Co:BaTiO3/Sn:BaTiO3 Heterostructure Thin‐Film Capacitors with Ultrahigh Energy Density and Breakdown Strength

Ferroelectric (FE) capacitors exhibiting ultrahigh power densities are widely utilized as electrostatic energy storage devices in pulsed electronic devices. One approach to maximize the discharge energy density (Ud) of capacitors is to increase the breakdown strength (Eb) accompanied with high maximum polarization (Pm) while suppressing the energy loss. However, the inverse relationship between Eb and Pm challenges the simultaneous enhancement of Eb and Ud. To overcome this limitation, FE/relaxor FE (RFE) heterostructure capacitors composed of Co‐doped BaTiO3 (BTCO) and Sn‐doped BaTiO3 (BTS) epitaxial thin film layers to decouple the Eb and Pm values are fabricated and the simultaneous enhancement of the Eb and Ud up to 7.9 MV cm−1 and 117 J cm−3, respectively is achieved. The high Eb and Ud values can be attributed to the suppression of the leakage current at the BTCO/BTS interface, a narrower hysteresis loop contributed by the BTS, and high Pm and Eb from the BTCO layer. Additionally, the BTCO/BTS heterostructure capacitors exhibit excellent fatigue endurance of up to 108 cycles and are thermal stable even at 160 °C. Through properly designing the FE and RFE layers, thermally stable and reliable FE/RFE heterostructure capacitors exhibiting high Ud and Eb can be realized.

DOI: 10.1002/aelm.202201141 medical applications. [1][2][3][4][5][6][7][8][9][10][11] The high power density and reliability of electrostatic capacitors are superior to those of electrochemical energy storage devices such as batteries, supercapacitors, and fuel cells. However, the low discharge energy density (U d ) of electrostatic capacitors, which is at least one order of magnitude lower than that of electrochemical capacitors, mainly hinders their wider energy storage applications that require miniaturization of electronic devices. [12] The U d and efficiency of an electrostatic capacitor (η) can be expressed as where P r is the remnant polarization, P m is the maximum polarization, E is the applied electric field, dP is the change in polarization, and U loss is the energy loss, i.e., the area inside the P-E hysteresis loop. [13,14] One strategy to maximize the U d value is to simultaneously increase the breakdown strength (E b ) and P m -P r values. Previous approaches to achieve high E b , P m -P r , and a narrow P-E hysteresis loop have mostly focused on developing relaxor ferroelectric (RFE) material-based solid solutions, because the RFEs generally possess a low P m -P r value and narrow P-E hysteresis loop. [1,[15][16][17] Several studies have focused on developing ferroelectric (FE)based solid solutions that exhibit ultrahigh U d from increased P m -P r but moderate E b values. [12,13,[18][19][20][21][22][23] For instance, a BiFeO 3 -BaTiO 3 -SrTiO 3 solid solution exhibited a high U d value of 112 J cm −3 with a moderate E b value of 4.9 MV cm −1 . [18] The Smdoped BiFeO 3 -BaTiO 3 solid solution exhibited an ultrahigh U d value of 152 J cm −3 with an E b of 5.2 MV cm −1 . [12] These capacitors exhibit such high U d values when the applied electric field approaches the E b value. However, as the magnitude of the applied electric field increases, the leakage current exponentially increases, leading to device breakdown. [24] A high E b value alone does not guarantee high U d because of the trade-off between E b and P m. An increase in E b should be achieved without sacrificing the P m value. However, the E b and P m values are inversely correlated with the dielectric constant (ε) [E b ∝ ε r -0.65 and P m ∝ ε r ], and the simultaneous enhancement of the P m -P r and E b values is highly challenging. [16] Ferroelectric (FE) capacitors exhibiting ultrahigh power densities are widely utilized as electrostatic energy storage devices in pulsed electronic devices. One approach to maximize the discharge energy density (U d ) of capacitors is to increase the breakdown strength (E b ) accompanied with high maximum polarization (P m ) while suppressing the energy loss. However, the inverse relationship between E b and P m challenges the simultaneous enhancement of E b and U d . To overcome this limitation, FE/relaxor FE (RFE) heterostructure capacitors composed of Co-doped BaTiO 3 (BTCO) and Sn-doped BaTiO 3 (BTS) epitaxial thin film layers to decouple the E b and P m values are fabricated and the simultaneous enhancement of the E b and U d up to 7.9 MV cm −1 and 117 J cm −3 , respectively is achieved. The high E b and U d values can be attributed to the suppression of the leakage current at the BTCO/BTS interface, a narrower hysteresis loop contributed by the BTS, and high P m and E b from the BTCO layer. Additionally, the BTCO/BTS heterostructure capacitors exhibit excellent fatigue endurance of up to 10 8 cycles and are thermal stable even at 160 °C. Through properly designing the FE and RFE layers, thermally stable and reliable FE/RFE heterostructure capacitors exhibiting high U d and E b can be realized.
One strategy for increasing the E b value of dielectric thin films without sacrificing the P m is to develop heterostructure thin-film capacitors using materials with different dielectric constants. [7,[25][26][27][28][29][30] In the heterostructures, the electric field across the layer with a high ε will be lower than that in the layer with a low ε, which will delay the polarization saturation that is extremely beneficial for a large change in the polarization value (dP). Furthermore, space charges are generated at the interface of the layers, and under an applied electric field, the electrons injected from the electrode are trapped at the interface, which efficiently suppresses the leakage current. [28] Sun et al. [29] reported that the heterostructure interface hinders the growth of the electric tree, i.e., the leakage current pathway, which leads to dielectric breakdown.
In this study, we propose a BaTiO 3 (BTO)-based heterostructure thin-film capacitor composed of FE Co-doped BaTiO 3 (BTCO) and RFE Sn-doped BaTiO 3 (BTS) to decouple the E b and P m values. We sequentially deposited FE BTCO and RFE BTS layers onto (111)-oriented Nb:STO substrates, because the (111)-oriented thin films were reported to exhibit high ε r and P m values. [31,32] For the top FE layer of the heterostructure, a cobalt-doped BaTiO 3 film (BTCO) was chosen, because acceptor dopants such as Co, Fe, and Ni induce the formation of polymorphs (tetragonal and hexagonal phases) in BTO, which is related to the high P m . [33][34][35][36][37][38][39] The BTS was chosen as the bottom RFE layer because Sn doping in BaTiO 3 was reported with a high U d (92.5 J cm −3 ) and E b (5.2 MV cm). [17] The FE/RFE heterostructure consisting of BTCO and BTS can exhibit high P m (due to polymorphs in both the BTS and BTCO layers), low P r (due to the RFE layer), and high E b values (due to the suppression of leakage current at the BTCO/BTS interface). Thus, by substituting Ti with two different cations (Co and Sn), FE and RFE layers were grown from Pb-free BaTiO 3 and developed ultrahigh-energy-density FE/RFE heterostructure capacitors.

Fabrication of Thin-Film Capacitors
BTS, BTCO, and BTCO/BTS epitaxial thin films were grown on (111)-oriented Nb:SrTiO 3 single-crystal substrates using the pulsed laser deposition technique. A KrF excimer laser (Coherent, COMPex 102) with a wavelength of 248 nm was used to ablate ceramic targets under an oxygen pressure of 10 mTorr. Ablation was performed at a laser repetition rate and fluence of 5 Hz and 1.5 J cm −2 , respectively. The substrate temperatures for depositing BTS and BTCO thin films were set to 800 and 750 °C, respectively. The BTCO and BTS targets were custom designed and prepared using a conventional solid-state method (Toshima Manufacturing Co. Ltd, Japan). The BaCO 3, TiO 2, and SnO 2 powders were used for synthesizing the BTS target, and BaCO 3, TiO 2, and Co 3 O 4 powders were used for synthesizing the BTCO target. To stabilize the hexagonal BTO phase in the BTCO layer, a nonstoichiometric ceramic target was used with 10% Ba deficient (Ba 0.9 Ti 0.9 Co 0.1 O 3 ). [40] Also, Ba deficiency can arise during the deposition. After deposition, the films were cooled to room temperature (RT) at a cooling rate of 10 °C min −1 . The thicknesses of the BTS and BTCO thin films were ≈120 and 80 nm, respectively. The optimization conditions for thin-film growth are provided in the Supplementary Information. The X-ray diffraction (XRD) patterns of BTCO and BTCO/BTS heterostructure thin films deposited at various thicknesses are shown in Figure S1 (Supporting Information). Patterned circular Pt top electrodes with thicknesses of 30 nm were deposited using an e-beam evaporator. Capacitors with diameters of 30 and 40 µm were used for the electrical analyses. A schematic of the Pt/BTCO/BTS/Nb:STO heterostructure capacitor is shown in Figure 1a.

Characterizations
The crystal structures of the films were analyzed by synchrotron X-ray diffraction (9C beamline, Pohang Light Source) using an X-ray beam with an energy of 10 keV and high-resolution transmission electron microscopy (HRTEM, JEM-ARM200F). Reciprocal space maps around the (113) and (211) planes of Nb:SrTiO 3 were recorded to identify the crystal structures and estimate the lattice parameters of the films. Further investigations on the crystal structure were also obtained from the selected area electron diffraction (SAED) patterns. Scanning electron microscopy (SEM; Hitachi, S-4700) and atomic force microscopy (AFM; XE-100, Park Systems) were used to investigate the surface morphology. AFM and SEM images are shown in Figures S2 and S3 (Supporting Information). The layer thicknesses were measured using cross-sectional SEM images and were confirmed using TEM images. Capacitance versus DC bias (C-V ) plots of the capacitors were measured using an impedance analyzer (Keysight Technologies, E4980A). The frequency-dependent dielectric constant and tangent loss were obtained with an AC voltage of 500 mV in the frequency range of 10 3 to 10 6 Hz. The leakage current-E and P-E hysteresis loops were measured using a ferroelectric tester (Radiant Technology, Precision Multiferroic). The P-E hysteresis loops were obtained under electric fields with various amplitudes at a frequency of 10 kHz until the capacitors reached breakdown. The P-E hysteresis loops of several capacitors to evaluate the statistics of the capacitors using a two-parameter Weibull analysis were also measured. To test the thermal stability of the capacitors, hysteresis loops were recorded under an electric field with an amplitude of 2.75 MV cm −1 for temperatures ranging from RT to 160 °C.

Results and Discussion
The XRD patterns of the BTCO, BTS, and BTCO/BTS heterostructure films are shown in Figure 1b. Clear (111) reflections from the substrate and thin films are observed in the patterns, indicating the (111) orientation of the epitaxial thin films. The crystalline quality of the epitaxial thin films was further investigated using the XRD rocking curves, as shown in Figure S4 (Supporting Information). The lattice parameters and crystal structures of the BTCO and BTS films were www.advelectronicmat.de estimated from the (113) and (211) reciprocal space maps shown in Figure 1c,d (Supporting Information). BTS was crystallized in a rhombohedral structure, whereas BTCO was crystallized in a tetragonal crystal structure. The estimated lattice parameters are given in Table S1 (Supporting Information). The rocking curve of the BTS epitaxial thin films exhibits a narrow peak with a full width at half maximum (FWHM) of 0.035°, which is placed on top of a broad peak with a FWHM of 0.51°, indicating the presence of strained and strain-relaxed regions with d-spacing spreading. However, the rocking curves of the BTCO and BTCO/BTS thin films were much broader than those of the BTS thin film. The relatively wider FWHM values of the BTCO film (0.69°) and BTCO/BTS heterostructure (0.44°) suggest that both films possess shorter structural coherence than the BTS film.
The oxygen defects arising from Ba deficiency and valency mismatch between Co 3+ and Ti 4+ ions can be attributed to short structural coherence and strain-relaxed growth. The valence state of Co was determined using the XPS spectra shown in Figure S5 (Supporting Information). Sn 4+ substitutes for Ti 4+ in BTS and oxygen defects arising from the valency mismatch do not exist in the BTS epitaxial thin film.
From the XRD pattern analyses of the BTCO thin film, we could not determine the presence of hexagonal (h)-BTCO because of the proximity of the (111) reflection from tetragonal (t)-BTCO and the (0006) reflection from h-BTCO. The (111) reflections from the BTCO, BTS, and BTCO/BTS heterostructures were found at 2θ values of 38.72°, 38.49°, and 38.69°, respectively. The corresponding lattice spacing (d 111 ) values of the BTS, BTCO, and BTCO/BTS films were 0.2294, 0.228, and 0.2282 nm, respectively, whereas that for bulk BaTiO 3 was 0.2348 nm. There is a considerable difference between the d 111 of BTS and that of the BTCO/BTS heterostructure, indicating compression along the out-of-plane direction with a shorter d 111 value. A single broad peak was observed in the BTCO/BTS heterostructure, suggesting that during the top BTCO deposition, the bottom BTS layer experiences additional interfacial tensile stress. The interfacial stress (σ) in BTCO/BTS film compared to the BTS film was also estimated using the following equation: [41] where E is Young's modulus, γ is Poisson's ratio, d f is the d-spacing of BTS film in BTS/Nb:STO, and d l is the d-spacing of the BTS layer in the BTCO/BTS heterostructure. The E and γ values of BaTiO 3 reported in the previous study are used for the calculation. [42] The calculated stress is ≈650 kPa. Owing to this internal stress, the BTS layer in BTCO/BTS heterostructure is much more relaxed with a shorter d 111 spacing, which impacts the magnitude of polarization and dielectric constant values of the BTCO/BTS heterostructures.
To investigate the presence of any additional phases, the BTCO and BTS layers in the BTCO/BTS heterostructure were analyzed using HRTEM and SAED.  Figure S6c (Supporting Information) also confirmed the presence of regions with different crystal structures. Thus, the FFT analysis and SAED patterns confirm that BTCO crystallized in the t-and h-phases.
The c-and a-axes lattice parameters of the BTS and BTCO layers in the BTCO/BTS heterostructure were estimated using the d 001 and d 110 values obtained from the FFT patterns (Figure 2b,c,e,f), as summarized in Table S2 (Supporting Information). The estimated c/a ratios of the BTCO and BTS films are 1.0194 ± 0.0005 and 1.0135 ± 0.0071, respectively. The d 0002 of the h-phase region of the BTCO layer was used to estimate the c-axis lattice parameter as 13.974 Å, which matches well with that of the bulk h-BTO. [38] It was reported that 10% Co doping in BTO triggers the transformation from the t-to h-phase in bulk BTO. [38] In this study, a 10% Co-doped BTO target was used, and epitaxial growth along the (111) direction stabilized h-BTCO domains in the tetragonal BTCO matrix instead of phase-pure h-BTO. Li et al. [39] reported that h-BTO stabilized in amorphous 0.92BaTiO 3 -0.08Bi(Mg 1/2 Zr 1/2 )O 3 thin films enhanced polarization values. We believe that the h-phase in BTCO allows us to enhance the P m and ε values of our BTCO/BTS heterostructure films. Figure 3a,b shows the E-and frequency-dependent ε r and tangent losses of the BTS, BTCO, and BTCO/BTS capacitors. As hypothesized, the BTCO/BTS heterostructure exhibits the highest ε r of 360 (at 10 kHz), presumably originating from the accumulation of free charges at the BTCO/BTS interface, whereas ε of the BTCO and BTS layers are 282 and 103, respectively. As the BTCO and BTS layers have different dielectric constants, free charges accumulate at the interface and eventually increase the dielectric constant of the BTCO/BTS system. The BTCO, BTS, and BTCO/BTS capacitors exhibit tangent loss values of 0.0314, 0.0227, and 0.0118 at 10 kHz, respectively, which supports our hypothesis of the suppression of leakage current at the BTCO/BTS interface.
As the free charges accumulated at the BTCO-BTS interface increased the dielectric constant, the quality of the interface and elemental distribution (to check if any elemental diffusion occurred at the interface) were investigated using TEM-EDS and are shown in Figure 3c,d. The composition of the film was mainly BTO; hence, it was difficult to distinguish each layer of the BTCO/BTS heterostructure using the HRTEM image. Therefore, we analyzed the elemental distribution, as shown in Figure 3d. Elemental mapping of Ba, Ti, Sn, Co, and O confirmed a clean and sharp interface. The BTCO was deposited at 800 °C, while the BTS film was at 750 °C. Because of this growth temperature difference, Sn and Co elements slightly diffused between the BTCO and BTS layers. Figure 4a shows the P-E loops of the BTS, BTCO, and BTCO/BTS films measured at 10 kHz before their electrical breakdown. The BTCO/BTS capacitor exhibited a high P m value of 68 µC cm −2 , whereas the P m values of the BTS and BTCO capacitors were 50 and 55 µC cm −2 , respectively. The higher P m value in the BTCO/BTS heterostructure film can be attributed to a combination of compression along the out-of-plane direction and the accumulation of free charges at the BTCO/ BTS interface. The epitaxial strain and crystalline structure are important factors in determining the P m values of BTCO, BTS, and BTCO/BTS heterostructure films.
The h-phase regions in the t-phase matrix of the BTCO layer can be attributed to the higher P m in the BTCO capacitor than that in the BTS. The compression along the (111) direction was higher in the BTCO film than in the BTS, which may be one of the reasons for the higher P m in BTCO. The RFE-BTS capacitor exhibited a low P r of 7.6 µC cm −2 , whereas the BTCO and BTCO/BTS heterostructure capacitors exhibited relatively higher P r values of ≈20 µC cm −2 due to their FE characteristics.
The P m -P r values and corresponding U d values at various applied electric fields are shown in Figure 4b,c. Despite the high P r value, the BTCO/BTS heterostructure exhibited the highest P m -P r value of 52 µC cm −2 and an ultrahigh U d of up to 117 J cm −3 . Such a high P m -P r value is attributed to the high P m value of BTCO/BTS capacitor. The BTS capacitor exhibited a P m -P r and U d of 47 µC cm −2 and 92.5 J cm −3 , whereas those of the BTCO capacitor were 43 µC cm −2 and 77 J cm −3 , respectively. The low U d value of the BTCO capacitor is due to an increase in P r in the high-electric field region above 3.5 MV cm −1 , accompanied by a moderate P m value, high FE hysteresis, and high tangent loss. As shown in Figure 4b   www.advelectronicmat.de the BTCO/BTS capacitor can be operated at a higher electric field of up to 7 MV cm −1 compared to 6.2 MV cm −1 for BTCO and 5.2 MV cm −1 for BTS capacitors, which is one factor that contributed to the highest P m -P r value in the heterostructure capacitors. It should be noted that to maximize the U d value, along with a high P m -P r value, the breakdown strength electric field should also be high.
The maximum electric field values set for the capacitors were limited by the breakdown strength, which was estimated using the classical Weibull analysis. The breakdown electric fields of the 15 capacitors were used for evaluation and fitted by Weibull analysis, as shown in Figure 4d. The BTCO/BTS capacitor has an excellent E b of 7.87 MV cm −1 , which is higher than that of BTS (5.97 MV cm −1 ) and BTCO (6.66 MV cm −1 ) capacitors. It is noteworthy that the BTCO capacitors have higher E b than BTS capacitors despite the high dielectric loss (0.0314). The higher coordination number of cations in hexagonal BTO rather than in rhombohedral or tetragonal BTO requires a higher electric field for breaking the MO (MBa/ Ti) bond, resulting in an increase in E b . [13,43] Because of the high E b of BTCO/BTS capacitor, it could operate at a high electric field of up to 7 MV cm −1 , resulting in a high U d value of 117 J cm −3 . The Weibull modulus β of ≈15 was obtained for the BTCO/BTS capacitor, ≈10 for the BTCO, and ≈21 for the BTS capacitors. A high β value indicates a small deviation from the central value and, therefore, better quality, and reproducibility. The E b of BTCO/BTS capacitors, as high as 7.87 MV cm −1 , can be primarily attributed to the suppression of the leakage current at the BTCO/BTS interface. In addition, the effect of thickness on the BTCO and BTCO/BTS heterostructure capacitors was also investigated. An increase in the thickness of the BTCO and BTCO/BTS thin films resulted in a drastic decrease in P m , P m -P r , U d , and E b , as shown in Figure S7 (Supporting Information).
To elucidate the origin of the high E b , the leakage current densities of the BTS, BTCO, and BTCO/BTS heterostructure films were analyzed, as shown in Figure 4e. The leakage current density of the BTCO/BTS heterostructure is extremely lower than BTCO and BTS capacitors. As discussed, the leakage current is a dominant factor in determining the E b and U d values, and the leakage current density characteristics differ for all three capacitors. The BTCO/BTS heterostructure capacitor exhibited the lowest leakage current and withstood a high electric field, whereas the BTCO film exhibited the highest leakage current density among the fabricated capacitors. The defect dipoles exhibit a significant effect on the leakage current as well as the energy storage performance. The A-site vacancies and B-site cation substitutions act as hole-trapping centers in ABO 3type perovskites and can enhance energy storage performance. The A-site vacancies reduce the hysteresis loss, resulting in a slim P-E hysteresis loop. [44] The A-site vacancies induced by Ba deficiency in the BTCO target are beneficial in suppressing the lossy hysteresis behavior. However, the leakage current in BTCO is higher than that in the BTS capacitor due to the multiple valencies of Co, which promotes charge migration. Through XPS analysis, we confirmed that Co exists with +2, +3, and +4 valence states in our BTCO film ( Figure S5, Supporting Information). The various Co valence state promotes the migration of charged carriers, mainly oxygen vacancies generated by Ba-deficiency. In addition, Ba vacancies can induce oxygen vacancies to preserve charge neutrality, increasing the leakage current. [3] In comparison, the BTCO/BTS film exhibited the lowest leakage current density (6.5 × 10 −2 A cm −2 at 4 MV cm −1 ), which is consistent with the low dielectric loss shown in Figure 3b. At an electric field of 1 MV cm −1 , the BTCO and BTS films exhibited similar leakage current densities; however, drastic variations occurred when the electric field was increased above 1 MV cm −1 . The leakage current abruptly increases for BTCO when E > 1 MV cm −1 ; however, no such abrupt variations occur in the heterostructure capacitor. When an external electric field is applied to the heterostructure capacitor, the electric field distribution across the layers in a heterostructure depends on the dielectric constant of each layer. The electric field across the layer with a high ε (BTCO layer, in our case) will be lower than that in the layer with a low ε (BTS layer), which will protect the layers from a breakdown due to the relatively lower electric field. In addition, the heterostructure interface acts as a barrier to suppress the formation of the electrical current leakage pathway. A detailed analysis of the I-V characteristics fitted using various conduction mechanisms is provided in Figure S8 (Supporting Information).
The energy conversion efficiencies η of BTCO, BTS, and BTCO/BTS capacitors are shown in Figure 4f. The η value of the BTCO/BTS heterostructure capacitor was as low as 76% at a lower applied electric field (<2 MV cm −1 ) compared with BTCO (95%) and BTS (92%). However, while the η values for BTCO and BTS capacitors steadily decrease with an increase in E, the η value of the BTCO/BTS heterostructure capacitor slightly increases, reaching 80% at a high E range from 2 up to 4 MV cm −1 . At E = 7 MV cm −1 , the η value of BTCO/BTS heterostructure is retained at 70%, compared to 47% of BTCO capacitor at E = 6.2 MV cm −1 and 78% of BTS capacitor at E = 5.3 MV cm −1 . BTS capacitors exhibit high η values due to their intrinsically low U loss linked to the intrinsic RFE characteristics. The FE BTCO layer exhibited high U loss and, therefore, low η values at a high E of 6.2 MV cm −1 . Thus, a high U d was achieved due to enhanced P m -P r and E b values of the BTCO/ BTS heterostructure capacitor.
The U d and E b values of the BTCO/BTS film are superior to those of our BTCO and BTS films and earlier reported heterostructure capacitors, as summarized in Figure 5a. The proposed BTCO/BTS heterostructure exhibited the highest U d and E b values among the various heterostructures reported in the literature. [22,25,27,29,30,[45][46][47][48][49][50][51] The heterostructure capacitor also shows competitive performance compared to the reported values of Pb-free solid solution and thin-film capacitors (Figure 5b). A further comparison of the U d and E b values with those of Pbbased solid solution thin films and multilayer thin films are shown in Figure S9 (Supporting Information).
For commercial applications, the thermal stability and cycling stability of the capacitors were investigated in detail. We measured the P-E hysteresis loops of the films at temperatures ranging from 20 to 160 °C (Figure 6a and Figure  S10, Supporting Information) at an applied E amplitude of 2.75 MV cm −1 . The corresponding U d and η values are shown in Figure 6b. With an increase in temperature, both www.advelectronicmat.de U d and η of the BTS and BTCO capacitors decrease because their P-E loops become lossy at higher temperatures ( Figure S10, Supporting Information). The BTCO and BTS capacitors were stable within the operating window of RT to 110 °C. The monotonic decreases in U d and η indicate that electronic conduction degrades the stability of single-layer capacitors. However, the BTCO/BTS bilayer capacitors exhibited enhanced thermal stability over a wider temperature range of RT-130 °C. We tested the heterostructure capacitor up to 160 °C. At 160 °C, the efficiency decreased by only 20%; however, the U d value drastically decreased from 30 to 5 J cm −3 . This drastic decrease in U d at 160 °C is due to the abrupt decrease in polarization, especially at ≈130 °C, closer to the Curie temperature of BTO.
The fatigue endurance of the capacitors was analyzed through the charge-discharge cycling process at an applied E amplitude of 1.86 MV cm −1 , as shown in Figure 6c,d. The P-E loops of the BTCO/BTS capacitors and BTS capacitors shown in Figure 6c and Figure S11 (Supporting Information), respectively, were nearly unchanged up to 10 8 cycles with excellent cycling stability. However, the BTCO capacitor endured cycling measurements only up to 10 5 cycles ( Figure S11, Supporting Information). The fatigue of capacitors usually originates from defects and space charges that pin domain movement. [63] Contrastingly, the multi-nanodomain structure of the BTS layer suppresses domain pinning; therefore, the BTS capacitors exhibit good reliability. However, the off-stoichiometry and oxygen defects in BTCO result in the poor fatigue endurance of BTCO capacitors. In the case of the BTCO/BTS heterostructure capacitor, the interface layer can prevent the diffusion of space charges and defects to pin the domain movement, resulting in excellent fatigue endurance. We also prepared BTCO/BTS capacitors with thicker BTCO (≈200 nm) and BTS (≈200 nm) layers; however, the U d and E b values were drastically reduced. We also prepared samples using thinner BTCO (≈50 nm) and BTS (≈70 nm) layers. In those cases, the capacitors were leaky. Therefore, a thorough  [22,25,27,29,30,[45][46][47][48][49][50][51] N corresponds to the number of heterostructure stacks. BTCO, Co-doped BaTiO 3 ; BTS, Sn-doped BaTiO 3 . b) Comparison of U d , Ε b values of BTCO/BTS capacitors with the reported lead-free solid solution and heterostructures (N > 2) based capacitors. [7,8,12,18,22,27,29,46,49,[52][53][54][55][56][57][58][59][60][61][62] www.advelectronicmat.de thickness optimization for each layer is required to suppress the leakage current and realize multilayer capacitors with even higher U d and E b values.

Conclusions
The lead-free BTCO/BTS heterostructure thin-film capacitors with an ultrahigh energy density of 117 J cm −3 and breakdown strength of 7.9 MV cm −1 were successfully fabricated. The outstanding U d and E b can be attributed to the following factors: (i) suppression of leakage current at the BTCO/BTS interface, (ii) narrow hysteresis loop of the RFE BTS component, (iii) compression along the out-of-plane direction, and (iv) high P m and E b of the FE-BTCO layer arising from the hexagonal phase of BTCO. Our heterostructure capacitor exhibited a high dielectric constant of 360 at a frequency of 10 kHz with a low dielectric loss of 0.0118. The energy conversion efficiency of the BTCO/ BTS heterostructure remained almost the same at ≈70%, even at an applied electric field of 7 MV cm −1 . Our heterostructure capacitor can also operate over a wide temperature range of up to 130 °C with excellent thermal stability. The fatigue endurance of the heterostructure capacitor was excellent, without significant degradation up to 10 8 cycles. Our structural design based on FE/RFE heterostructure capacitors effectively maximizes the discharge energy density and minimizes the energy loss with excellent fatigue endurance and thermal stability by suppressing the leakage current at the FE/RFE interface. Furthermore, multilayer capacitors fabricated using BTCO, and BTS layers with optimized thickness have the potential to result in even higher U d and E b values.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.