Electroceramics for High-Energy Density Capacitors: Current Status and Future Perspectives

Materials exhibiting high energy/power density are currently needed to meet the growing demand of portable electronics, electric vehicles and large-scale energy storage devices. The highest energy densities are achieved for fuel cells, batteries, and supercapacitors, but conventional dielectric capacitors are receiving increased attention for pulsed power applications due to their high power density and their fast charge–discharge speed. The key to high energy density in dielectric capacitors is a large maximum but small remanent (zero in the case of linear dielectrics) polarization and a high electric breakdown strength. Polymer dielectric capacitors offer high power/energy density for applications at room temperature, but above 100 °C they are unreliable and suffer from dielectric breakdown. For high-temperature applications, therefore, dielectric ceramics are the only feasible alternative. Lead-based ceramics such as La-doped lead zirconate titanate exhibit good energy storage properties, but their toxicity raises concern over their use in consumer applications, where capacitors are exclusively lead free. Lead-free compositions with superior power density are thus required. In this paper, we introduce the fundamental principles of energy storage in dielectrics. We discuss key factors to improve energy storage properties such as the control of local structure, phase assemblage, dielectric layer thickness, microstructure, conductivity, and electrical homogeneity through the choice of base systems, dopants, and alloying additions, followed by a comprehensive review of the state-of-the-art. Finally, we comment on the future requirements for new materials in high power/energy density capacitor applications.


INTRODUCTION
To limit global warming to <1.50°C, as set out in the Paris agreement, carbon dioxide emissions need to decrease ∼45% by 2030 and reach net-zero by 2050. 1,2 Technologies based on renewable resources such as sun, wind, and tides will play a pivotal role to meet these targets. Although the increasing deployment of renewable energies is encouraging, there still are many barriers to the replacement of power generation from traditionally high CO 2 -emitting sectors based on coal and gas, which is still a critical and large portion of the energy generation, due to the intermittent nature of renewables. Hence, to simultaneously move away from fossil fuels and to circumvent the unpredictability inherent in clean energy resources, it is necessary to integrate energy-harvesting technologies with energy storage devices. Energy storage, therefore, is emerging as a key enabler for sustainable renewable technologies, particularly for the electrification of transportation but also in more specialized applications such as heart defibrillators and active armor. 3 Technologies already exist to store energy, such as batteries, electrochemical supercapacitors, and electrostatic capacitors. 4−16 The latter are electrical energy-storage devices belonging to the category of passive components, which are ubiquitous in electronics. Indeed, every year more than 3 trillion multilayer ceramic capacitors (MLCCs) are manufactured from BaTiO 3 (BT), the prototypical ferroelectric (FE) ceramic. 17 −22 In comparison with Li-ion batteries or fuel cells, the nonpolarized electrostatic or dielectric capacitors possess high power density (∼10 4 −10 5 W/kg) resulting from their faster charging/discharging characteristics (∼μs), which are advantageous for power electronics in electrical vehicles (EVs) and pulse power applications (Figure 1a). 4,23−27 Hence, electrostatic capacitors are emerging as promising candidates for energy storage devices, where high power density in combination with high energy density are important technological requirements, as illustrated by the exponential rise in publications devoted to energy storage involving electrostatic ceramic capacitors, Figure  1b. Apart from high energy density and fast charging− discharging rate, other properties such as temperature/ frequency stability, fatigue resistance, lifetime reliability, equivalent series resistance, and manufacturing cost are equally important for dielectric capacitors used in practical applications. New electroceramics are, therefore, required to facilitate nearengine power electronics, exhibit ultrafast charging, and have more durable EV performance at high temperature and voltage. Thus, future electroceramics must (i) deliver high energy density (W rec > 10 J cm −3 ) and conversion efficiencies (η > 90%); (ii) endure wider temperature ranges (−50−250°C) and frequency ranges (1−1000 Hz); (iii) exhibit greater reliability (>10 5 cycles) and fatigue resistance (<5% change over capacitor lifetime); and (iv) be compatible with cost-effective internal electrodes and be easily integrated with other components.
Historically, many different dielectric materials, ranging from paper and plastic to ceramics, have been employed in the fabrication of electrostatic capacitors. Nowadays, capacitors are fabricated from either polymers or ceramics because they offer the best combination of properties in terms of capacitance, dielectric loss, breakdown strength (BDS), and for the latter, thermal stability.
The prospects of employing ceramic capacitors for energy storage can be traced back to the 1960s work by Jaffe 28 from the Clevite Corp., USA. One decade later, Burn and Smyth 29 from Sprague Electric Company evaluated the energy storage performance in SrTiO 3 (ST) and BT with applied electric fields up to 400 kV cm −1 . Until that point, quantitative data of energy storage on these materials were limited to fields generally smaller than 150 kV cm −1 due to the relatively low dielectric BDS of the fabricated ceramics. They emphasized that the maximum energy density for a ceramic should be obtained for thinner dielectric layers due to the lower probability for the occurrence of defects (such as pores, voids, or microcracks), which are well-known sources of dielectric breakdown. Later in 1990, Love, 30 also from Sprague Electric Company, revisited energy storage in ceramic capacitors and highlighted empirical design principles to achieve enhanced energy storage in capacitors, as shown in Table 1. Commercial C0G-type capacitors are manufactured from low relative permittivity (ε r ) linear dielectrics but may achieve an energy storage of 1 J cm −3 , by virtue of their intrinsically high BDS. The significance of the BDS, to achieve high energy storage becomes apparent in the case of X7R-type capacitors, fabricated from high ε r BT. An important correlation between dielectric BDS and the thickness (t) can be extracted from Table 1. Indeed, by halving the t of the dielectric layers, the energy storage appears to increase >3 fold. This effect has been recently captured by Yang and co-workers, 31 who compiled BDS data from literature for several dielectric materials of different t and observed decay inversely proportional to (t) a , where a was determined as 0.5. Finally, when comparing the energy storage of Z5U and X7R, it becomes apparent that high ε r alone is not a sufficient parameter to achieve high energy storage. Interestingly, Love 30 stressed that the capacitor industry was rather conservative in terms of perfecting the BDS of ceramics to reach values near those of single-crystals, which would significantly enhance the energy storage in ceramic capacitors.
Love 30 proposed that maximum energy storage density can be achieved in intermediate rather than high ε r materials since they exhibit larger BDS. Fletcher and co-workers 32 convincingly postulated that greater energy storage densities can indeed be achieved in FE materials, whose Curie temperature (T c ) is adjusted to ensure that the material is operated in the paraelectric regime, where it shows a relatively small zero-field ε r , an approach already mentioned by Jaffe in 1961. 28 In 2009, Ogihara and co-workers 33 proposed the use of socalled weakly coupled relaxors, such as 0.7BaTiO 3 −0.3BiScO 3 (0.7BT−0.3BS), to fabricate energy storage devices. This new conceptual approach aimed at exploiting the extraordinary temperature stability of ε r exhibit by this family of materials. When compared with commercial X7R capacitors, 0.7BT− 0.3BS capacitors displayed superior performance, reaching a recoverable energy density (W rec ) of 6.1 J cm −3 at 730 kV cm −1 . Again, the large dielectric BDS played a decisive role in this performance. More recently, in 2019 Wang, Reaney and coworkers 34 unveiled a novel approach to enhance energy storage characteristics via the fabrication of chemically heterogeneous but electrically homogeneous ceramics, with W rec reaching 10.5 J cm −3 , as detailed later in this review.
Here, we present the principles of energy storage performance in ceramic capacitors, including an introduction to electrostatic capacitors, key parameters for evaluating energy storage properties, microstructural considerations, and critical electrical factors. Second, we will review the current state-of-the-art for lead and lead-free electroceramics for energy storage capacitors with bulk ceramics, ceramic multilayers (MLs), ceramic films and glass ceramics evaluated separately. Third, we will describe strategies for optimizing energy storage in electroceramics. Finally, we will demonstrate, with appropriate examples, a guide to the future development for electroceramics in energy storage capacitors.

Electrostatic Capacitors
The simplest dielectric capacitor consists of two parallel metallic plates separated by an insulator, which becomes polarized under the application of an electric field. This is the defining behavior of a dielectric material. The actual capacitance, C (i.e., ability to store charge), of an ideal capacitor is given by the ratio of the charge, Q, stored on each metallic plate and the applied voltage, V, as shown by eq 1.
Nevertheless, from a practical viewpoint, a more useful equation to compute the C of a real device, as illustrated in Figure 2, encompassing a dielectric material between two parallel plates of area, A, separated by a distance, d, subject to a V, can be obtained through the application of Gauss's law where ε is the permittivity of the dielectric, and a measure of its polarizability. Combination of eqs 1 and 2 provides the relationship: From eq 3, it becomes immediately apparent that the ability of dielectric capacitor to charge and, therefore store energy, is ultimately associated with ε of the dielectric.

Key Parameters for Evaluating Energy Storage Properties
During the application of a V, the electrostatic energy stored, W, in the dielectric can be estimated by where Q max is the maximum charge achieved at the end of the charging cycle and dq is the incremental charge increase during the charging cycle. The volumetric energy density, W st (i.e., the  energy stored per volume unit, A d), is a common key performance indicator, expressed by where E is the electric field and D max is the electric displacement in the material under the maximum applied field, E max . The electrical displacement (D) corresponds to the charge density (Q/A) on the metallic plates and is expressed by D = ε 0 E + P (Figure 2), where P is the polarization (surface charge density). For high ε materials, D is approximately equal to P, and it follows that D = εE = ε 0 ε r E, where ε 0 is the permittivity of free space (= 8.854 × 10 −12 F m −1 ) and ε r is the relative permittivity, which is the ε/ε 0 ratio. This approximation allows stored energy density (W st ) to be defined in terms of P, as follows where P max is the maximum polarization reached at the E max . From a practical viewpoint, eq 6 is prevalent in the calculation of W st because several experimental methods exist to determine P under an applied E. In 1961, Jaffe 28 pointed out that the recoverable energy (W rec ) corresponds to the area above the discharging curve, whose upper limit is given by the P max . Essentially, the mathematical integration of the area above a polarization-electric (P−E) loop provides an estimate of W rec , as schematically illustrated in Figure 3 for four distinct types of polarization response.
For linear dielectrics such as Al 2 O 3 , where ε r is independent of the applied E. The calculation of W rec from the P−E response illustrated in Figure. 3a, is given by which clearly shows that W rec is dependent on ε r and E. Parts b− d of Figure 3 show cases where polarization responses deviate from linearity, and consequently, the computation of W rec needs to be carried out using eq 6. The response illustrated in Figure 3b is typical of a classical FE material, such as BT, where the hysteresis is linked to polarization switching of macroscopic FE domains, as explained in detail in the review by Damjanovic. 35 Already in 1961, Jaffe 28 stressed that in FEs, charging energy is mainly absorbed by domain switching and is retained as remanent polarization (P r ). The typically high remanence of classical FEs can be effectively minimized via chemical doping, giving rise to the response shown in Figure 3c, which is characteristic of relaxor-ferroelectrics (RFEs), such as doped-BT and Pb(Mg 1/3 Nb 2/3 )O 3 . 36 It is now generally accepted that relaxor behavior originates from the response of polar nanoregions (PNRs) to an alternating E. RFEs remain unsaturated at high applied E, and therefore, any increment of the E will have a contribution to energy storage. Remanence-free materials are therefore, preferable for achieving high W rec . Linear dielectric materials meet this requirement but due to their low ε r , energy storage is limited. Antiferroelectrics (AFEs) display low-remanence under low E but at large E the P−E loop opens due to the stabilization of an FE with respect to AFE phase and they display a saturated polarization, as illustrated in Figure 3d. In principle, therefore, as suggested by Jaffe, 28 AFEs should afford advantages for high energy storage, providing that dielectric breakdown issues are eliminated (i.e., the BDS should be high enough to induce the AFE-FE phase transition).
From the above, it becomes evident that nonlinear dielectric materials such as FEs, RFEs, and AFEs exhibit energy dissipation (W loss ); therefore, the W rec is actually the most important parameter, as schematically illustrated in Figure 3c ( where W loss is the energy loss during discharging, which correlates to the area enclosed by the P−E loop (Figure 3c green area). Electric-field induced polarization can be determined via the measurement of charge, current, and voltage responses, typically achieved using either the Sawyer−tower, the virtual ground, the   Table 2. For details of each method, the reader is referred to Prume and co-workers. 37 Prume, Schmitz, and Tiedke proposed that overall the virtual ground method offers the highest precision for the measurement of FEs.

Key Factors for Optimizing Energy Density
The microstructural features of electroceramics, such as density, grain size, secondary phases and core−shell structures, play an important role in energy storage properties. Simultaneously, the intrinsic electrical response, e.g., band gap, alongside the electrical microstructure, i.e., the distribution of conductive and resistive elements, are equally critical factors for the optimization of energy density. The following section reviews these factors, and gives examples of where and how they may be optimized. 2.3.1. Intrinsic Band Gap. The band gap (E g ) is the forbidden energy between the top of the valence band and bottom of the conduction band. E g is commonly used to define insulator (E g > 4.0 eV), semiconductor (0.0 eV < E g < 4.0 eV), and metal (E g = 0.0 eV). For semiconductor, the intrinsic BDS can be defined as where BDS is direct proportional to E g . 38 Thus, semiconductors with wider E g have higher intrinsic BDS. The electronic structure and band gaps of semiconductor can be studied theoretically using, e.g., linear discriminant analysis, or experimentally, e.g., absorbance spectroscopy and diffuse reflectance spectroscopy. 39 A general rule of thumb is that the activation energy (E a ) for conduction is approximately half E g . Both may be increased by doping or through the formation of solid solutions, often delivering higher BDS and W rec . 40,41 For example, the highest E g ∼ 3.58 eV among all different kinds of lead-free electroceramics was found in NaNbO 3 (NN), as shown in Figure 4a. 42 Thus, NN was introduced into Na 0.5 Bi 0.5 TiO 3 (NBT) and BiFeO 3 −BaTiO 3 (BF−BT) to enhance E g . The E g for BF−BT−xNN ceramics increased from 2.5 eV up to 2.95 eV for x ≤ 0.15, as shown in Figure 4b accompanied by significant enhanced W rec ∼ 8.12 J cm −3 under electric field ∼400 kV cm −1 , along with greater thermal stability (±10%, −50 to +250°C) and ultrafast discharge rate (t 0.9 < 100 ns), Figure 4. 43 2.3.2. Electrical Microstructure. The distribution of regions with different conductivity and ε r are important aspects of the so-called "electrical microstructure" of electroceramics. 44 In many instances, such as the core−shell microstructure or grain boundary response of BT based ceramics, the distributions are markedly heterogeneous and lead to localization of the electrical field strength in lower ε r regions or pathways for breakdown through interconnected conducting regions. In 2019, electrical homogeneity was for the first-time proposed by Wang, Reaney and co-workers in the BF−BT system as a key factor to optimize BDS and as a consequence W rec . Electrical heterogeneity was effectively eliminated by alloying with a third end-member so that it became more difficult to form a conductive pathway at high field, resulting in higher BDS and W rec . 34 A homogeneous electrical microstructure may be obtained in many different ways such as heat-treatment in the appropriate atmosphere (N 2 , Air, O 2 ) provided the type and magnitude of electrical conductivity is affected by oxygen stoichiometry. Practically, however, in production, a suitable dopant strategy is utilized once the conduction type is known (p vs n type). For example, the conductivity of BF-ST-based compositions is suppressed by doping with 3 mol % Nb on the B-site to compensate for Bi volatilization and the formation of oxygen vacancies (V O .. ), through variation of the Fe valence (Fe 3+ to Fe 4+ ). 45 For materials with more than one bulk-like region, e.g., phase mixtures, core−shell microstructures, or surface layers, alternating current (AC) impedance spectroscopy (IS) is able to show multiple responses and the resistance (R) and C can be extracted. 46−53 Both the volume fraction and difference in magnitude of R and C for multiple electrical responses are equally important in influencing energy storage performance. Given the importance of the electrical microstructure, a brief outline of the role of IS is described and its advantages with respect to direct current methods are emphasized.
Direct current (DC) electrical measurements are the most commonly employed technique to evaluate the electrical characteristics of materials. However, they merely give the overall response instead of the properties of specific regions (e.g., grains and grain boundaries) unless microprobe techniques are employed. 54,55 Such techniques are useful but the sample volume is small, which casts doubt on their ability to represent global behavior and they are difficult to implement experimentally.
An alternative and much more convenient technique is IS. In IS measurements, an AC signal with small voltage over a wide range of frequency, typically 10 −2 to 10 7 Hz, is applied on the sample. 44, 56 The small voltage prevents any permanent change to the sample as well as yielding a (near) linear relationship between input and output. The wide range of frequencies allows separation of the response of different electro-active regions according to their relaxation times. For energy storage capacitors, impedance is capable of: (i) establishing the contributions to the electrical microstructure (grains, grain boundaries, core−shell structure and electrode−sample interface) and determine their individual conductivity and ε r which give an insight into the distribution of electrical components within the sample; (ii) verifying the origin of the dominant electrical behavior (i.e., grains, grain boundary or interfacial layer response); 57,58 and (iii) determining the conduction mechanism and charge carrier type which helps further interpret the electrical response of the material. 47 Impedance can be defined as a complex number which usually contains both resistive and reactive (capacitive and/or inductive) components: Different electro-active regions of a material are characterized by a R and a C, usually in parallel. Then the electric relaxation time or time constant, τ, of each region can be expressed as its R and C RC τ = (12) at the frequency of maximum loss, ω max , it holds the relation: Due to their different R and C values, electro-active regions can be separated in the frequency domain. Once the value of R and C are extracted, they can then be assigned to appropriate regions of the sample. Normally the impedance measurement needs to be taken over a temperature and/or oxygen partial pressure (pO 2 ) range to gain a better understanding of the conduction mechanism and the charge carrier. The associated activation energy, E a , can be estimated using Arrhenius equation where σ is the conductivity, σ 0 is pre-exponential factor, k is the Boltzmann constant, and T is temperature. E a may be related to predominant charge carrier and conduction mechanism. The type of charge carrier may also be determined to some extent by the pO 2 dependence of conductivity, i.e., p-type: conductivity increases with increasing pO 2 ; n-type: conductivity decreases with increasing pO 2 ; ionic charge carrier: conductivity is independent with pO 2 . 2.3.3. Density and Porosity. The density of the ceramic materials plays an essential role on electrical performance, especially BDS. Ceramics with higher density tend to support higher E closer to the intrinsic/theoretical BDS. In contrast, low density ceramics exhibit conductive pathway composed of pores/voids which result in short circuit under modest field strengths. The relationship between the voltage across the pore and the external E based on a "slab" model is shown below where V c and V ext are the voltage applied cross the cavity pore and external applied voltage, ε c and ε d are the permittivity of the cavity and the dielectric, respectively, 59,60 and td and tc are the thicknesses of the dielectric and cavity, respectively. Thus, the local E increases markedly for materials with larger pores and pore volumes, resulting in lower BDS. High density electroceramic materials are commonly obtained by optimization of the sintering conditions, including sintering temperature/time and heating/cooling rate. For ceramics that are difficult to densify using a conventional approach, sintering aids are often added. 61−64 Higher density ceramics may be obtained by the addition of ZnO, 65 CuO, 66 and MgO, 62 which enhances BDS and W rec . For K 0.5 Na 0.5 NbO 3 − Bi(Mg 2/3 Nb 1/3 )O 3 (KNN−BMN), small amounts of CuO help densify ceramics through the formation of a transient liquid phase, as reported by Qu and co-workers ( Figure 5). 64 The sintering temperature was also reduced from 1150 to 930°C, allowing compatibly with Cu or Ag/Pd internal electrode in MLs and giving rise to W rec ∼ 4.02 J cm −3 at 400 kV cm −1 for 0.9KNN-0.1BMN with 1% mol CuO. 66 Different sintering technologies, such as spark plasma sintering (SPS), two-step sintering, 67 and the formation of coatings using chemical methods, 68−81 have also been shown to improve density and give rise to higher BDS and W rec .
2.3.4. Grain Size. The effect of grain size (G) on energy storage properties has been discussed for several electroceramics because of the relationship between BDS and G, expressed in eq 16 where a is the exponent values being in the range of 0.2− 0.4. 31,82−84 Waser explained that leakage current in fine-G ceramics is lower than coarse-G ceramics due to the high grain boundary density which act as barriers for charge carriers. 85 Thus, dielectric materials with high density and fine-G are required to optimize energy storage. G may be tailored by chemical doping and the formation of solid solution. It may also be modified by the application of an ultrathin coating on the primary particles prior to sintering via chemical coating methods, e.g., SiO 2 on BT ceramics. 67,77,86−89 The optimization on E max and W rec via grain size-engineering for several materials is illustrated in Figure 6. For example, an average G ∼ 10 μm was reported for BF−BT ceramics, which was reduced to <2 μm after A-site Nd doping, as shown in Figure 7. Meanwhile, improved W rec ∼ 1.8 J cm −3 and η ∼ 88% were obtained for 15 mol % Nd−BF−BT and 40 mol % Nd−BF−BT, respectively. 90 Similar optimization behavior has also been found in KNN−BMN and KNN−ST ceramics, resulting in BDS ∼ 400 kV cm −1 and W rec > 3.5 J cm −3 . 91,92 2.3.5. Core−Shell Structure. Core−shell subgrain microstructures are observed in many lead-free ceramics, due to either kinetic limitations of the diffusion process (typical for BT based ceramics) or immiscibility on cooling from high temperature for perovskite end members with dissimilar ion size and bonding (BF based ceramics). 22,34,45 The effect of core−shell microstructures on energy storage performance is still unclear. In BTbased ceramics, the cores are often more conducting than the doped shells and core to core conductive pathways lead to breakdown. 93−96 For BF based ceramics, the defect chemistry of the cores and shells remains to be elucidated, but initial work suggests that further dopants are needed to create electrical homogeneity and thus eliminate the conducting pathways. 34,45 The theoretical modeling has reported a positive influence of core−shell microstructure but none have been unambiguously validated experimentally. 97

STATE-OF-THE-ART IN ELECTROCERAMICS FOR ENERGY STORAGE
3.1. Bulk Ceramics 3.1.1. Lead-Based Ceramics. Lead-based ceramics are used commercially as energy storage materials for high-power pulsed capacitors due to their excellent W rec and η. 98−101 The energy storage properties of RFE and AFE lead-based ceramics are summarized in Table 3.  102 Li and co-worker probed the effect of domain structure on W rec and thermal stability of Figure 8. 0.2PMN−0.8PST ceramics exhibited W rec ∼ 0.85 J cm −3 with excellent thermal stability which was attributed to the coexistence of ferroelectric domains and PNRs. 111 3.1.1.2. Lead-Based Antiferroelectrics. PbZrO 3 (PZ) is the first known AFE and exhibits a double P−E hysteresis loop below T C . However, the high critical switching field required for an AFE−FE phase transition at room temperature limits applications for energy storage. Chemical substitution to reduce switching field is an effective strategy to overcome the problem and three well-known PbZrO 3  When Pb ions are replaced by ≤30 mol % La on the A-site in accordance with a lead vacancy (V pb .. ) ionic compensation model, an orthorhombic AFE phase similar to PbZrO 3 occurs for Zr rich compositions. However, only PLZT with <10 mol % La is typically utilized for energy storage applications 141−147 since higher concentrations have lower polarization and therefore lower W rec . Li and co-workers prepared (Pb 0.97 La 0.02 )-(Zr 0.95 Ti 0.05 )O 3 ceramics via a solid-state reaction route, yielding W rec ∼ 0.83 J cm −3 and η ∼ 70% under an electric field of 55 kV cm −1 . 114 Jo and co-workers found that AFE and RFE behavior can both be obtained by substitution of La and excess PbO in PLZT, resulting in the enhancement of W rec by promoting a slim and slanted hysteresis loop. Both high W rec ∼ 3.04 J cm −3 and η ∼ 92% were obtained along with no performance degradation after 10 5 cycles. 112 Tuning the Zr/Ti ratio has also shown to be an effective way to improve W rec of PLZT ceramics. Qiao and coworkers reported slimmer P−E loops giving a W rec ∼ 3.38 J cm −3 in (Pb 0.895 La 0.07 )(Zr x Ti 1−x )O 3 ceramics by changing the Zr/Ti ratio ( Figure 10), which was attributed to the reduction of tolerance factor and "enhancement of antiferroelectricity". 115 Mn doping is also suggested to improve W rec of PLZT by "enhancing antiferroelectricity". 143   (Zr,Sn,Ti)O 3 with 2 mol % La has been most commonly studied in which the Zr/Ti/Sn ratio is varied to give a complex phase diagram that contains FE tetragonal (F T ), high-temperature nontilted FE rhombohedral (F R(HT) ), low temperature FE rhombohedral (F R(LT) ) AFE tetragonal (A T ), and AFE orthorhombic (A O ) phases.
We note that the A T phase in both PLZT and PLZST has been shown to be incommensurate by a number of researchers and might be better described as a A O phase in which there is disorder of antipolar coupling. 168−170 PLZST is AFE for concentrations with <15 mol % Ti. For compositions with A O structure, the critical phase switching fields are above BDS but ceramics with the A T phase undergo electric field-induced AFE-FE switching at room temperature, for which the switching field increases with increasing Sn concentration ( Figure 12). 118 Adjusting the Zr/Sn/Ti ratio leads to optimization of W rec in PLZST ceramics 116,118,148,157 with an increase in Sn concentration leading to a reduction in switching field (forward threshold electric field, E F , and backward threshold electric field, E B ) and weakening ferroelectricity. 171    with η of 82% for (Pb 0.97 La 0.02 )(Zr 0.5 Sn 0.44 Ti 0.06 )O 3 ceramics with good temperature stability. 118 Recently, a ferrielectric (FIE) configuration was reported in PLZST which consists of ferroelectric ordering segments with either magnitude or angular modulation of dipoles. 172 The net polarization of field-induced FE order can be tailored by adjusting the Sn/Ti ratio.
The performance of PLZST ceramics is also influenced by the Pb/La ratio. 117,122,126 The AFE phase becomes more stable with a commensurate increase in the AFE−FE switching field as La concentration increases. The energy storage properties of (Pb 1−1.5x La x )(Zr 0.5 Sn 0.43 Ti 0.07 )O 3 ceramics were optimized (W rec of 4.2 J cm −1 ) by Dan and co-workers for compositions with x = 0.03 due to a large switching electric field and high BDS. 3,126 Furthermore, doping with Ba and Sr (A-site) improves fatigue behavior and temperature stability, suppresses the stress sensitivity, and enhances energy storage. 148 . Wang and co-workers reported a unique E-induced multiphase transition in PLZS for which a conventional AFE−FE phase transition at low E, followed by a second FE-FE phase transition at a higher E, leads to an increase in polarization. 131 W rec of 10.4 J cm −3 and η of 87% were achieved at 400 kV cm − 1 for (Pb 0.98 La 0.02 )-(Zr 0.55 Sn 0.45 ) 0.995 O 3 ceramics, along with superior discharge current density of 1640 A cm −2 and ultrafast discharge speed (75 ns discharge period) (Figure 13a,b). 131 Subsequently, a recordhigh W rec ∼ 11.2 J cm −3 and a high η ∼ 82% were realized in (Pb 0.98−x La 0.02 Sr x )(Zr 0.9 Sn 0.1 ) 0.995 O 3 ceramics, as illustrated in (Figure 13c,d). The substitution of Pb by Sr gave rise to an   Chemical Reviews pubs.acs.org/CR Review increase in BDS and AFE/FE switching fields, leading to further enhancement of energy storage performance. 130 3.1.2. Lead-Free Ceramics. In the last decades, extensive research has focused on lead-free electroceramics due to concerns over the toxicity of lead/lead oxide-based compounds. 174 Table 4.
The most effective approach, however, is the introduction of a Bi-based perovskite end member in which the B-site contains multiple cations. Doping with a Bi ion that has a lone pair electronic 6s 2 configuration into the Ba-site increases P max . P r is minimized by forming a so-called "weakly coupled relaxor" state in which long-range polar coupling is disrupted by the multiple B-site ions which have different valence and ionic radius to Ti. Using this principle, Hu and co-workers 208 reported high W rec of 4.49 J cm −3 with a η of 93% for 0.6BT−0.4 Bi(Mg 0.5 Ti 0.5 )O 3 (BT−BMT) ceramics that were temperature stable to 170°C (variation W rec < 5% and η < 4%).
Of greater commercial potential, Yang and co-workers reported similar properties with BT−0.06 Bi 2/3 (Mg 1/3 Nb 2/3 )O 3 (BT−0.06B 2/3 MN) but in compositions compatible with Ag/Pd electrodes due to the presence of only 4 mol % Bi on the Asite. 209 Similar energy storage properties, W rec ∼ 4.6 J cm −3 and η ∼ 92% (Figures 14a,b) to BT−BMT were obtained for BT− 0.06B 2/3 MN ceramics which also benefited from the highest BDS, ∼520 kV cm −1 , among all reported BT-based compositions. 209 BT−0.06B 2/3 MN is not electrically homogeneous but BDS and W rec were still optimized by reducing, though not completely eliminating, the difference between the bulk and grain boundary responses with respect to undoped BT, Figures 14c,d. 209 3.1.2.2. SrTiO 3 -Based Ceramics. ST, which is an incipient ferroelectric, is considered as a promising candidate for energy storage applications due to its relatively high permittivity (ε r ∼ 300) and low dielectric loss (<1%) at room temperature. Extensive efforts have been made to improve the energy storage performance of ST-based ceramics, including (i) doping with Ba, 213−216 Dy, 217 Mg,218 Ce, 64 231 233 Table 5.
From a review of the literature, doping commonly increases both ε r and P max but decreases the BDS, sintering aids increase BDS but lower the P max . The highest energy densities have therefore been achieved through either dopants and/or alloying with "relaxor end-members" which also act as sintering aids. Adopting these protocols, a W rec of 3.1 J cm −3 with η ∼ 93% was obtained for 0.9(Sr 0. 7 Figure 16. 91 The energy storage properties of KNN-based materials are summarized in Table 6.
3.1.2.4. BiFeO 3 -Based Ceramics. BF-based ceramics are best known as multiferroics but have also been explored for hightemperature ferroelectric and piezoelectric applications due to  The t of the bulk ceramics is commonly >0.1 mm. their high T C and large spontaneous polarization. 256−259 Compared with other lead-free ceramics, BF-based were not initially considered as good candidates for energy storage applications due to their high leakage current. 260 p-type electrical conductivity due to the volatilization of Bi and the variation of Fe valence states has been reported frequently in BF system, which limits the BDS and restricts energy density. 261−263 However, the introduction of dopants and alloying with endmembers suppresses the leakage current, reduces the electrical conductivity, increases intrinsic E g and induces transitions from a FE to RFE state. The energy storage properties of BF-based materials are summarized in Table 7.
BF−BT-based materials are purported as potential energy storage compositions due to their excellent BDS and high P max . 34,45,90 Undoped BF−xBT ceramics exhibit optimized ferroelectric and piezoelectric properties in a mixed-phase region of 0.25≤ x ≤ 0.35. 264 The majority of studies have focused on this region and modified compositions either by (i) A and/or B-site chemical doping, including Nd, Nb, Zn 0.5 Zr 0.5 , Zn 2/3 Nb 1/3 , and Li 0.5 Nb 0.5 or (ii) alloying with a third endmember, such as Nd(Zn 0. The importance of electrical homogeneity was first postulated in 2019 by Wang, Reaney, and co-workers as a key factor to optimize BDS as well as W rec . 34 Two electrical components with       Recently, superior energy density through tailored dopant strategies was achieved in BF−ST−xNb−yBMN ceramics, by promoting electrical homogeneity, enhancing E a and suppress-   Figure 18h. 45 3.1.2.5. Na 0.5 Bi 0.5 TiO 3 −Based Ceramics. NBT-based ceramics are promising candidates of lead-free dielectrics due to their high P max and T c . However, their large hysteresis and low BDS are not ideal for high energy density capacitor applications. 277 69,71,74,335,336 The energy storage properties of NBT-based materials are summarized in Table 8. Notably, Li and co-workers reported that 0.55NBT−0.45-(Sr 0.7 Bi 0.2 )TiO 3 (SBT) achieved W rec of 2.5 and 9.5 J cm −3 with η > 90% for bulk ceramic and MLs at 200 and 720 kV cm −1 (Figure  19a,b), respectively. 222 Superior W rec ∼ 7.02 J cm −3 and η ∼ 85%, were also reported for 0.78NBT−0.22NN ceramics at E max ∼ 360 kV cm −1 , Figures 19c,d, with <10% variation from 25− 250°C and from 0.1 to 100 Hz. 297 Recently, Ji and co-workers 337 proposed that the key factors for designing an ideal RFE with high energy density were as follows: (i) utilization of a highly polar base system (e.g., NBT); (ii) disruption of long-range polar coupling through forming solid solutions with, e.g., SBT and BMN without sacrificing average ionic polarizability, Figure 20a, and (iii) simultaneously inducing or retaining electrical homogeneity with a highly resistive single component in IS (∼250 kΩ cm at 660°C), Figure 20b. These factors combined to give E max ∼ 470 kV cm −1 , W rec ∼ 7.5 J cm −3 , and η ∼ 92% for 0.62NBT−0.30SBT− 0.08BMN, Figure 20c,d. 3.1.2.6. AgNbO 3 −Based Ceramics. AFEs have long been considered as the prime candidate for energy storage capacitors due to their large P max and small P r . There are only a handful of lead-free AFE systems, with AN showing particular promise because it possesses a large saturation polarization of 52 μC cm −2 under an E max ∼ 220 kV cm −1 . 343 Recent research on AN ceramics has focused on stabilizing the AFE phase so that switching field is moved to higher fields while simultaneously optimizing P max . 344,345 There have been a number of recent studies on AN focusing on (i) substitution of aliovalent B-site oxides such as MnO 2 and WO 3 ; 346 (Figure 21a,b). 357 A-site doping with ions smaller in radius than Ag is suggested to decrease tolerance factor and enhance AFE stability, while donor doping is compensated by A-site vacancies which reduce antipolar and polar coupling of the AFE and field induced FE phases, respectively. Some authors postulate that substituting B-site ions  Chemical Reviews pubs.acs.org/CR Review with a lower polarizability than Nb also stabilizes the AFE phase and moves the switching field higher. 348,350,359,363 The underlying principles are schematically represented in Figure 21c. The energy storage properties of AN-based materials are summarized in Table 9 Table 10. Zuo and coworkers 42 have also proposed the concept of an "AFE relaxor" to explain the energy storage properties of 0.78NN−0.22NBT ceramics. They argue that the local AFE phase transforms reversibly into an FE monoclinic phase at∼ 400 kV cm −1 , giving a large ΔP (P max > 50 μC cm −2 and P rem < 5 μC cm −2 ). W rec of ∼12.2 J cm −3 was reported with η ∼ 69%, at 680 kV cm −1 , Figure  22. 42 However, the term "AFE relaxor" has little physical significance since an antipolar phase cannot form short-range polar features characteristic of a relaxor. 0.78NN−0.22NBT  Chemical Reviews pubs.acs.org/CR Review may, therefore, be better described as either a relaxor or a shortrange AFE phase that undergoes a field induced transition. This intriguing behavior is interesting, but it is the large E max (680 kVcm −1 ) that is most likely responsible for the exceptional W rec rather than the intrinsic crystal chemistry. The underpinning reasons for the large E max most likely relate to the defect chemistry, band gap and electrical homogeneity, consistent with the key factors proposed by Ji and co-workers. 337 3.1.3. Glass Ceramics. Glass-ceramics are composed of one or more crystallized phases (ceramics) dispersed uniformly in amorphous phase (glass). They often exhibit the combined properties of ceramics and glass depending on the induced crystalline phases and their microstructures. Glass-ceramics are prepared by melting the requisite raw materials, cooling to room temperature to form a glass, followed by two step annealing to induce crystal nucleation (approximately at the glass transition temperature, T g ) and growth > T g Figure 23. 179,382 The microstructure of a glass ceramic is typically dominated by a largely 2D and 3D defect-free (e.g., no grain boundaries) glass phase and a uniformly distributed (provided the system undergoes homogeneous rather than heterogeneous nucleation) ceramic phase. W rec and η are both large due to the high BDS associated with the absence of 2D and 3D defects accompanied by a near zero value of P r . The energy storage properties of glassbased glass ceramics are summarized in Table 11.
As discussed above, the crystallization of glass ceramics is controlled by the annealing procedure, where the annealing temperature and time are critical for nucleation and growth of the ceramic phase, the microstructure and the properties. Generally, the volume fraction of crystalline phase increases with increasing annealing temperature and time, accompanied by an increase of ε r and decrease of BDS. The optimized W rec is a balance between ε r and BDS. Chen and co-workers 383 391 Each constituent oxide in the glass matrix has an important effect on the crystal phase, microstructure, BDS, and energy storage properties. For example, SiO 2 is an important and active studied constituent oxide in glass matrix. With increasing SiO 2 content, ε r of SrO−Na 2 O−Nb 2 O 5 −SiO 2 (SNN-Si) glass ceramics first increased and then decreased as shown in Figure  25a, which was attributed to the change of volume fraction Sr 6 Nb 10 O 30 (Figure 25b). The optimal ε r of 120 and BDS ∼ 1700 kV cm −1 were obtained with 35 mol % SiO 2 (Figures  25a,c), resulting in the highest theoretical W rec of 15.2 J cm −3 . 392 Wang and co-workers reported that, as K 2 O concentration increased in K 2 O−BaO−Nb 2 O 5 −SiO 2 glass ceramics, grain boundary R and activation energy decreased, indicating the decrease of interfacial polarization, leading to the enhancement of BDS to ∼1900 kV cm −1 and W rec ∼ 12 J cm −3 . 393 They also reported that substitution of Sr for Ba in SrO−BaO−K 2 O− Nb 2 O 5 −SiO 2 led to the formation of solid phase Sr 0.5 Ba 0.5 Nb 2 O 6 and improvement of dielectric properties. 394 The highest BDS of ∼1800 kV cm −1 and W rec of 17.5 J cm −3 were achieved with Sr = 0.4 due to a uniform and dense microstructure and lower    396 Compared with alkali based compositions, alkali-free glass compositions were found to deliver lower dielectric loss and fewer defect microstructure. For example, Smith and co-workers reported both ultrahigh E max ∼ 12000 kV cm −1 and W rec ∼ 35 J cm −3 in BaO−B 2 O 3 −Al 2 O 3 −SiO 2 glass. 397 The effect of Al/Si ratio on the modification of the microstructure and properties of SrO−BaO−Nb 2 O 5 −SiO 2 −Al 2 O 3 glass ceramics was also studied by Xiu and co-workers. 398 Additionally, rare-earth oxides, such as  405 are also commonly substituted into glass formulations for energy storage applications. Rare-earth oxides are mainly reported to act as nucleating agents 399,402 or crystal growth inhibitors. 400,406 Zhang and co-workers revealed that La 2 O 3 leads to a homogeneous microstructure in the BaO−SrO−TiO 2 − Al 2 O 3 −SiO 2 glass-ceramics which improved BDS ∼ 1600 kV cm −1 and W rec ∼ 3.2 J cm −3 (2.5 times of the glass-ceramics without La 2 O 3 ). 399 Zheng and co-workers reported that 0.5 mol % La 2 O 3 in SrO−BaO−Nb 2 O 5 −B 2 O 3 −ZnO glass-ceramics also optimized ε r (∼130) and W rec ∼ 7.1 J cm −3 through achieving a BDS ∼ 1100 kV cm −1 due to a reduction in crystallite size and precipitation of high ε r phase, Sr 0.5 Ba 0.5 Nb 2 O 6 . 400 A similar effect was reported for Sm 2 O 3 by Chen and co-workers in the SrO−BaO−Nb 2 O 5 −B 2 O 3 −ZnO glass ceramics with W rec of 8.2 J cm −3 at 1100 kV cm −1 . 402 Moreover, Yb 2 O 3 is reported to eliminate the impurity phases and form a uniform microstructure in BaO−SrO−TiO 2 −Al 2 O 3 −B 2 O 3 −SiO 2 glass-ceramics, leading to W rec of 3.5 J cm −3 , ∼ 1.8 times higher than undoped compositions. 404 Apart from the conventional annealing, novel methods such as microwave treatment have been reported to improve the energy storage properties of glass ceramics. Zhang and co-workers found that microwave treatment restrained the formation of the dendritic microstructure in Ba x Sr 1-x TiO 3 −(Ba−B−Al−Si−O) (BST−BBAS) glass-ceramics (Figure 26), leading to the improvement of BDS from 1200 kV cm −1 to 1500 kV cm −1 , corresponding to W rec of 2.8 J cm −3 (950°C anneal). 408 Xiao and co-workers further reported that the precipitation of impurity phases in the K 2 O−SrO−Nb 2 O 5 −SiO 2 −Al 2 O 3 − B 2 O 3 glass-ceramics was limited by controlling the crystallization time using microwave sintering, with optimum ε r , BDS of 1400 kV cm −1 and maximum theoretical W rec (∼9 J cm −3 ) obtained after 10 min. 407 3.1.4. Summary of State-of-the-Art in Ceramics. The debate over whether lead-free electroceramics can replace their lead-based counterparts has been ongoing for over two decades. Lead based compositions generally outperform their lead-free counterparts on most metrics. Moreover, lead-free compositions are disparate with a large number of different formulations potentially required to cover the properties achieved with essentially doped PZT. Provided, however, that the perform-ance, reliability and cost of lead-free are competitive with PZT, it is highly likely that lead-free electroceramics will begin to replace their lead-based equivalents and attain large scale production in the coming years as a consequence of environmental legislation. 31,103,415 Of all the applications, lead-free high energy density capacitors are the most likely to see large-scale production since (i) the performance of lead-free compositions is approaching that of lead-based; (ii) reduction in intrinsic electrical properties may be compensated by increasing the BDS often through decreasing layer thickness (see section 3.1.2); and (iii) the capacitor industry is dominated by lead-free BT-based MLCCs, and thus, there is an expectation that the related products will not contain lead. 34,45,90,209,276,416 This latter statement does not hold for piezoelectric ceramics market which is dominated by PZT and its derivatives. 174,417−419 The energy storage performances, E max , ΔP, W rec , and η, for lead-based and lead-free ceramics are summarized and plotted in Figure 27 (note: glass ceramics are not included). A comparison of W rec vs E max for different lead-based/lead-free bulk ceramics is displayed in Figure 27a. Lead-based bulk ceramics have the advantage of both high E max (up to ∼400 kV cm −1 ) and W rec (up to ∼12 J cm −3 ) with respect to lead-free candidates. NN-based ceramics currently offer the highest W rec under high E (>350 kV cm −1 ) for lead-free compositions, followed by AN-, NBT-, BF-, and KNN-based materials. BT and ST-based ceramics display the lowest W rec in spite of their high E max of ∼450 kV cm −1 , but Chemical Reviews pubs.acs.org/CR Review they are perhaps the most appealing dielectrics commercially since they are the current basis of MLCC production. ΔP vs W rec is compared in Figure 27b. Lead-, NBT-, and BFbased materials exhibit extraordinarily high ΔP (up to ∼60 μC cm −2 ), followed by AFE NN-and AN-based (up to ∼40 μC cm −2 ) and KNN-based materials (up to ∼35 μC cm −2 ) with the lowest (up to ∼25 μC cm −2 ) for BT-and ST-based materials. Bicontaining electroceramics such as NBT and BF, have been heavily studied recently as potential lead-free electroceramic materials due to their large polarization. 260,266,281,282,284,420−423 A high ΔP (60 μC cm −1 ) is obtained in NBT and BF based by reducing the P r through chemical substitution with other perovskite end members to form a relaxor with an ultraslim P−E loop. To some extent therefore, the advantage of a high intrinsic polarization end member such as BF is weakened. Intermediate ΔP ∼ 40 μC cm −2 values are observed for AFEs such as AN-and NN-based materials but P max is often limited as compositions exhibit polarization saturation as a function of applied field. Figure 27c compares W rec vs η for a wide range of compositions. ST-based materials display the best η (∼90%) due to their linear-like dielectric behavior. For BT-, NBT-, and lead-based materials, η varies with the material composition since it is a function of many factors. Dielectric loss associated with defects such as V O .. play a role but primarily at high field and high frequency, energy is dissipated during the transition to a field induced long-range ordered state which is manifested by the opening of the P−E loop. If the transition is smeared over for the operational range through alloying or doping to create a socalled "weakly coupled relaxor state", η > 90% can be achieved. 43,209,337 High leakage current and electrical conductivity are considered as major challenges in BF and KNN-based materials but are addressed by appropriate doping, e.g. donor doping to mitigate p-type conductivity in BF-based ceramics. 45 In addition, AFE based materials generally suffer from opening of the polarization loop above switching field to form a field induced FE phase which is detrimental to η. Compositional modifications to AN and NN ceramics aim not only to push the AFE-FE transition to higher field and stabilize the AFE phase but also to disrupt the long-range ordering in the field induced FE phase, thereby creating a slimmer portion of the P−E loop (higher η) than that being observed in unmodified materials. 366 Apart from the parameters discussed above (E max , ΔP, W rec , and η), temperature and frequency stability are also important for practical applications. In the future, high energy density ceramic capacitors will be placed closer to the core engine electronics to optimize the equivalent circuit resistance. Therefore, the temperature requirement for energy storage ceramics is anticipated to increase. According to the white paper "Multilayer Ceramic Capacitors for Electric Vehicles" published by Knowles capacitors in 2017, 424 the explosive development of EVs has prompted the appearance of new 200°C-stable C0G type I dielectric ceramic capacitor on the market. However, these materials still do not fulfill the required high power/ voltage, energy density, and temperature requirements (∼250°C ) to facilitate use near-engine. Better frequency stability from 100 Hz to 100 kHz is required to reduce power fluctuations when capacitors are used for DC/DC conversion for battery charging and DC/AC conversion for propulsion. 425−429 Enhanced frequency stability also enables the capacitor to be compatible with diodes and thyristors for power switching and control. 430 The temperature and frequency stabilities of many high energy density ceramics are evaluated, as shown in Figure 28. Only two compositions to date deliver W rec > 3.5 J cm −3 up to 250°C, 0.57BF−0.33BT−0.1NN 43 and 0.78NBT−0.22NN. 297 Most other compositions either do not sustain or do not have properties reported >200°C. Typical issues associated with operating at higher temperature include, widening of the P−E loop or early breakdown due to high leakage current and electric field and/or temperature-induced phase transitions. High leakage currents above 200°C typically arise from oxygen vacancy diffusion. 45 Most compositions have been shown to deliver W rec at a few hundred Hz but higher frequencies (>kHz) are rarely reported. Wang and co-workers, for example, discussed frequency stability from 10 −2 to 10 2 Hz for 0.57BF-0.3BT−0.13BLN with W rec ∼ 8 J cm −3 and η ∼ 81% at 400 kV cm −1 . All electroceramics for capacitors appear to deliver a charging−discharging speed at or faster than 1 μs. Short times of τ 0.9 ∼ 0.15 μs (90% of energy discharge in 0.15 μs) were reported by Li and co-workers in BT

Ceramic Multilayers and Films
3.2.1. Ceramic Multilayers. Ceramic MLs are fabricated by a series of processing steps which include slurry preparation, tape-casting, screen printing, lamination, cosintering, and termination, as shown in Figure 29. 18,176,422,431,432 This fabrication technology is a powder-based approach that accommodates scale-up from laboratory research to commercial manufacturing. The market of ceramic MLs ∼ $5.3 billion in 2017 but will reach ∼ $7.8 billion by 2024 for electronic applications, including but not limited to mobile phones, laptops and motor vehicles. 433 Advanced high energy density ceramic MLs, based on AFEs and RFEs materials, are being developed to facilitate power electronics within hybrid electric vehicles which require higher W rec and operating temperature. Simultaneously, research into low cost internal electrodes is required so that the highest performant ceramics can be developed. The energy storage properties for different ceramic MLs are summarized in Table 12.
Several lead-based AFE ceramic MLs have been reported using different internal electrodes. A giant power density ∼2000 kW cm −3 and W rec ∼ 3 J cm −3 was obtained in Pb-(Zr 0.95 Ti 0.05 ) 0.98 Nb 0.02 O 3 MLs using Pt as internal conductive electrode. 434 Optimized performance of W rec ∼ 3.  Figure 31a,b, as well as demonstrating good temperature stability from 75 to 175°C . 438 NBT−0.45SBT ceramic MLs have also been reported to exhibit W rec ∼ 9.5 J cm −3 with η ∼ 95% at 720 kV cm −1 , 222 which were further improved by forming a solid solution with a third perovskite end-member, 10%BMN to give W rec ∼ 18 J cm −3 with η > 90% at 1000 kV cm −1 , by Ji and co-workers, Figures 31c,d. 337 The highest among all lead/lead-free ceramic MLs, W rec ∼ 21.5 J cm −3 at ∼1030 kV cm −1 , however, was achieved for textured NBT−0.3SBT ceramic MLs. 439 Texturing was achieved through the use of <111>-oriented ST platelets which reduced fieldinduced strain at high field thus enhancing the BDS greatly. Combination of texturing with alloying with a third end member (BMN) in NBT−SBT may well represent an exciting path to achieve yet higher energy densities.
Wang, Reaney, and co-workers have employed a range of different chemical dopants and alloying additions to investigate BF-(B,S)T-based ceramic multilayers. 34,45,90,276,416 In 2018, Nd-doped BF-0.3BT was reported to exhibit W rec ∼ 6.74 J cm −3 (more than 3 times higher than the bulk value) and η ∼ 77% at 540 kV cm −1 with a layer t of 33 μm. 90 On the other hand, alloying BF−0.3BT with 8 mol % Nd(Zr 0.5 Zn 0.5 )O 3 resulted in W rec ∼ 10.5 J cm −3 with η of 87% at 700 kV cm −1 with a dielectric layer t of ∼17 μm, Figure 17. 34 Further studies focused on promoting electrical homogeneity, which was considered to prevent conductive pathways developing in these composition, thereby avoiding the breakdown at high field and facilitating the  Figure 32. 276 3.2.2. Ceramic Films. Higher BDS and W rec compared to MLs fabricated through powder-based technology have been reported for ceramic films deposited on LaNiO 3 /Si (100) or Pt/ Ti/SiO 2 /Si substrates by physical vapor or chemical deposition techniques, such as radio frequency magnetron sputtering, 449 spin coating, 101,450,451 pulsed laser deposition, 452 and chemical solution deposition. 453 The BDS of ceramic films is significantly improved due to the reduction of t (<1 μm) often attributed to fewer defects (grain boundaries) and/or pore/void concentration. Not only have higher figures of merit been reported for ceramic films, but several researchers have proposed novel underlying mechanisms (beyond reduction in defect density) behind the enhancement, such as the formation of polymorphic nanodomains. 452 However, nonpowder based techniques are difficult to scale up into MLCCs, which may constrain these extraordinary results to lab-based fundamental research rather than promising practical output for commercial exploitation. The energy storage properties for ceramic films are summarized in Table 13.
Lead-based ceramic films have been studied heavily in the past decade using different preparation methods, particular for PLZT.    In the past few years, there has been increased focus on leadfree ceramic films due to concerns over toxicity of PbO. NBT and BF based relaxor compositions have dominated research and have competitive or even superior energy storage performance to lead-based films. Mn doped NBT ceramic films on a LaNiO 3 /Si (100) substrates with t = 1.2 μm were reported to exhibit excellent W rec ∼ 30.2 J cm −3 under E max ∼ 2310 kV cm −1 . 455 Similar properties, W rec ∼ 33.3 J cm −3 under E max ∼ 2300 kV cm −1 , were also obtained for Fe doped NBT− K 0.5 Bi 0.5 TiO 3 ceramic film. 456 Recently, even higher W rec ∼ 50.1 J cm −3 , η ∼ 63.9% accompanied by fast charge−discharge speed (∼210 ns) were achieved simultaneously at ∼2200 kV cm −1 in relaxor 0.6NBT−0.4 Bi(Ni 0.5 Zr 0.5 )O 3 films. 457 In addition, an ultrahigh W rec ∼ 112 J cm −3 with η ∼ 80% was reported by Pan and co-workers in BF−BT−ST ceramic films recently ( Figure  34). Polymorphic nanodomains with competitive rhombohedral and tetragonal phases with competitive free energy were considered critical for the extraordinary electrical properties. BF was chosen as a main component due to its large spontaneous polarization. BT was introduced to form a solid solution to encourage coexistence of rhombohedral and tetragonal phases and finally ST was incorporated to further disrupt the long-range polar coupling and induce polymorphic nanodomains. By tuning the ratio of BF, BT and ST, a highly disordered composition is produced with rhombohedral and tetragonal nanodomain. The experimental observations were validated by phase-field simulations in the optimized composition, 0.20BF−0.25BT−0.55ST. 452 Apart from these perovskite lead-free relaxor candidates, HfO 2 -based ceramic films have also been explored and demonstrate promising energy storage properties, stabilities/reliabilities, scalability, and integration. W rec ∼ 46 J cm −3 with excellent temperature stability (up to 175°C ) and cyclic fatigue resistant (up to 10 9 time) was reported by Park in a 9.2 nm thick Hf 0.3 Zr 0.7 O 2 film 458 and W rec ∼ 63 J cm −3 with η ∼ 85% were realized in 50 nm thick Al doped HfO 2 ceramic films with excellent temperature and frequency stability. 459 3.2.3. Summary of State-of-the-Art in Ceramic MLs and Films. The energy storage performances, W rec , η, and ΔP, between bulk ceramics, ceramic MLs, and ceramic films are shown in Figure 35. The highest W rec (∼130 J cm −3 ) and E max (∼5800 kV cm −1 ) are obtained in ceramic films, followed by ceramic MLs (∼21 J cm −3 and E max ∼ 1000 kV cm −1 ) and bulk  ceramics (W rec (∼12 J cm −3 and E max ∼ 650 kV cm −1 ) and scale primarily with t of the dielectric layer (Figure 35d). W rec in ceramic films is also improved by higher ΔP (up to ∼120 μC cm −2 ) with respect to ceramic MLs (up to ∼70 μC cm −2 ) and bulk ceramics (up to ∼60 μC cm −2 ), as shown in Figure 35b. η for bulk ceramics, ceramic MLs and ceramic films varies significantly with composition and relates to factors such as, energy dissipation through a field induced transition to a longrange polar state, domain switching, polarization rotation, and leakage current relating to the presence of V O .. and associated defect dipoles.
From a commercial perspective, energy storage performance of lead-free ceramic MLs has improved significantly in the past few years with BF and NBT based ceramic MLs now rivalling lead-based ceramic MLs, delivering W rec > 15 J cm −3 at ∼1000 kV cm −1 . However, the selection of inner electrodes for these materials is limited to Pt, currently too costly for mass production. Commercial focus therefore, currently remains mainly on modified BT compositions which are compatible with Ni, Ag and Ag/Pd electrodes depending on composition and pO 2 during fabrication. A huge stride forward in the industry would be the development of a low cost equivalent to Ag/Pd

STRATEGIES FOR IMPROVING ENERGY STORAGE PROPERTIES
The review of the state-of-the-art of ceramics, MLs, and films presented in section 3 clearly points to a set of criteria that are required to optimize energy storage performance. Though some of these have been alluded to in section 3 as part of the review of the state-of-the-art, it is worth collating these principals for RFE  Chemical Reviews pubs.acs.org/CR Review and AFE materials to act as guide for future materials development.

Optimization through an Induced Relaxor State
The most commonly utilized strategy to optimize energy storage properties is through inducing a relaxor state within a system that contains highly polarizable ionic species. It is typically carried out through strategic doping or alloying to form a pseudoternary solid solution. Levin and co-workers proposed that long-range ferroelectric correlations can be effectively "blocked" by using designed dopants with enhanced local  Chemical Reviews pubs.acs.org/CR Review polarizability. 498 If this concept is married to dopant strategies to induce or maintain a homogeneous electrical microstructure, ΔP and BDS can be optimized leading to a large W rec . These two simple precepts can be applied to most systems; e.g., a frequency dispersion of ε r was observed after doping xBi 2/3 (Mg 1/3 Nb 2/3 )O 3 (B 2/3 MN) into BT ceramics, along with reduction on maximum dielectric constant (ε m ) and associated temperature (T m ). 209 After fitting the ε r and T m using modified Curie−Weiss law (as follows) (17) γ was found to be within the range of 1.61 for BT−0.06B 2/3 MN ceramics with a Burns temperature ∼154°C. In addition, XRD revealed a transformation from tetragonal to an average pseudocubic structure as increasing x concentration, coupled with a reduction in P r , confirming a relaxor state at room temperature, Figure 36a,b.
An optimum W rec ∼ 4.55 J cm −3 at 520 kV cm −1 was recorded for BT−0.06B 2/3 MN ( Figure 36 452 In summary, the inferior energy storage performance of all FEs can be improved by forcing a RFE state through strategic doping or alloying, Figure 38. 34,66,90,92,199,204,272 P max is often unsaturated in RFE and increases with E which means ΔP is not only a function of the slimness of the P−E loop but also of the applied E. RFEs are therefore, among the most promising candidates for capacitors in power electronics (E max > 300 kV cm −1 ).

Optimization of Antiferroelectrics
Many recent publications have focused on optimization of energy storage in lead-free AN-based materials, 347,353 although similar strategies date back to early studies of lead-based PLZT and PLZST AFE ceramics. 118,126,143−145 The most comprehensive study of AN AFE ceramics was performed by Lu and coworkers in which they used A and B-site substitutions to develop, Ag 0.97 Nd 0.01 Nb 0.80 Ta 0.20 O 3 which yielded W rec ∼ 6.5 J cm −3 at 370 kV cm −1 with η∼ 71%, Figure 39. 366 In their study, Lu and co-workers defined several key points required to optimize AN-based ceramics:  Chemical Reviews pubs.acs.org/CR Review i) Optimization of P max through local strain/field coupling around the smaller (with respect to Ag) Nd ion the A-site and its compensating V A , Figure 39a.
ii) Stabilization of the AFE structure through a combination of Nd and Ta doping which leads the induced AFE/FE transition to higher fields. Figure 39(a). iii) Inducing a slim hysteresis curve in the field induced region of the P−E loop. This was also achieved, through Nd doping that disrupted polar and antipolar coupling which manifested itself as a decrease in domain width from ∼1 to 0.5 μm and streaking of ±1/4(001) c superstructure reflections in electron diffraction patterns for x = 0.03, Figure 40. All the above maximize the area of the P−E loop to the left of the curve in the positive quadrant and thus optimize W rec , Figure  39b,c.
Stabilization of the AFE structure was also confirmed by Firstprinciples calculation and Ginzburg−Landau−Devonshire (GLD) phenomenology, as illustrated in Figure 41. Figure 42 summarizes the properties for many AFE systems. Most conventional AFEs exhibit low W rec (∼1.5−2 J cm −3 ) and η (∼40%), which can be optimized to W rec > 3 J cm −3 and η > 50% (enhanced-AFEs) by strategies described in the work of Lu and others. 348,[350][351][352]357,359,363,366 Similar values of ΔP (30−40 μC cm −2 ) are found for AFEs and enhanced-AFEs which reflects an intrinsic limitation of AFE materials, i.e. when antiparallel polar coupling is fully switched to polar under electric field, P max reaches saturation and is difficult to enhance unlike for RFEs (section 4.1). η is also difficult to further improve (>80%) due to the hysteresis above the AFE/FE switching field. However, at intermediate electric fields (∼300 kV cm −1 ), much higher ΔP and W rec are obtained for AFEs compared with RFEs, indicating that AFEs are more suitable for low/intermediate-voltage energy storage applications.     233 in which SiO 2 coating inhibits grain growth, thereby modifying the microstructure and reducing DC current leakage. The effect of SiO 2 layer thickness on BT particles has been systematically investigated (Figure 43a-c). The highest W rec (∼4.8 J cm −3 ) and η (∼99.1%) were obtained for 20 wt % SiO 2 coated BT, as shown in Figure 43d−f. 88 Other coating materials such as Al 2 O 3 and La 2 O 3 , 80,184 are also reported to optimize energy storage properties, as listed in Table  14.

Layered Structure.
Layer-structures composed of multiple materials have been reported to optimize energy storage properties and are typically tape cast, followed by lamination. Both BDS and ε r are optimized with the final properties related to the type of electroceramic material and the thickness of each layer. The BDS of BT-based ceramics was enhanced to >300 kV cm −1 by laminating layers between BT−x wt % SiO 2 layers (t ∼ 20 μm) and BT layer (t ∼ 25 μm). 441 ε r decreased but this was compensated by an increased BDS with increasing SiO 2 concentration in the BT−x wt % SiO 2 layers. ST + L i 2 C O 3 ( t ∼ 5 0 μ m ) a n d 0 . 9 3 N B T − 0.07Ba 0.94 La 0.04 Zr 0.02 Ti 0.98 O 3 (t ∼ 33 μm) layered structure were also fabricated via tape-casting with improved W rec ∼ 2.72 J cm −3 at 294 kV cm −1 . 502 Enhanced W rec ∼ 2.41 J cm −3 at 237 kV cm −3 was also obtained for layer-structure ceramics with ST + Li 2 O 3 (t ∼ 50 μm) and NBT−0.06BT (t ∼ 50 μm), Figure  44a.b. 442 The interface between the ST + Li 2 O 3 and the NBT− 0.06BT layer was further investigated using the finite element analysis. BDS was improved by reducing the breakdown paths between the ST + Li 2 O 3 and the NBT−0.06BT layer, Figure 44c−e 442 with electrical field redistribution and interface blocking playing essential roles. 503 BDS was also influenced by the difference in ε r and thickness ratio between the adjacent layers.

Lead-Based Energy Storage Ceramics
Lead-based ceramics have great potential as energy storage materials in modern microelectronics where high voltage and temperature are required, such as in pulsed power and power electronic applications. Lead-based AFE-type ceramics exhibit extremely high energy density but optimizing BDS, η and minimizing electrostrain is problematic. Low BDS (<300 kV cm −1 ) is often attributed to the volatilisation of lead/lead oxide which leads the formation of lead vacancy (V pb .. ) and V O .. that results in current leakage. Such issues may be partially solved by a combination of improved processing and dopants but achieving the values of BDS observed in lead-free materials has proved elusive. The low η in lead-based AFE-type ceramics (<80%) is mainly a result of opening of the hysteresis loop at high field due to the stabilization of a field induced FE phase. This results in a change in crystal class from tetragonal (AFE T )  The lack of popularity in researching lead-based compared with lead-free materials in the academic community has meant that exploration of novel systems is rather limited, but there are, for example, interesting mixed Pb-and Bi-based systems with high ε r and a spontaneous polarization that would mirror some of the design principles adopted in lead-free ceramics, particularly in solid solutions which combine AFEs and relaxor end members. In addition, further work is required to understand crystal structure and phase transition behavior. Many systems have incommensurate modulations and their influence on AFE/FE switching needs to be explored further using in situ XRD and Raman (temperature/electric field), as well as utilizing advanced aberration corrected TEM to study the local structure.

Lead-Free Energy Storage Ceramics
Lead-free candidates, including BT, ST, BF, KNN, NBT, AN and NN-based systems, are extensively studied and summarized in this review. Research into lead-free materials far outweighs that in lead-based, due to how the potential environmental legislation surrounding manufacturing and the end use of leadbased products has influenced funding bodies and awards. As a result, the optimization of energy storage properties has progressed rapidly in the last 5 years. Successful strategies to improve properties include, disrupting long-range polar coupling particularly if the average ionic polarizability is increased or unaffected, construction relaxor feature (PNRs) in FEs and AFEs, enhancing E g and as a consequence E a , reducing the total electrical conductivity and promoting electrical homogeneity through the use of strategic dopants to modify defect chemistry. If these strategies are married with a reduction in the dielectric layer thickness, high values of W rec ∼ 20 J cm −3 and η ∼ 90% can be achieved. Recent work on texturing of ceramic MLs has also proved successful in enhancing W rec but the complexity of this approach may inhibit commercial uptake. However, the two overriding issues with the majority of lead-free compositions, particularly those whose W rec are >10 J cm −3 are (i) the need to find an effective low cost internal electrode system that permits their commercial exploitation (currently almost all ML data is quoted with Pt internal electrodes) and (ii) pushing their operating window to >200°C and >100 Hz. Interestingly, electrostrain, a major drawback in lead-based materials, is broadly speaking not an issue in most of the lead-free RFEs and AFEs since the measured values of strain are often significantly lower (<0.2%) even at high fields. As with lead containing ceramics, there are only a few comprehensive investigations of the energy storage mechanisms which require high field in situ studies to be performed. A greater understanding of the role of defect chemistry, doping and alloying is also required, particular on how this influences, E g , resistivity and electrical homogeneity, and thus the W rec and η, thermal stability and cyclic reliability. In addition, in commercial MLCCs, the ripple current, equivalent series R, failure mode, voltage rating, the reliability in high humidity need to be evaluated and explored.

Glass Ceramics
Glass-ceramics have the advantages of facile manufacture, high W rec , ultrahigh η (low energy dissipation), ultrafast charge− discharge speed, excellent temperature/frequency stability. However, there are still challenges/problems. Generally, increasing the volume fraction of the crystal phase will increase the ε/P but decrease BDS. It is critical to balance the ε r and BDS to obtain the highest W rec . The mechanism of crystallization and control of crystal phase/microstructure is still ambiguous, which should be further investigated using, advanced TEM and in situ XRD/TEM as a function of applied field and temperature. Furthermore, although the theoretical W rec (>15 J cm −3 , due to Chemical Reviews pubs.acs.org/CR Review ultrahigh BDS, >1100 kV cm −1 ) of glass-ceramics are much higher than other bulk ceramics and even MLs (15−20 J cm −3 ), the measured/calculated W rec by P−E loops and discharging processes is low (<2 J cm −3 ) due to the lower applied electric fields. As a result, we recommend using the same test method (P−E loops and discharging process) to evaluate the practical energy storage performance for glass-ceramics, consistent with other dielectrics. Author Contributions ‡ G.W., Z.L., and Y. L. contributed equally to this work.

Notes
The authors declare no competing financial interest.

Biographies
Dr. Ge Wang is currently working as a postdoctoral research associate in the Functional Materials and Devices group at the University of Sheffield (UK). He obtained his Ph.D. degree from the Department of Materials at the University of Manchester in 2017, specifically on functional and structural behavior of lead-free ferroelectrics. His research focuses on ferroelectric, piezoelectric, energy density ceramic capacitors, lithium-ion batteries, high entropy oxide materials, and structural analysis using synchrotron X-ray diffraction.