high permittivity materials (1000-20,000) based on ferroelectric ceramics with a dissipation factor usually in the range of 0.01-0.03. They normally present nonlinear temperature and frequency dependence on their dielectric permittivity and exhibit a predictable change with time and voltage.
Class III dielectrics are the basis for barrier layer capacitors. For such types of dielectrics, each grain in the dielectric consists of a conductive core and a thin insulating shell, or a barrier layer. This unique structure often results in extremely high capacitance but low operating voltage (<25 V) [2].
Among all these classifications, modified Class II dielectrics (ferroelectric ceramics) are inclined to candidates for a hightemperature dielectric with relatively high stable permittivity and high resistivity. However, their temperature stability is still a big weakness. This property is often evaluated by the temperature coefficient of capacitance (TCC). The formula is described as follows: where C base is the capacitance at a base temperature and C T represents the capacitance at any temperature in the whole operating range. However, it is worthy to note that the base temperature is not specified. In most cases, base temperature is usually chosen at 25°C, while many researchers tried to select another higher temperature as a standard temperature, such as 150°C or 200°C for those high-temperature dielectrics with temperature limit reaching 250°C and beyond, which seems to be more rational.
The internationally used standard EIA-198-1-F-2002, designated by Electronic Industries Alliance (EIA), made a temperature characteristic code by two letters and one figure to evaluate temperature stability of capacitors. For Class II dielectrics, the first letter and the figure indicate the minimum and the maximum working temperatures, respectively; while the second letter represents the maximum change allowed in capacitance, namely, the permitted TCC. Some representative temperature characteristic codes are listed in Table 1. Taking the code X7R as an example, the maximum change in capacitance of these dielectrics varies not more than ±15% from −55°C to 125°C. However, traditional Class II ferroelectric dielectrics are far from enough for high-temperature capacitors under harsh conditions mentioned above. The upper operating temperatures of the most widely used commercial Class II capacitors (EIA-X7R and EIA-X8R) are not more than 150°C, while the operating temperature for HTE needs to reach 200-300°C. The lack of reliable hightemperature capacitor hinders the manufacture of high-temperature electronic devices. Therefore, there is an urgent need to develop new type Class II dielectric materials that meet the needs of hightemperature use.
Compared with a large number of ferroelectric materials, leadbased perovskite materials often have high dielectric permittivity, as shown in Fig. 2 [17][18][19]. Hence, lead-containing perovskite dielectrics are more likely to be used in the manufacture of ceramic capacitors with large capacitance than tungsten bronze structured and bismuth layer structured compounds. Actually, the investigations for practical applications of lead-based relaxor capacitor materials can date back to 1980s, when Pb(Fe 1/3 W 2/3 )O 3 , Pb(Mg 1/3 Nb 2/3 )O 3 , and Pb(Zn 1/3 Nb 2/3 )O 3 -based dielectrics have been intensively studied. These dielectric materials often exhibit large permittivity (10,000-25,000) with Y5V or Z5U characteristics, while their splendid permittivity is hard to maintain well over 100°C [4]. Further researchers found that Bi-Pb based materials, such as BiScO 3 -PbTiO 3 or BiScO 3 -Pb(Mg 1/3 Nb 2/3 )O 3 -PbTiO 3 [2,17], presented greater temperature stability at high temperature. Nevertheless, considering the environmental concern as well as government legislation against the usage of lead and other hazardous substances, such as the directive 2002/95/EC on the restriction of hazardous substances from the European Union [20], lead-based materials cannot become mainstream in the commercial capacitor market, and their application prospects are still uncertain [2].  In recent years, tremendous efforts have been devoted to the development of competitive lead-free perovskite materials for potential high-temperature application. Among them, multi-ferroic BiFeO 3 with high Curie temperature (T c ∼ 825°C) was investigated widely as one of the most popular materials [21,22]. However, the large leakage current and difficulty in sintering process are their main weaknesses [21]. Extensive researches attempted various methods to improve the sintering behaviours and ferroelectric properties of samples, such as an optimised processing route [23], the addition of other members [24,25], and ion substitutions [22,26,27]. Also, some materials for capacitors were also studied [28][29][30], i.e. high permittivity varying <30% up to 200°C was achieved in BiFeO 3 -SrTiO 3 solid solution [29]. However, it should pay attention that the existence of BiFeO 3 in these systems often leads to large dielectric loss, resulting in self-heating of the capacitor, which limits its practical application, especially in the high temperature range. In general, candidates based on BaTiO 3 or Bi 0.5 Na 0.5 TiO 3 are more competitive. Therefore, in this review, we mainly present perspectives on recent research into new lead-free dielectric ceramics based on BaTiO 3 , Bi 0.5 Na 0.5 TiO 3 and some other components, whose operating temperatures are well-beyond the upper limit of traditional BaTiO 3 -based Class II dielectrics.

BaTiO 3 -based materials
Barium titanate (BaTiO 3 , BT) is a kind of typical normal ferroelectric with perovskite structure. Since its discovery during world war II, researchers have paid a great deal of attention to it [2]. Up to now, BT is still the base material of choice for ceramic dielectrics because of its high permittivity and stable dielectric properties.
There are three phase transitions for pure BaTiO 3 ceramic when cooled from high temperature: cubic to tetragonal at ∼120°C, tetragonal to orthorhombic at 0°C, and orthorhombic to rhombohedral at −90°C, as shown in Fig. 3 [2]. The sharp dielectric anomalies at phase transition temperatures bring about highly and strongly temperature-sensitive dielectric properties and a serious deterioration in permittivity, which is far from ideal for a capacitor dielectric. Additionally, due to the low Curie temperature (T c ∼ 120°C), BT is hard to be competent in the high-temperature application. Hence, there are extensive researches to modify its dielectric behaviour. For example, shift and suppress the dielectric peaks by adding dopants [4] or altering the average grain size [31]. However, only these modifications are difficult to meet the temperature-independent capacitance variation over 200°C.
Effective ways of modification, generally used in improving the temperature-dependent dielectric response of BT-based ceramics, can be basically divided into two categories. One is through the construction of the 'core-shell' structure by utilisation of various doping elements which are partially dissolved in BT grains. In recent year, our group systematically investigated the effect of Y 2 O 3 addition on non-reducible BT-based ceramics and resolved the 'core-shell' structure with respect to Y 2 O 3 doping [32]. As can be seen in Fig. 4 [32], the 'core-shell' grain consists of the core region, pure BT with a ferroelectric tetragonal structure, surrounded by the non-ferroelectric shell which contains some additive elements. The dielectric constant of two phase coexistence could be described by the Lichtenecker formula given as where V c and V s are volume fraction of the core and shell regions; ɛ is the total dielectric permittivity; ɛ c and ɛ s represent the dielectric permittivity of the core and shell regions, respectively. According to the Lichtenecker formula, the ferroelectric core and nonferroelectric shell are responsible to the high-temperature region and low-temperature region of dielectric response, respectively, which resulting in a broadening dielectric peak and enhanced temperature stability. Except for the volumetric ratio of core-toshell, coherency of interface and local stress field in 'core-shell' ceramics can also influence temperature-dependence of dielectric properties. In the case of small ionic radius rare-earth-substituted samples (Dy 2 O 3 , Ho 2 O 3 , Er 2 O 3 , and Yb 2 O 3 ), which predominantly occupied the B-site, the appropriate doping amount can shift T c of BT-based samples towards a higher temperature due to the increased tensile stress resulting from the lattice mismatch of the core-shell interface. Some other effects of microstructural features of the 'core-shell' structure on functional properties are listed in Table 2 summarised by Acosta et al. [33]. The other useful approach is making BT-based ceramics as relaxors by adding large amounts of components with a similar structure. It is known to us that most relaxor dielectrics are characterised by a high permittivity and a diffuse phase transition [19,34], which is desired for the temperature-stable capacitor. The diffuse phase transition behaviour is closely related to the peculiar structure of polar nanoregions (PNRs) surrounded by a non-polar matrix on the scale of nanometres [13]. These inhomogeneities are due partly to the coexistence of two or more cations with different charges and chemical bonding characteristics in A-site or B-site of the ABO 3 perovskite structure, such as the classic relaxor ferroelectrics Pb(Mg 1/3 Nb 2/3 )O 3 and Bi 1/2 Na 1/2 TiO 3 . Hence, forming solid solutions with different charges or polarities is also effective to flatten the dielectric permittivity stability. An example for Bi(Mg 2/3 Nb 1/3 )O 3 (BMN) modified BT [35] is shown in Fig. 5. With BMN increasing, the peak at T c was depressed and disappears eventually, bringing about flat ɛ r -T curves. Nowadays, more and more researchers attempt to obtain high-temperature BT-based ceramics beyond 300°C in this way. Nevertheless, the significant frequency dispersion dependence permittivity of relaxors, which has a negative influence on frequency stability at low temperature, still needs to be improved.
Search for new compounds with dielectric temperature stability of BT has resulted in the discovery of a number of systems. A brief summary is as follows.

BaTiO 3 -BiMeO 3 solid solutions
BiMeO 3 compounds, where Me can be Sc 3+ , Al 3+ , (Mg 0.5 Ti 0.5 ) 3+ , (Mg 0.5 Zr 0.5 ) 3+ , and so on, are considered to be a kind of attractive end member for ferroelectric ceramics owing to their broad diffused dielectric behaviour as well as relatively high permittivity. It has been shown that BiMeO 3 additions improved the temperature-dependent dielectric behaviour of BaTiO 3 available, which can be ascribed to the weakly coupled polarisation in nanopolar regions originated from the substitution of Ba 2+ and Ti 4+ by Bi 3+ and Me 3+ . The weakly coupled polarisation limited the ability to obtain long-range dipole ordering, thereby making dielectric anomalies more suppressed and diffusive [1,17,36]. Therefore, solid solutions of BaTiO 3 -BiMeO 3 , which are highly promising for high-temperature applications, have recently received a great deal of interest.
The binary solid solution BaTiO 3 -BiScO 3 (BT-BS) as an attractive high-temperature dielectric system was studied by many researchers. In Tinberg and Trolier-McKinstry's work [37], a pure BT-BS perovskite phase was achieved in the films when the composition of BS was in 20-45 mol.%. Following this work, Ogihara et al. [36] successfully prepared a stable perovskite phase in (1−x)BT-xBS bulk samples and demonstrated the weakly coupled polarisation behaviour in PNRs. With small amounts of BS (x = 0.03-0.1), an interesting core-shell structure was observed in the BT-BS system. The shell regions enriched in BS and the core regions of BT were responsible for the lower-and highertemperature dielectric anomalies, respectively. However, with BiScO 3 increasing from x = 0.1 to x = 0.4, there is a gradual change from proper ferroelectric behaviour to highly diffusive and dispersive relaxor characteristics, resulting in temperature-stable dielectric properties. When x = 0.3, a high, temperatureindependent permittivity (1000 from 0 to 300°C) and a highelectrical resistivity of ∼10 12 Ω cm at 250°C was obtained, reported by Ogihara et al. [1], which is superior to traditional commercial X7R or X8R capacitors at high-temperature regions. Additionally, Lim et al. [38] also studied a kind of hightemperature capacitor material on the base of 0.6BT-0.4BS by modification of Bi 1/2 K 1/2 TiO 3 , which exhibited high-dielectric permittivity (ɛ r ∼ 1700 at room temperature) and low-dielectric loss over the temperature range from 100 to 300°C, with flat temperature coefficients of permittivity (TCɛr), TCɛ r = −800 ppm/°C. Furthermore, an electrical resistivity of 10 9 Ω cm and RC constant of 0.8 s at 300°C were reported and the energy density can reach 4.0 J/cm 3 at 220 kV/cm. Except for BiScO 3 , BiAlO 3 was also taken as end-member to form solid solutions for capacitor applications. In 2010, our group designed the BaTiO 3 -BiAlO 3 (BT-BA) binary system and investigated its structure description and dielectric behaviours [39,40]. The temperature-dependent relative permittivity of 0.9BT-0.1BA composition depicts a typical relaxor behaviour with noticeable frequency dispersion and strongly diffusive and broad dielectric maximum, which suggests the positive effect of BA on BT ceramic for temperature stability. Later, Liu et al. [41] synthesised the binary (1−x)BT-xBA ceramic system by solid-state reaction and sol-gel processing routes, respectively. They found that the temperature stability in permittivity was significantly enhanced with BA contents increasing. Besides, a wider operating temperature window (TCC ≤ ± 15%) was obtained in the ceramics fabricated by the sol-gel method. Among all the compositions, 0.7BT-0.3BA possessed the best dielectric property, with a moderate dielectric permittivity (ɛ r = 660) and low-dielectric loss (tan δ = 1.2%) at room temperature, 1 kHz, and an ultra-broad temperature range (−55°C to 440°C, ΔC/C 25°C ≤ ±15%). However, ceramics using the starting powders produced by conventional solid reaction methods failed to achieve this temperature stability level. Also, an investigation of (1−x)BaTiO 3 -xBi(Zn 0.5 Ti 0. 5 [42] suggested an attractive lead-free system with dielectric temperature stability as well. They found a decreased tetragonality and decreased dielectric transition temperature with a more diffused transition peak as the BZnT content increased >10 mol.%. Based on this work, further investigations of modified BT-BZnT for better temperature stability were carried out [43][44][45][46]. Raengthon and co-workers found that the incorporation of BiInO 3 , BiScO 3 , and NaNbO 3 into the BT-BZnT system was effective in flattening the ɛ r -T curves [43][44][45]. These additives lead to higher substitute levels and weaken the coupling of dipoles in nanopolar regions. Also, as a result, the improved temperature stabilities of the dielectric response were obtained in all these systems [36]. In addition to these typical solid solution systems of BaTiO 3 -BiMeO 3 , many such compounds also show promising dielectric properties at 150 or 200°C and above, including BT-Bi(Zn 0.5 Zr 0.5 )O 3 [47], BT-Bi(Mg 0.5 Zr 0.5 )O 3 [48,49], BT-Bi(Mg 0.5 Ti 0.5 )O 3 [50,51], BT-Bi(Mg 2/3 Nb 1/3 )O 3 [35,52], BT-Bi(Mg 2/3 Ta 1/3 )O 3 [53], BT-Bi(Li 1/3 Zr 2/3 )O 3 [54], and their dielectric properties are partially shown in Table 3.

Ba 0.8 Ca 0.2 TiO 3 -BiMeO 3 solid solutions
Although BT-based dielectrics modified by Bi-based perovskite materials displayed a flat and broad temperature-dependent dielectric response, their temperature stability can be more appealing by introducing Ca 2+ further.
In 2013, Zeb and Milne [74] presented their work of Ba 0.8 Ca 0.2 TiO 3 in solid solution with Bi(Mg 0.5 Ti 0.5 )O 3 (BCT-BMT). BaTiO 3 ceramic extends the temperature range of the tetragonal perovskite phase by introducing Ca 2+ , which exhibits a great stability at elevated temperature, better than the unmodified BT-BMT. For the composition of 0.5BCT-0.5BMT, a moderate and near-flat relative permittivity (ɛ r = 800 ± 10%) spanning from 45 to 550°C was obtained, with tan δ ≤ 0.025 over the temperature range 100-430°C. Better dielectric properties can be acquired in solid solutions with a higher concentration of BMT, which has a permittivity of a value of 1000 within the variation of 15% from 80 to 500°C and small dielectric loss (tan δ ≤ 0.02) from 160 to 500°C. The optimum resistivity is of the order of 10 10 Ω m at 250°C and 10 7 Ωm at 400°C.
However, the work mentioned above is just a beginning for investigations of BCT-BMT high-temperature dielectrics. Zeb et al. [55] further developed an ultra-wide temperature stable system based on the 0.45BCT-0.55BMT modified by NaNbO 3 (NN). Despite the incorporation of NN decreased overall relative permittivity, the temperature at maximum permittivity (T m ) are decreased too. Compared with unmodified 0.45BCT-0.55BMT, the T m of 0.45BCT-0.25BMT-0.3NN can reach from 160°C to −70°C [47], which is significant for dielectric stability over a wide temperature range. For this composition, the values of permittivity are about 550 varying not more than ±15% across the temperature range of −60°C to 300°C in combination with tan δ ≤ 0.02. Also, the optimum dielectric properties were obtained by adding 20 mol.% NaNbO 3 , with ɛ r = 600 ± 15% from −70°C to 500°C. Meanwhile, the dc resistivity for all compositions was on the magnitude order of 10 10 Ωm at 250°C, and 10 7 Ωm at 400°C. Later, Chen et al. [56] [57]. Of particular significance that an excellent dc resistivity of 10 9 Ω m at a temperature of 300°C was experimentally reported for 0.7Ba 0.8 Ca 0.2 TiO 3 -0.3Bi(Zn 0.5 Ti 0.5 )O 3 , which possessed high-temperature stability up to 400°C with ɛ r ∼ 1000 [58]. Their detailed properties are also listed in Table 3.

Other systems based on BaTiO 3
Except for the BaTiO 3 -BiMeO 3 solid solutions and their modification, some other BT-based ceramics can operate stably beyond 200°C likewise.
Yao et al. [59] successfully developed the capacitor dielectrics based on 0.9BaTiO 3 -0.1Bi 0.5 Na 0.5 TiO 3 (0.9BT-0.1BNT) with 2.0 mol.% Nb 2 O 5 , which satisfy the EIA-X9R specification (−55 to 200°C, TCC ≤ ±15%) with a permittivity of 1640 and a dielectric loss of 0.02 at 25°C. The good temperature stability was partly attributed to the inhomogeneous distribution of Nb 5+ from the grain boundary to the interior, namely, the core-shell structure, which is supported by transmission electron microscopy (TEM) pictures with Energy dispersive X-ray spectroscopy (EDS) data. Based on this work, many researchers attempted to modify BT-BNT-Nb 2 O 5 by introducing additional doping elements. Sun et al. investigated the role of 3d elements (Fe/Co/Ni/Mn) in the BT-BNT-Nb 2 O 5 system for capacitor application [79][80][81]. The incorporation of 3d acceptors, such as Ni and Co, can stabilise the core-shell structure by affecting the reaction between Nb and BT-BNT, which is crucial for the high dielectric temperature stability and most of these systems could meet EIA-X8R or X9R criteria. Li and Zhang [60] also analysed the role of doping CoO played in a BT-BNT-Nb 2 O 5 system with glass flux and CeO 2 for the realisation of dielectric stability in the wide temperature range. A higher permittivity was obtained with a nearly flat response from temperatures of −50 to 350°C.
In an earlier work reported by Li et al. [82], further enhanced the temperature stability of dielectric properties in 0.9BT-0.1BNT-Nb 2 O 5 can be achieved by adding MgO. The 1.5 mol.% MgOdoped composition can meet ΔC/C 20°C varying 6% over the temperature range from −55 to >200°C. In the study of Zhang et al. [61], ZnO was doped into the 0.85BT-0.15BNT-Nb 2 O 5 matrix, which realised the temperature-stable high relative permittivity over an ultra-broad temperature range. It is interesting that the overall permittivity increases first and then decrease with a higher concentration of ZnO. The increase of dielectric permittivity may be due to the increasing concentration of associated defect pairs. The decrease of permittivity can be attributed to the limited motion of the domain wall when the ZnO content is beyond the solubility limit. For all specimens, the sample doped with 5 wt.% ZnO has ɛ r = 2404 ± 13% over a wide temperature range from −55°C to 375°C. However, their insulation resistance is only 2.1 × 10 10 Ω m at room temperature, which is not enough for reliability and availability in high-temperature applications.
A subsequent study by Li et al. [62] investigated the effect of doping Pr 6 O 11 in 0.85BT-0.15BNT-Nb 2 O 5 ceramics. The substitution of Pr in Ba-site and Ti-site can decrease the spontaneous polarisation of BaTiO 3 and then weakened the ferroelectricity. Since the polarisation of the BT system mainly originated from the deviation of Ti 4+ from the central site of the octahedron, the substitution of Pr 3+ for Ti 4+ prevented the distortion of the octahedron, thereby reducing the polarisation. In the case that Pr 3+ substitutes into Ba-site, smaller Pr 3+ could decrease the unit cell volume and make the movement of Ti 4+ ion more difficult, therefore weakened the ferroelectricity. With 0.6 mol.% Pr 6 O 11 addition, temperature-dependence of capacitance can be <15% in the range of −55°C to >300°C with permittivity of 1608 at 25°C. However, the loss factor at high temperature is high (tan δ > 0.05 over 100°C), which is still needed to be reduced further.
In the case of other works on doping effects of BT-based perovskite dielectric materials, it was found that the addition of acceptors, such as MnO and Cr 2 O 3 , can improve the resistance, because conduction electrons can be trapped by the acceptors [4,83,84]. In addition, some researches revealed that the doping of amphoteric rare-earth ions (R 3+ ), such as Y 3+ , Dy 3+ ,and Ho 3+ , which could enter either the Ba-site or Ti-site, can realise a longer lifetime owing to the formation of the donor-acceptor complex [32,[85][86][87]. In the light of works by Hennings et al. [86,87], [(R′ Ti ) ⋅ (R Ba ⋅ )] acts as a strong barrier against the migration of oxygen vacancies even under a DC bias, thus, a slow degradation rate of the capacitor can finally be obtained. These researches may guide those BT-based high-temperature capacitor materials to gain more attractive performance. 8 IET Nanodielectr., 2018, Vol.
Owing to the distinctly different displacement of Na + and Bi 3+ ions as well as complex octahedral tilting, BNT ceramics exhibit a complicated crystal structure [89]. Figs. 6a and b present the BNT crystal structure of two models. At room temperature, BNT was identified as rhombohedral with R3c space group which has a − a − a − antiphase rotations of oxygen octahedron. However, highresolution single crystal x-ray diffraction measurements by Gorfman and Thomas [90] suggest that the structure of BNT would better be described by the monoclinic space group Cc. A similar opinion was also proposed by Aksel et al. [91] and Ma et al. [92].
Up to now, the phase structure of BNT is still controversial. Anyway, its unique structure has great effects on dielectric properties.
Over the past few years, numerous efforts have been made on BNT-based dielectrics, not only as a possible alternative to leadcontaining piezoceramics or electrostrictive materials due to their appealing electromechanical properties [93][94][95][96][97][98][99][100] but also as one of the most promising candidates of high-temperature dielectrics for capacitor applications. Fig. 7 presents permittivity and dielectric loss as a function of temperature measured at different frequencies for pure BNT ceramics. As a kind of relaxor ferroelectric, BNT ceramics have obvious diffuse phase transitions with two anomalies in temperature-dependent permittivity (T m1 ∼ 200°C, T m2 ∼ 320°C), which are considered to account for dielectric temperature stability [19,93]. However, the nature of these two anomalies for BNT-based solid solutions has long been disputed [101][102][103][104]. Jo et al. [105] proposed that the commonly observed two dielectric anomalies in BNT-based dielectrics can be attributed to the thermal evolution of PNRs with two different types of symmetries and their transformation into each other with temperature, which is widely accepted now. Hence, if the number density, correlation length, and the polarisability of PNRs could be modified by chemical substitutions, a flat and broad plateau in temperature-dependent permittivity would be obtained. In this way, temperature-insensitive dielectrics with high permittivity in potential capacitors application can be probably achieved.
Here, we briefly survey recent advances in BNT-based dielectrics for high-temperature capacitors, which are categorised into three classes.

Bi 0.5 Na 0.5 TiO 3 -based binary materials
In 2004, the lead-free binary solid solutions (1−x)Bi 0.5 Na 0.5 TiO 3 -xNaNbO 3 (BNT-NN) was investigated for x = 0-0.08 by Li et al. [106]. For the samples with x = 0.01-0.03, two distinct anomalies are also observed, accompanied by shifting to lower temperatures. At x = 0.08, only a broad dielectric plateau extending from ∼200 to 400°C can be found. It was also revealed that the dielectric relaxor behaviour of BNT ceramics can become clearer by adding a small amount of NN. On the basis of the study by Li et al. [53], Wu et al. [107] further investigated the effects of higher NN concentration on the relaxor and dielectric properties of BNT ceramics. As observed from the ɛ r -T curves, the 0.8BNT-0.2NN ceramic revealed a higher dielectric permittivity and more stable temperature-dielectric curve than other compositions. Therefore, they considered that 0.8BNT-0.2NN ceramics are suitable for applications in high-temperature multilayer ceramic capacitors. In addition, the decrease of polar R3c phase and the increase of weakly polar P4bm phase with NN increasing were revealed by the Rietveld refinement in the study of Xu et al. [63], which is critical for weakening the ferroelectricity and flattening the temperaturedependent dielectric response. For the samples with x = 0.25-0.35, TCC was found <11% in an ultra-wide temperature range of −60°C to 400°C with moderate permittivity (ɛ r ∼ 1000) and low-dielectric loss (tan δ ≤ 0.02, −60 to 150°C), promising for temperature-stable capacitor applications.
Actually, the binary solid solution BNT-NN was patented with a series of other BNT-based high-temperature dielectric materials by Bridger et al. in 2010 [64]. Except by utilisation of NN, Bridger et al. also took KTaO 3 as the end member of BNT-based binary solid solutions in their high-temperature dielectrics patent. Among all the compositions, 0.9Bi 0.5 Na 0.5 TiO 3 -0.1KTaO 3 exhibits a relatively flat permittivity from 80 to 260°C with a value of ∼2500 though permittivity outside this range rapidly decreases [64]. Besides the research of Bridger et al., König et al. [108] studied that the temperature-dependent dielectric anomalies and the overall permittivity values of Bi 0.5 Na 0.5 TiO 3 -KTaO 3 binary compositions. Their study showed that the formation of the solid solutions initially started with the formation of the BNT-rich and KTaO 3 -rich phases, while with increasing KTaO 3 , there was a fatter ɛ r -T curve with a reduction of the temperatures of the dielectric anomalies, and a decrease of the dielectric losses. For the composition of x = 0.2, ɛ rmid = 2000 ± 15%, in the temperature range 80-300°C (estimated by Zeb et al. [6]) with tan δ ≤ 0.02 over a narrower temperature range, from 200 to 300°C.

Bi 0.5 Na 0.5 TiO 3 -BaTiO 3 -based materials
The Bi 0.5 Na 0.5 TiO 3 -BaTiO 3 (BNT-BT) solid solution, firstly found by Takenaka et al. [109], was regarded as one of the brightest lead-free piezoelectric ceramics owing to its outstanding dielectric and piezoelectric properties [93]. Fig. 8 presents the schematic phase diagram of (1−x)BNT-xBT proposed by Ma et al. [110], which shows a series of complex structural transformations with the variation of the BT content from 0 to 12 mol.% (i.e. R3cto-P4bm transition at around x = 0.06 and P4bm-to-P4mm transition at around x = 0.11 at room temperature). It is noted that the binary system at morphotropic phase boundar (MPB) (x = 0.06-0.07) possesses highly enhanced dielectric properties and therefore many researchers used 0.94BNT-0.06BT or 0.93BNT-0.07BT as a based material system.
In 2007, Zhang et al. [111] attempted to modify 0.94BNT-0.06BT by using K 0.5 Na 0.5 NbO 3 (KNN), and the obtained leadfree piezoceramics deliver a giant strain (0.45%). What's more, it was discovered that materials with higher amounts of KNN exhibit high electrostrictive constants and temperature-insensitive fieldinduced strain [95]. The accumulated knowledge in the literature dictates that the origin of the large and temperature-insensitive strain response in the BNT-BT-KNN system was related to the formation of the 'non-polar' or 'weakly-polar' phase, which was identified as ergodic relaxor state now [97,112]. In other words, the size and polarizability of initially nanoscopically polar regions were changed by introducing KNN. Based on the investigations of Zhang et al. [111] and Jo et al. [112], Dittmer et al. [65] studied the ternary system, (1−x)(0.94BNT−0.06BT)-xKNN, for compositions 0.09 ≤ x ≤ 0.18, as the lead-free high-temperature dielectric. They observed a temperature-dependent permittivity plateau originated from the two anomalies in the base material and attained a high permittivity at 150°C (>2000) with a normalised permittivity ΔC/ C 150°C varying no >10% from 43 to 319°C. Meanwhile, the large resistivity on the order of magnitude of 10 8 Ω m and the RC constant of about 1 s at 300°C also demonstrated that these materials are promising as a dielectric for high-temperature applications.
To clarify the role played by KNN in affecting the electrical properties and relaxor characteristics in BNT-BT, impedance spectroscopy analysis (IS) was used by Zang et al. [113]. They found that the addition of KNN altered the stability of lowtemperature PNRs and high-temperature PNRs. Increased random local fields by increased substitution levels disrupt the correlations among PNRs, so the number density and polarisability of PNRs decreases, accompanied by decreasing freezing temperature (T f ) and Burns temperature (T B ), which leads to a suppression of both anomalies in ɛ r -T plots, as shown in Fig. 9.
Moreover, Acosta et al. [15] developed ultra-wide stable dielectrics with <15% of variation in the temperature range from −69 to 468°C based on the BNT-BT-KNN-CaZrO 3 system and found that the effect of incorporating CaZrO 3 into BNT-BT is similar to that with KNN, though the valences of Zr 4+ and Ti 4+ are the same. This phenomenon also can be attributed to increased random fields and local strain by increasing weakly ions Zr 4+ and Ca 2+ in BNT-BT, which reduce correlation lengths of PNRs and finally result in a stable permittivity over wide usage temperature.
Following this mechanism, more and more high-temperature capacitor dielectric systems based on BNT-BT have been developed. NaNbO 3 (NN) is a kind of critical end-member with very complicated phase transition sequence and structure characters and some of its peculiar properties were intensively studied by our groups [114,115]. Due to these intrinsic characteristics of NN, some researchers expected that the addition of NN may affect the coupling of PNRs and enhance the temperature stability at elevated temperature. Xu et al. [116] obtained high-temperature stable energy storage performance in 0.92BNT-0.08BT through weakening the polar phase by introducing a certain amount of NaNbO 3 . For the composition of 0.9(0.92BNT-0.08BT)-0.1NN, an energy storage density of 0.71 J/cm 3 at 7 kV/mm and a good temperature stability around 25-150°C were obtained. In 2017, our group reported the (1−x) (0.94BNT-0.06BT)-xNN system as capacitor dielectrics working far beyond 200°C, those with composition of x = 0.10 showed a high permittivity of >2700 at 1 kHz with a ΔC/C 150°C varying no more than 15% in a wide temperature range of 54-318°C [66]. Also, all specimens show the high-resistance value on the magnitude order of 10 9 Ω·m and the RC constants at the level of 10 2 s at 150°C. In addition, Na 0.73 Bi 0.09 NbO 3 (NBN) was also introduced into BNT-BT ceramics to alter the temperature stability of dielectric permittivity and the energy storage property by Xu et al. [67]. For the optimised sample of 0.8(0.92BNT-0.08BT)-0.2NBN, a wide temperature range of −12 to 300°C with TCC ≤ 15% was achieved. At the same time, 0.85(0.91BNT-0.09BT)-0.15NBN ternary ceramics present a permittivity of 1680, the dielectric loss of 0.004 at 150°C with Δɛ/ɛ 150°C varied no more than 10% up to 340°C. Recently, other modified systems, such as BNT-BT-NaTaO 3 [68], exhibit stable TCC ≤ ±15% spanning over 300°C as well, whose attractive property is also presented in Table 3.
In addition to the above excellent works, the doping roles of Nb-donor and Fe-acceptor in BNT-BT-KNN solid solutions were also investigated by Jo et al. [117]. It was found that the strain and polarisation characteristics were enhanced and suppressed by the acceptor and donor dopants, respectively. In other words, Nb-donor dopants can destroy the ferroelectric order, while the acceptor doping of Fe can lead to the stabilisation of a ferroelectric order due to the formation of (Fe′ Ti − V O ⋅ ⋅ ) ⋅ . The effect of (Fe′ Ti − V O ⋅ ⋅ ) ⋅ is a little different from the observation in doped PZT ceramics [83,[118][119][120].
Except for the modification of the BNT-BT system as described above, there is another interesting method to obtain a very high permittivity high-temperature capacitor with good temperature stability. By investigating the effect of Bi nonstoichiometry on Bi 0.5+x NaTiO 3 ceramics, Sung et al. observed abnormal high permittivity and dielectric loss of BNT with Bi deficiency at low frequency, which can be ascribed to the contribution from space charges [121]. Shi et al. [69] therefore designed the Bi-deficient solid solution series (Bi 0.5−y Na 0.5 ) 0.94−x Ba 0.06 (Bi 0.2 Sr 0.7−0.1 ) x TiO 3 (BNBT-BST) to obtain large permittivity with small deviation over a wide temperature range. A plateau in ɛ r -T plots with the significantly enhanced dielectric permittivity of ∼5000 by Bi-deficiency was achieved successfully. The composition of x = 0.26, y = 0.07 exhibits ɛ r = 4884 ± 1.5% over the temperature range from 73°C to 230°C and their working temperature spans from ∼50 to ∼270°C for a 10% tolerance. The very large value of RC time constant of 5.96 s at a temperature of 300°C was finally acquired.

Bi 0.5 Na 0.5 TiO 3 -Bi 0.5 K 0.5 TiO 3 -based materials
The (1−x)Bi 0.5 Na 0.5 TiO 3 -xBi 0.5 K 0.5 TiO 3 [(1−x)BNT-xBKT] system was first synthesised in 1996 by Elkechai et al. [122] and their enhanced piezoelectric properties were discovered in the composition close to the rhombohedral-tetragonal phase boundary. Later, Sasaki et al. [123] proposed that the MPB between rhombohedral BNT and tetragonal BKT was located near x = 0.2. In spite of both BNT-BT and BNT-BKT showing attractive performances at MPB, 0.8BNT-0.2BKT seem to have better piezoelectric properties than 0.94BNT-0.06BT, such as higher T d . In addition, since the nature of the low-temperature polar phase and high-temperature 'nonpolar' phase in BNT-BKT is very similar to those in the BNT-BT system, altering the number density of the polar phase and stabilising the 'non-polar' phase by an appropriate chemical modification, which was applied to the BNT-BT-based systems, may also realise the stability of temperature-dependent dielectric properties.
In 2010, Seifert et al. [124] investigated the ferroelectric behaviour of 0.8BNT-0.2BKT modified by KNN. It follows that the ferroelectric order is disrupted significantly with the addition of KNN and the destabilisation of the ferroelectric order is accompanied by an enhancement of a temperature-insensitive large signal d 33 up to 200°C. Anton et al. [125] reported that the effect of KNN is independent of the distance to the MPB. For instance, similar high strains and other dielectric properties can be achieved at and far off the MPB. Subsequently, Dittmer et al. [75] investigated the effect of KNN on 0.6BNT-0.4BKT solid solutions off the MPB after reporting BNT-BT-KNN based hightemperature dielectrics. Owing to the different results presented by temperature-dependent Ramen scattering and neutron diffraction, all those samples are also identified ergodic relaxors with two types of PNRs providing different relaxation mechanisms, the same as BNT and BNT-BT ceramics. The optimum properties for application as high-temperature dielectrics are obtained in a material with x = 0.15 at <10% variation of the relative permittivity of about 2100 between 54 and 400°C.
In terms of the overall dielectric temperature characteristics, BNT-BKT-KNN is competitive because it not only has high and stable permittivity but also low-dielectric loss over a wide temperature range, which surpassed many BT-based dielectrics. Moreover, their submicron scale grain sizes (500-600 nm) for all compositions are conducive to adapt to the trend of miniaturisation of electric devices. Nevertheless, their resistivity and RC constant at 300°C, which is at the level of 10 5 Ω m and 1 ms, respectively, are still needed to be further improved by continuous investigations [75].

Other materials
In addition to the popular BT-and BNT-based materials, Bi 0.5 K 0.5 TiO 3 and K 0.5 Na 0.5 NbO 3 are also competitive for hightemperature dielectrics. There are many similarities between these two kinds of materials. For instance, both of them are difficult to sinter because of the high volatility of the potassium component at sintering temperatures [126][127][128], while fine grains can be acquired after densification [129][130][131]. Besides, their wonderful ferroelectric and piezoelectric properties are attractive for many real applications [93,129,130,[132][133][134][135][136][137]. For examples, in the aspect of piezoelectric materials, a flat panel micro speaker was fabricated from co-fired multilayer KNN-based piezoceramics by Chu et al. [134], and BKT-and KNN-based multilayer actuators were also constructed by Nagata et al. [135] and Kim et al. [136]. In the aspect of capacitor materials, Du and co-workers developed a series of novel energy storage dielectrics based on KNN for application of pulsed power electronics [129,130,138,139].

Bi 0.5 K 0.5 TiO 3 -based materials
Bi 0.5 K 0.5 TiO 3 (BKT) was synthesised by Popper et al. in 1957 [140], but Buhrer [141] first proved its ferroelectricity [79]. Similar to BNT, BKT is also a kind of A-site complex perovskite relaxor ferroelectrics, while it is a perovskite structure of tetragonal symmetry at room temperature with a relatively high Curie temperature of 380°C and a second phase transition at around 300°C. In recent years, some researchers have focused their attention on BKT-based materials for high-temperature capacitor applications.
Kruea-In et al. [70] investigated the relaxor behaviour of (1−x)Bi 0.5 K 0.5 TiO 3 -xBiScO 3 [71]. With the increasing BZN content from 0 to 20 mol.%, the temperature of maximum permittivity, T m , decreased from 150 to 70°C. The sample with 20 mol.% BZN displayed a wide temperature range of stable permittivity, such that ɛ r = 805 ± 15% from −20 to 600°C (1 kHz) and dielectric loss were <0.02 from 50 to 450°C at 1 kHz. Additionally, values of dc resistivity were of the order of ∼10 9 Ωm at 300°C, which is promising in the context of developing new high-temperature capacitor materials.

K 0.5 Na 0.5 NbO 3 -based materials
Solid solution K 1−x Na x NbO 3 (KNN) between ferroelectric KNbO 3 and anti-ferroelectric NaNbO 3 has been thoroughly investigated. The existence of MPBs and polymorphic phase transitions is regarded as one of their advantages and many excellent works have been devoted to obtaining the very high piezoelectric response of KNN-based materials by constructing special phase boundaries [132,133,142,143]. On the other hand, the relatively high Curie temperature (T c ∼ 400°C) of KNN [93] also makes it a promising candidate for high-temperature applications.
Similar to BT, KNN with better temperature stability can also be achieved by forming relaxor ferroelectrics with the incorporation of BiMeO 3 . According to many investigations, the two phase transitions of orthorhombic-tetragonal and tetragonalcubic in K 0.5 N 0.5 NO 3 ceramic can be getting more and more diffused with a higher concentration of BiMeO 3 . Du et al. [144] developed the 0.96K 0.5 Na 0.5 NbO 3 -0.04BiScO 3 solid solution, which shows a broad permittivity maximum near 2500 from 100 to  [145]. An operational temperature window with 15% variation in permittivity from 150°C to 350°C was obtained at 100 kHz for the composition of x = 0.01.
Furthermore, a study of the K 0.5 Na 0.5 NbO 3 -LiTaO 3 -xBiScO 3 (KNN-LT-xBS) ternary system was also reported by Skidmore et al. [73] and Zhu et al. [146]. For small amounts of BS (<2 mol.%) modifications, a slight decrease in T c was observed. However, a diffuse ɛ r -T peak appeared when the composition of BS is >2 mol.%. The optimum property was obtained for adding 5 mol.% BS, which exhibited broad and stable relative permittivity curves as a function of temperature, ɛ r = 1100 ± 10% from ∼20 to 450°C, with tan δ < 0.025 from 200 to 350°C.
Besides the BiMeO 3 modified KNN-based materials, stable dielectric properties at a broad temperature usage range can be obtained by another way. Skidmore et al. [147] presented phasediagram for the K 0.5 Na 0.5 NbO 3 -LiTaO 3 (KNN-LT) solid solution series based on the combined results of temperature-variable X-ray powder diffraction and dielectric measurements. As shown in Fig. 10, with the increasing LT content, the temperature of orthorhombic/monoclinic to tetragonal phase transition was a shift to a lower temperature, while T c still remains around 430-440°C [147]. A flat dielectric response was therefore obtained. This approach of modification to achieving wide-ranging temperature stability and moderate ɛ r is impactful as well. For the composition of adding 7 mol.% LT, the operational temperature range expanded to −15°C to 300°C with ɛ r = 630-700 ± 15%. While a wider working temperature window range from −50 to 350°C with ɛ r = 480 ± 15% can be acquired with 10 mol.% LT addition [6]. In fact, for KNN materials, doping equipollent ions, such as Li + , Ta 5+ , as well as Sb 5+ , can also induce the formation of new phase boundaries by shifting phase transition point [142,[148][149][150]. For example, Li + , which replaces the A site, can decrease the orthorhombic-tetragonal phase transition temperature (T O-T ) and increase the Curie point (T c ), while Ta 5+ and Sb 5+ substituted in the B-site can decrease both T O-T and T c [142,[148][149][150]. These rules can be used for reference in the design of the KNN-based hightemperature material system.

Summary and perspective
High-temperature capacitor dielectric materials are receiving wide attention and research interest due to the urgent requirements or potential applications in some fields, such as deep oil drilling, aviation, automobiles and so on. From this review, we summarise and compare various lead-free perovskite capacitor dielectrics based on BT, BNT, BKT, and KNN, which display stable relative permittivity over wide temperature ranges with upper temperature limits beyond 200°C. Their temperature-related dielectric performances are presented in Fig. 11 and detailed properties are summarised in Table 3. However, in most studies, the permitted changes in dielectric constant are not able to be within the variation of 15% at low temperature (≤−55°C) and usually accompanied by poor frequency stability at low temperature, high loss factor and a low resistivity at high temperature. These obstacles that can be foreseen should be overcome by more efforts in order to facilitate the implementation of practical high-temperature capacitor materials further. In addition to the temperature stability, for capacitors, voltage stability and dielectric breakdown strength measurements and failure mechanisms are still critical, especially applied in a high-power pulse-forming network. In previous works by Randall et al. [17], the differences of Bi-Pb based hightemperature dielectrics in voltage saturation behaviour at various temperatures between room temperature and 350°C with an applied dc bias was reported. Actually, similar works are eagerly required for high-temperature lead-free capacitor materials as well.
Moreover, although more and more high-performance capacitor dielectrics with high and stable permittivity have been investigated and developed, many challenges still need to be faced with the transfer of these joint achievements into applications. The biggest challenge is their successful implementation into the MLCCs, which is necessary to enhance the volumetric efficiency but has totally different fabrication process compared with bulk samples [2,4]. Only in this way can we advance these high-temperature dielectrics into real applications. To make BNT-BT-KNN capacitor materials to be used in the form of thick films and multilayers, Zang et al. [16] further investigated the behaviour of this high-temperature dielectric material in the form of thick films in comparison with bulk ceramics. The 18KNN sample is suitable for applications ranging from 25 to 350°C despite the overall dielectric constant is decreased owing to the porosity and clamping effect. Subsequently, with the cooperation with Taiyo Yuden Company Ltd, Groh et al. [14] manufactured MLCCs made of the BNT-BT-KNN system by adding CuO to lower sintering temperature. The cross-sectional microstructure of MLCC, which is presented in Fig. 12, reveals the compact structure with no interdiffusion between the electrode and dielectric layers. The deviation of capacitance was smaller than ±10% between 40 and 225°C. Su et al. [151] fabricated 1210-sized MLCC components by the BNT-NN system based on the previous work by Xu et al. [63]. Capacitance variation ΔC/C 25°C was <11% in the temperature range of −55 to 205°C with ɛ r ∼ 850. Additionally, the tentative MLCC chips of 1812 case size based on the 0.90[0.94(0.75BNT-0.25NaNbO 3 )-0.06BT]-0.10CaZrO 3 system were successfully manufactured by our group as well, and the macro-and microstructures are shown in Figs. 13a and b, respectively. Wonderful dielectric temperature stability (ɛ r ∼ 560 ± 15%) was displayed ranging from −60 to 385°C with low loss (tan δ ≤ 0.02) Fig. 10 Phase diagram for structural phase transitions in (1−x) K 0.5 Na 0.5 NbO 3 -xLiTaO 3 [147] Fig. 11 Bar plots for summarising and comparing promising dielectric temperature stability of different systems [6,63,71,72] 12 IET Nanodielectr., 2018, Vol. from −55 to 340°C, as shown in Figs. 13c and d. As meaningful attempts, these investigations may pave the way for applications of Bi 0.5 Na 0.5 TiO 3 -based capacitor dielectrics. Nevertheless, for most of the new type Class II high-temperature capacitor dielectric systems, the development of MLCCs is largely lacking. Hence, not only should researchers focus attention on parameters of hightemperature dielectric material systems, but also attempt to combine with the real application process further. High-temperature MLCCs with more layers and smaller size are two development trends in the future, in order to adapt the miniaturisation and high integration of electronic devices. To gain more layers in limited volume, the thickness of dielectric layers should be as thin as possible. Fabrication technologies for thindielectric-layer MLCCs are being developed. Nowadays thickness of 0.5 µm for a single dielectric layer has been successfully achieved by Murata Company Ltd. Taking into account the reliability of a dielectric thickness <1 µm, finer dielectric and electrode powders are needed [4,152]. Nano-sized dielectric particles are indispensable, which can be synthesised by wet processes, such as hydrothermal synthesis, co-precipitation process, or sol-gel process. Moreover, fine electrode powder synthesised by the chemical vapour deposition method or wet chemical process is also important because the roughness of the electrode layer strongly affects the breakdown voltage of MLCC. For the formation of thin layers, using a tape caster seems to meet a limitation; hence, some researchers have been moving to produce MLCCs using film methods for ensuring their reliability [152]. In the process of dielectric layers with fine dense grains, a suitable sintering technology was also indispensable. For example, two-step sintering proposed by Chen and Wang [153] is deemed to be a costeffective preparation for the nanocrystalline ceramic capacitor and has been widely used by many researchers [154][155][156]. Furthermore, it is worthy to notice the effect of grain size on dielectric properties and reliability characteristics with continuous size reduction of capacitors, which plays a vital role in the functionality of devices [31,157]. In BT, for instance, the dielectric properties are maximised (ɛ r ∼ 6000) at intermediate grain sizes (∼1 µm). Buessem and co-workers [158,159] attributed the superior permittivity to the internal residual stress which changes the domain architecture, while more researchers believed that the physical origin may relate to domain wall displacement [160,161]. With decreasing grain sizes from 10 to 1 µm, smaller size of 90°d omain and higher activity of the 90° domain walls can make a higher permittivity; while below 1 µm, especially for nano-sized grains, the domain density of BT dropped because of the increased volume proportion of domain walls [160,161]. For BT-based nanocrystalline materials, though there is a sharp decrease in permittivity compared with intermediate-grain-sized ceramics, some advantages can be obtained as well [157,[162][163][164]. In a classical study by Arlt et al. [162], more depressed and flatter Curie peak can be observed with the decrease of grains from 6800 to 280 nm. Subsequently, Emelyanov et al. [163] proposed that the effective permittivity has a relationship with the grain size and thickness of grain boundary layer (dead layer). Thus, the diffused dielectric behaviours for nanoceramics are related to their high grain boundary density. Besides, Gong et al. [157] found that ceramics with smaller grains can also lead to higher insulation resistivity and reliability under high temperatures and DC bias voltage due to a higher volume of the grain boundary. These investigations suggest that nanocrystalline ceramics may not only help to miniaturise devices but also contribute to more appealing performance. In addition to the perspective of dielectric system design and new technological process, some other aspects, such as packaging and interconnect [165,166] dealing with temperature over 200°C, are also significant for achieving better application of HT-MLCCs in aviation or automotive fields.