Large electrocaloric effect in lead-free ferroelectric Ba_0.85Ca_0.15Ti_0.9Zr_0.1O₃ thin film heterostructure

Electrocaloric (EC) effects of large thermal changes across the phase transitions in ferroelectric (FE) materials when subjected to electric field stimuli promise eco-friendly and energy-efficient solid-state cooling technologies.^1–4 Theoretically, the choice of an ideal EC material depends on the parameters that quantify EC effects, namely, the adiabatic temperature change ΔT and the isothermal entropy change ΔS (or isothermal heat Q). At a temperature T, ΔS and ΔT of a FE material due to an applied electric field E may be generally obtained from the Maxwell relation $∂S∂ET=∂P∂TE$ using^1,2

where ρ and C_p are the mass density (ρ) and the specific heat capacity of the material, respectively, and E₁ and E₂ are the initial and final applied electric fields, respectively. It is noted that, in Eq. (2), the value of C_p has been assumed to be constant with no temperature and field dependence, as typically accepted in the literature (references in Table I). The polarization gradient $∂P/∂T$ is defined as the pyroelectric coefficient of the material. Equations (1) and (2) imply that the temperature dependence of polarization is coupled with a thermal change on varying the electric field. This indirect method has been successfully used to predict the performance of EC materials in the literature.¹ 

In recent times, artificially engineered FE heterostructures, in particular, epitaxial thin film heterostructures, are considered more desirable compared to their bulk ceramic counterparts due to their higher EC performances⁵ and inherent advantages for easy integration in microelectronic chip-cooling applications^6,7 with higher efficiency energy-recovery strategies.⁸ The recent surge in thin film electrocalorics was initiated in 2006 following the breakthrough report of “giant” EC effects (|∆T| = 12 K at |∆E| = 480 kV cm⁻¹ at 495 K) in PbZr_0.95Ti_0.05O₃ (PZT) thin films.⁹ The success of thin films was attributed to their high breakdown fields that allowed the application of large electric fields (|ΔE| ≈ 400 kV cm⁻¹–1200 kV cm⁻¹) across the film thicknesses to drive much higher EC effects than had been previously possible in bulk materials (|ΔE| < 50 kV cm⁻¹).¹ Furthermore, due to the effective clamping of the thin film material to the underlying-layer or substrate, especially in epitaxial thin films, it was possible to change the discontinuous first-order phase transition observed in bulk FE materials to a second-order continuous transition (thus reducing the hysteresis losses),¹⁰ which have been used to tune giant EC effects in FE thin films.^7,11 Electromechanical response via epitaxial strain engineering has been utilized to enhance EC effects by bringing separate transitions into close proximity,¹² or in order to create transitions in incipient FE materials,¹³ all implying that EC effects in films might not necessarily be similar to those in bulk materials.

In the last decade, giant EC effects have been indirectly estimated in thin films of a vast array of lead-based perovskites as listed in Table I.^14–19 However, due to the environmental concerns regarding the toxicity of lead, research has been redirected in the search for lead-free materials with comparable properties to replace the lead-based materials in technological applications. However, to date, the EC performance of lead-free perovskite thin films (see Table I)^20–22 or polymeric materials such as P(VDF-TrFE) (|∆T| = 12.0 K at |∆E| = 2093 kV cm⁻¹ at 353 K)²³ have not been able to meet the required performance level for replacing lead-based materials in commercial applications.

In terms of lead-free FE materials, the perovskite oxide, Ba_0.85Ca_0.15Zr_0.1Ti_0.9O₃ (BCZT), is particularly promising due to its significant dielectric properties (dielectric constant ∼3000 to 8000) and ferroelectric properties (remanent polarization P_r ≈ 10 µC cm⁻²–15 µC cm⁻² and coercive field E_C ≈ 1.5 kV cm⁻¹–3 kV cm⁻¹) near room temperature, in combination with a large piezoelectric coefficient (≈620 pCN⁻¹),^24–27 which is comparable to lead-based Pb_xZr_1−xTiO₃ (PZT) materials.²⁸ Enhanced polarization properties have been demonstrated in the specific composition of bulk BCZT ceramics, which is near its morphotropic phase boundary where BCZT is easily electrically poled and switchable upon application of external stress or electric fields.^29,30 Recently, enhanced FE properties have been reported in epitaxial BCZT (001) thin films grown on conductive Nb-doped (001)-SrTiO₃ single-crystal substrates (P_r = 21.3 µC cm⁻² at E_C = 60 kV cm⁻¹)³¹ and in epitaxial La_0.67Ca_0.33MnO₃/BCZT heterostructures (P_r = 10.3 µC cm⁻² at E_C = 220 kV cm⁻¹) grown on MgO (100) substrates.³² In addition, Shirsath et al. reported a novel interfacial charge-induced mechanism to observe a giant EC effect (|∆T| = 23 K at |∆E| = 620 kV cm⁻¹ at 370 K) in bilayer thin films of (Bi_0.5Na_0.5)TiO₃–BaTiO₃ (BNBT) and Ba(Zr_0.2Ti_0.8)O₃–(Ba_0.7Ca_0.3)TiO₃ (BCZT), while single-layer BCZT thin films exhibited nominal EC effects (|∆T| = 2.3 K at |∆E| = 620 kV cm⁻¹ at 370 K).³³ All these reports have raised the stakes for investigating EC effects in epitaxial BCZT thin films, which have thus far not been investigated.

In this work, we report the observation of a large EC effect (|∆T| = 13.5 K under |∆E| = 1000 kV cm⁻¹ at 430 K) across the ferroelectric–paraelectric phase transition in the lead-free perovskite BCZT thin film. An epitaxial BCZT thin film capacitor was fabricated using La_0.7Sr_0.3MnO₃ (LSMO) top and bottom electrodes on single-crystal SrTiO₃ (STO) (100) substrates using a pulsed laser deposition (PLD) technique. In-depth structural analyses using x-ray diffraction and x-ray spectroscopy are found to reveal the single-crystalline nature and stoichiometric composition of the heterostructure. Cross-sectional electron microscopy imaging revealed the cube-on-cube epitaxial growth and defect-free interfaces in the BCZT/LSMO heterostructure. Both temperature dependent x-ray diffraction and dielectric measurements revealed a broad electro-structural phase transition near 430 K in the BCZT/LSMO heterostructure. From detailed theoretical analyses of the experimental data, it is confirmed that the phase transition around 430 K is second-order in nature, unlike the first-order transition observed in the bulk BCZT materials. Thermodynamic analyses of polarization revealed extremely broad adiabatic temperature change ΔT(T) curves over a wide range of temperatures (330 K < T < 480 K) resulting in enhanced relative cooling powers, which are higher than those reported so far in most electrocaloric materials. We propose that an interfacial strain-induced enhanced tetragonal distortion of the BCZT layer gives rise to these hitherto unobserved enhanced EC effects in the BCZT/LSMO heterostructure system. The work provides fundamental understanding into the ferroic phase transitions in epitaxial lead-free thin films and provides a key insight into the design of artificial heterostructures with novel and enhanced EC properties.

An epitaxial BCZT thin film heterostructure using LSMO top and bottom electrodes was fabricated on a single-crystal SrTiO₃ (STO) (100) substrate using a commercial pulsed laser deposition (PLD; Neocera Pioneer 120 Advanced) system. Briefly, high-purity ceramic targets of La_0.7Sr_0.3MnO₃ and Ba_0.85Ca_0.15Ti_0.9Zr_0.1O₃ were prepared from commercially bought high purity (99.99%) powders from Alfa Aesar via solid state reaction followed by cold pressing and sintering at 1200 °C. The prepared targets were characterized for phase purity, structure refinement, and composition, as detailed in the supplementary material. The LSMO and BCZT targets were then sequentially ablated using a KrF excimer laser (Lambda Physik, λ = 248 nm, frequency = 10 Hz, fluence = 30 kJ m⁻²) inside a deposition chamber equipped with a multi-target carousel that allowed for the in situ deposition of multilayers with clean interfaces. A distance of 5 cm was maintained between the substrate and the targets during the depositions. Prior to growing the LSMO layer, the STO substrate was annealed inside the PLD chamber at 800 °C under a background oxygen pressure (pO₂) of 500 mTorr for 2 h. In the optimized synthesis process, an initial layer of LSMO was deposited on an STO substrate at 800 °C under a background pO₂ of 10 mTorr. This initial LSMO layer acted as the bottom electrode during the polarization of the BCZT thin films. The subsequent BCZT layer was deposited at 750 °C with pO₂ of 0.1 Torr. A shadow mask was used during the BCZT layer deposition to preserve an open access to the LSMO bottom electrode. After the BCZT layer deposition, top LSMO electrodes of 200 µm in diameter and 50 nm thickness were deposited using a shadow mask at 750 °C under a pO₂ of 10 mTorr. After deposition, the PLD chamber was flooded with pure oxygen (500 mTorr), and the samples were gradually cooled down to room temperature (∼4 h). The thicknesses of the LSMO and BCZT layers were kept constant at ∼100 nm. While several samples were deposited to optimize the growth conditions, here, we only present the results from the optimized BCZT/LSMO sample.

The crystallinity and crystallographic orientations in the heterostructures were characterized by x-ray diffraction (XRD) with a Rigaku Smart Lab 9 kW XG diffractometer equipped with a five-axis goniometer and a temperature variable thin film sample stage using collimated parallel beam Cu Kα radiation (λ = 1.5406 Å). Thermal conductive Apiezon H-grease was used to fix the samples on the thin film temperature stage of the XRD and care was taken to avoid sample misalignment. X-ray photoelectron spectra (XPS) were measured using a commercial Omicron (model 1712-62-11) spectrometer. The data were collected at room temperature using non-monochromatic Al Kα (1486.7 eV) x-ray source operating at 150 W (15 kV and 10 mA). Cross-sectional transmission electron microscopy (TEM) sample of thickness ∼100 nm was prepared by FEI Nova 600 Dual beam focused ion beam (FIB/FEG-SEM). A protective Pt-layer was used to preserve the features of the heterostructure during FIB milling for TEM sample preparation. High-resolution TEM (HRTEM) was performed in FEI Tecnai F20 (equipped with Oxford X-Max EDS detector) operated at 200 kV. Scanning transmission electron microscopy (STEM) with EDS mapping was also carried out to investigate the chemical composition of the heterostructure. DigitalMicrograph was used to analyze the HRTEM images and the simulated corresponding fast Fourier transform (FFT) images. The dielectric permittivity vs temperature measurements were carried out in an inert sealed bespoke chamber with a heating and cooling rate of 2 °C min⁻¹ (using 336 Temperature Controller, Lakeshore), while the dielectric properties were recorded between frequencies of 1 kHz and 1 MHz using an impedance analyzer (4294A, Agilent Technologies). The ferroelectric polarization measurements of the fabricated BCZT thin film capacitor (using LSMO top and bottom electrodes) were performed at different temperatures using a commercial Precision LC Ferroelectric tester (from Radiant Technologies, Inc.) equipped with a microprobe station. During the hysteresis measurements, a constant standard bipolar input profile was used with a 100 ms period. The leakage current densities (J_L) in the BCZT thin film capacitor were measured at different temperatures by applying a stress voltage of 10 V for a soak period of 100 ms at each voltage step. The heating was carried out using a custom-made heater setup to measure polarizations at different temperatures. The polarization data were collected independently from several (more than 20) top electrodes keeping the same bottom electrode, and the results were found to be consistent. The FE hysteresis loops reported here have been corrected for contributions from leakage currents.

The single crystalline nature of the BCZT and LSMO phases in the BCZT/LSMO heterostructure is evidenced from the XRD θ–2θ pattern shown in Fig. 1(a). In all cases, only strong (00l) (l = 1, 2, and 3) diffraction peaks of the tetragonal BCZT phase (c = 4.012 Å and a = b = 4.005 Å, see Fig. S1 of the supplementary material) are observed along with the (l00) (l = 1, 2, and 3) peaks of the pseudo-cubic perovskite LSMO phase (a = 3.87 Å, JCPDS 01-089-4461) and the single-crystal STO (100) substrate (cubic, a = 3.905 Å), confirming the cube-on-cube epitaxial growth, with no traces of impurity peaks within the resolution limits of the XRD. Due to the close proximity of the BCZT (002) and the LSMO and STO (200) peaks, a close-up view of the BCZT/LSMO/STO heterostructure and the single-layer LSMO/STO thin film is shown in the inset (I) of Fig. 1(a), where the dotted vertical line marks the position of the BCZT (200) peak from the bulk BCZT XRD data (see Fig. S1 of the supplementary material). As compared to the unstrained bulk BCZT (002) peak, the occurrence of the BCZT (002) peak at lower 2θ value in the BCZT/LSMO heterostructure implies the elongation of the out-of-plane lattice parameter (a_per) of the BCZT unit cell. Inset (II) in Fig. 1(a) shows the asymmetric XRD 2θ–ω scan (or detector scan) about the BCZT (111) plane in the heterostructure. The occurrence of the (111) peak at higher 2θ value as compared to the bulk position (marked by dotted vertical line) suggests a contraction of the in-plane lattice parameter (a_par) of the BCZT unit cell in the film. This is easily understood since the BCZT unit cell (tetragonal, a = b = 4.005 Å, c = 4.012 Å, see Fig. S1 of the supplementary material) may experience an in-plane compressive strain (i.e., out-of-plane tensile strain) in order to match the slightly smaller lattice parameter of the underlying LSMO layer (a = 3.894 Å), as calculated from the XRD θ–2θ scan in the inset (I) in Fig. 1(a).³⁴ From the XRD symmetric and asymmetric scans shown in Fig. 1(a), the calculated values for a_per = 4.028(6) Å and a_par = 3.935 (5) Å give a slight out-of-plane tensile strain ε_per = (a_per − c)/c ≈ 0.4% and correspondingly a large in-plane compressive strain ε_par = (a_par − a)/a ≈ −1.8% in the BCZT layer in the BCZT/LSMO heterostructure. This results in an order of magnitude higher out-of-plane tetragonal distortion (i.e., a_per/a_par − 1) of ∼2.3% in the BCZT unit cell in BCZT/LSMO as compared to the bulk BCZT unit cell (c/a − 1 = 0.1%). Since BCZT is a highly electrostrictive material, the larger tetragonal distortion of the BCZT layer in BCZT/LSMO could drastically affect its polarization and the EC properties.³⁵ The epitaxial growth of the individual layers of BCZT and LSMO on the STO substrate was further confirmed from XRD ϕ (phi or azimuthal) scans³⁴ performed about the STO (110), LSMO (110), and BCZT (110) crystallographic planes, respectively, as shown in Fig. 1(b). These crystallographic planes of the BCZT, LSMO, and STO phases were specifically chosen for the XRD ϕ scans so that there is no contribution from underlying phases in the BCZT/LSMO heterostructure. The repeated occurrence of sharp peaks at regular intervals of 90° confirms the fourfold symmetry of the individual layers and cube-on-cube epitaxial growth of the heterostructure. The rocking curve (ω scan), as shown in Fig. 1(c), performed about the BCZT (100) plane showed an extremely narrow width (FWHM = 0.2°), which further confirmed the in-plane orientation of the BCZT layer in the BCZT/LSMO heterostructured epitaxial film. X-ray photoelectron (XPS) spectra were collected ex situ from the BCZT/LSMO heterostructure to determine the composition and the site occupancy of the ions in the BCZT unit cells. Figures 1(d) and 1(e) show the core level XPS spectra of Ba 3d, Ti 2p, Ca 2p, and Zr 3d for the BCZT layer and the core level XPS spectra of La 3d, Mn 2p, Sr 3d for the LSMO layer in the BCZT/LSMO heterostructure, respectively. The composition of the BCZT/LSMO heterostructure as calculated from the XPS analyses matched the nominal composition of BCZT and LSMO phases, as also confirmed from the XRD analyses [Fig. 1(a)]. The details of the fitting parameters of the XPS spectra for the BCZT/LSMO heterostructure are given in Table S1 of the supplementary material. The cationic site occupancy ratio (B/A site) for the perovskite ABO₃-type BCZT unit cell as calculated from the XPS spectra in Fig. 1(d) was found to be (Ti+Zr/Ba+Ca) = 0.98, which is very close to the ideal ratio of 1 from the BCZT unit cell. The above XRD and XPS analyses confirmed the single-crystalline nature and the stoichiometric composition of the BCZT/LSMO heterostructure.

Cross-sectional microstructure of the BCZT/LSMO/STO heterostructure revealed sharp and flat interfaces with uniform thicknesses of the individual layers, as illustrated in the representative TEM image in Fig. 2(a). Detailed HRTEM images of the heterostructure showed scattered threading dislocations running from the LSMO–STO interface and terminated as “saw-tooth” like columns in the BCZT layer as shown in the inset of Fig. 2(a), indicating the highly strained state of the individual layers in the BCZT/LSMO heterostructure. The HRTEM images of the interfaces in the BCZT/LSMO/STO heterostructure in Figs. 2(b) and 2(c) show atomically sharp and flat phase boundaries with continuous lattice fringes from the bottom to top layers, respectively. The images clearly demonstrate the single crystalline nature and the cube-on-cube epitaxial growth morphology of the BCZT/LSMO heterostructure. Similar HRTEM images captured at different locations along the interfaces showed no evidence of any structural defects such as lattice misfits or twin planes, as shown in the representative inverse FFT image in Fig. 2(d) and the linear-dotted FFT pattern in Fig. 2(e) of the LSMO/STO interface [captured from a region marked with a red box in Fig. 2(b)]. The HRTEM image in Fig. 2(f) clearly exhibits the single-crystalline defect-free growth of the BCZT layer captured from a region away from the BCZT-LSMO interface. Lattice parameters calculated from the HRTEM image of the BCZT layer confirmed the tetragonal distortion of the BCZT unit cell (i.e., a_per > a_par), which is also evidenced in the tetragonal nature of the FFT pattern shown in the inset of Fig. 2(f). From the TEM analyses, it is concluded that the BCZT layer and the LSMO layer have almost identical lattice parameters near the interface in the BCZT/LSMO heterostructure, suggesting a highly strained state in the tetragonal BCZT layer.

Figure 3(a) shows the temperature dependent XRD θ–2θ patterns near the BCZT (002), LSMO (200), and STO (200) peaks for BCZT/LSMO heterostructure as measured from 303 K to 633 K at regular intervals of 20 K (as shown in the scale bar). The XRD patterns in Fig. 3(a) have been purposefully shifted from right (633 K) to left (303 K) for visual clarity. The STO (200) peaks are deconvoluted into two peaks for contributions from Cu K_α1 and K_α2 lines. It is clearly observed from the XRD patterns in Fig. 3(a) that, while there is no change in the peak profiles or intensities of the STO (200) substrate peaks, a distinct change in the peak profile and intensity is observed in the BCZT (002) peak as detailed in the close-up view in the inset to Fig. 3(a). Since the in-plane lattice parameter (a_par) of BCZT could be strongly affected by the lattice mismatch of the underlying LSMO layer in BCZT/LSMO, we plot the in-plane lattice parameters (a_par) for the BCZT unit cell in the temperature range of 303 K–633 K in Fig. 3(b) for predicting the structural phase transition in the BCZT layer in BCZT/LSMO heterostructure. Note that the error bars in Fig. 3(b) are calculated from the full-width half maxima of the BCZT (002) peaks shown in Fig. 3(a). The room temperature lattice parameter of BCZT matches with that observed earlier in Fig. 1(a) for BCZT/LSMO. From the figure, it is observed that the lattice parameter increases steadily as the temperature increases from 303 K to 427 K; beyond that temperature, the lattice parameter does not change as drastically from 427 K to 633 K. From the temperature dependent XRD analyses, we can attribute this change in lattice parameter [as marked by linear fits in Fig. 3(b)] at the Curie temperature, T_C = 427 K, to the structural phase transition from tetragonal to cubic in the BCZT layer in the BCZT/LSMO heterostructure. The mean linear coefficient of thermal expansion (α) calculated from the slopes of the linear fits of the data in Fig. 3(b) (solid blue lines as guides to the eye) in the range T < T_C is α_(T<Tc) = 3.2 × 10⁻⁶ K⁻¹ and T > T_C is α_(T>Tc) = 1.0 × 10⁻⁶ K⁻¹, respectively, which is close to the values for BCZT ceramics.³⁶ Based on the temperature dependent XRD analyses, the crystal structures of BCZT unit cells with the tetragonal phase at 300 K (T < T_C) (a = b = 3.935 Å, c = 4.027 Å, α = β = γ = 90°) and the cubic phase at 503 K (T > T_C) (a = b = c = 4.024 Å, α = β = γ = 90°) were simulated, as shown in Sec. I of the supplementary material. The temperature dependent XRD analyses of the bulk BCZT target revealed the tetragonal to cubic phase transition at T_C = 353 K (see Fig. S2 of the supplementary material). The higher structural transition temperature of T_C = 427 K for the BCZT/LSMO heterostructure as compared to the bulk BCZT (T_C = 353 K) can be attributed to the higher degree of tetragonal distortion (∼2.3%) in the BCZT unit cell arising due to the epitaxial strain from the lattice mismatch of the underlying LSMO layer in BCZT/LSMO as compared to that in the unstrained bulk BCZT (0.1%) unit cell. The enhancement of T_C in the BCZT system has been reported earlier in BCZT nanowire structures with exceptionally high tetragonality.²⁷ To our knowledge, the observation of the structural phase transition has not yet been evidenced in epitaxial BCZT thin films, possibly due to the challenges in detecting the extremely small degree of distortion in the BCZT unit cell above and below T_C along with the presence of strong twinning in the lattice. Since the polarization direction of the tetragonal BCZT structure coincides with one of the three original cubic (100) directions, it is difficult to resolve the structural change using XRD.³⁷ Thus, the observation of structural transition as understood from Fig. 3 can be attributed to the oriented growth of the epitaxial heterostructured film with (100) orientation.

Figure 4(a) shows the temperature dependence of the dielectric constant ε_r′ (upper panel) and dielectric loss tanδ (lower panel) for the LSMO/BCZT/LSMO thin film capacitor as measured at different frequencies from 30 kHz to 1000 kHz, respectively. From Fig. 4(a), it is clearly observed that all the ε_r′ (T) curves show broad transition peaks at T_C = 428 K at all frequencies in the BCZT/LSMO heterostructure. The transition peak at T_C = 428 K is also determined from the linear fitting of the Curie–Weiss law as shown on the right-hand axis of Fig. 4(a) for 1000 kHz and found to be the same for all other frequencies. It is noted that the T_C in Fig. 4(a) exactly matches with that observed from the XRD analyses in Fig. 3; confirming that the dielectric transition is associated with the structural phase transition from tetragonal to cubic in the BCZT/LSMO heterostructure. However, the ε_r′(T) curves for BCZT/LSMO show much broader phase transition as compared to the ε_r′(T) curves for the bulk BCZT target (see Fig. S3 of the supplementary material), thus potentially increasing the operating temperature regime of the EC devices based on BCZT/LSMO heterostructure.¹ Furthermore, T_C = 428 K as observed in the ε_r′(T) curves for BCZT/LSMO is much higher than that observed in the bulk BCZT target (T_C = 352 K) (see the supplementary material), which could be again be attributed to the higher degree of tetragonal distortion (∼2.3%) in the BCZT unit cell in BCZT/LSMO as compared to that in the unstrained bulk BCZT (0.1%) unit cell.²⁷ Interestingly, the broad phase transitions and dielectric dispersions in Fig. 4 point to a second-order type phase transition across T_C in the BCZT/LSMO film as compared to the first-order phase transition typically reported in the bulk BCZT ceramics.³⁸ 

Figure 4(b) shows the leakage current density vs the electric field (J_L–E) curves for the BCZT/LSMO thin film capacitor (shown schematically in the inset with LSMO top electrodes of diameter 200 µm with area 31400 µm² and BCZT layer thickness 100 nm) measured at different temperatures below and above T_C (428 K). From Fig. 4(b), it is observed that the values of the J_L at the highest field of E = 1000 kV/cm at 300 K is nearly two orders of magnitude lower than those reported for BCZT thin films^27,39 and comparable to the low leakage properties of high quality epitaxial PLZT thin films.²⁸ The extremely low leakage currents in the BCZT/LSMO thin film capacitor as seen in Fig. 4(b) possibly imply that the ensuing polarization properties are not significantly affected by the leakage effects. Furthermore, the perfectly symmetric nature of the branches of the J_L curves can be attributed to the similar intrinsic properties (such as interface state densities and potential barrier heights) of the LSMO top and bottom electrodes of the BCZT/LSMO device structure, resulting from the high-quality defect-free epitaxial growth of the BCZT/LSMO heterostructure. We have reported earlier that LSMO top and bottom electrodes greatly reduce the interfacial polarization caused by space charges at the film/electrode interfaces, thus resulting in higher remanent polarization in perovskite films.²⁸ Figure 4(c) shows the temperature dependence of J_L as measured for different electric fields from 0 kV cm⁻¹ to 1000 kV cm⁻¹ for the BCZT/LSMO heterostructure. From Fig. 4(c), it is clearly observed that the leakage increases exponentially above T_C = 428 K, which further confirms the FE to paraelectric phase transition in the BCZT/LSMO heterostructure (as show by linear and exponential fits) about T_C = 428 K.

Figure 5(a) shows the ferroelectric hysteresis P(E) loops for the BCZT/LSMO heterostructure measured at different temperatures from 300 K to 505 K under the maximum driving electric fields of 1000 kV cm⁻¹. The well-saturated (saturation polarization, P_sat = 30.7 µC cm⁻²), square, and highly symmetric P(E) loop at 300 K in the BCZT/LSMO exhibits a large remanent polarization of P_r = 19.8 µC cm⁻², which matches well with those of the reported epitaxial BCZT films with the same composition.^31,40 However, at 300 K, the squareness of the P(E) loop (i.e., P_r/P_sat = 64%) and the coercive field of E_C = 280 kV cm⁻¹ in BCZT/LSMO are much higher than those reported for BCZT thin films and close to the reported value for epitaxial LCMO/BCZT heterostructure,^31,41–43 which can be attributed to large domain sizes in the high-quality single-crystalline BCZT/LSMO heterostructure and the defect-free nature of the interfaces, as observed earlier in their XRD and TEM analyses (Figs. 1 and 2), respectively. From Fig. 5(a), it is clearly observed that as the temperature increases, the P(E) loops become narrower accompanied by the continuous reduction of P_r and E_C values; while above T_C = 430 K, the P(E) loops show signatures of paraelectric phase transition. Interestingly, the P(E) loops in Fig. 5(a) at higher temperatures show low values of remanent polarization even in the paraelectric phase (i.e., T > T_C). Such P(E) loops are typically observed in relaxor FE materials including BCZT ceramics and thin films and have been attributed to the existence of random local polar nano-regions in the paraelectric phase of these materials.^25,33 The significantly low values of J_L at different temperatures as seen in Fig. 4(c) further confirm that the shape of P(E) loops in Fig. 5(a) is not affected from leakage effects in the BCZT/LSMO heterostructure. Figure 5(b) shows the temperature dependent polarization P(T) curves under different external electric fields from 0 kV cm⁻¹ to 1000 kV cm⁻¹ obtained from the upper branches (first quadrant) of the P(E) hysteresis loops in Fig. 5(a). The temperature dependent pyroelectric coefficients $∂P∂TE$ as calculated from the $∂P∂TE$ vs T curves for different applied electric fields (as shown in Fig. S4 of the supplementary material); indicated peaks around T_C = 430 K due to the phase transition from FE to paraelectric phase in the BCZT thin film and the values are higher than those reported for bulk BCZT near the phase transition.^44,45 The temperature dependent isothermal entropy change ∆S(T) shown in Fig. 5(c) and the adiabatic temperature change ∆T(T) shown in Fig. 5(d) were calculated from the P(T) curves as plotted in Fig. 5(b) for different fields starting from 0 kV cm⁻¹ to 1000 kV cm⁻¹ using Eqs. (1) and (2) and considering the density (ρ = 6050 kg m⁻³) and assuming the specific heat capacity as constant (C_p = 540 J kg⁻¹ K⁻¹) for the BCZT thin film (C_p varies by ∼8% in 320 K < T < 480 K for single crystals).⁴⁶ The maxima for the ∆S(T) and ∆T(T) curves span over a relatively broad temperature range (330 K–480 K) as seen in Figs. 5(c) and 5(d), which could be indicative of a second-order type phase transition of the BCZT thin film. A large EC response is recorded with the maximum ΔT = 13.5 K and ΔS = −16.9 J kg⁻¹ K⁻¹ at T_C = 430 K for ΔE = 1000 kV cm⁻¹ for the BCZT/LSMO heterostructure. To the best of our knowledge, the ΔT measured indirectly in the BCZT/LSMO heterostructure constitutes a record among all the lead-free oxide perovskite materials reported to date (see Table I).

In order to investigate the nature of the FE phase transition, the P(E) data were fitted to the Ginzburg–Landau mean field theory for polarization where the Landau free energy, f (P, E), can be expressed as^55–57 

where $aT$⁠, $bT$⁠, and $cT$ are the material constants. Close to the phase transition, the condition for minimization of free energy gives

where a(T), b(T), and c(T) can be determined by fitting the isothermal P(E) data using Eq. (4), as shown in Fig. 6(a). From the mean field model, the conditions that determine whether a phase transition is second-order are as follows:⁵⁸ 

The phase determining coefficients a(T) and b(T) as shown in the inset of Fig. 6(a) are consistent with the above conditions, implying that there is indeed a second-order type phase transition present in the BCZT/LSMO heterostructure thin film.

Since the broadness of the ∆S(T) curves in Fig. 5(c) has been attributed to the second-order type phase transition in BCZT/LSMO, which is unlike the first-order transition in bulk BCZT ceramics, we used the phenomenological universal curve method proposed by Franco et al.^59–61 for the BCZT/LSMO system. Following this postulation, the universal curve was constructed by normalizing the ∆S(T) curves at different fields in Fig. 5(c) with their maximum values [i.e., ∆S′ = ∆S(T)/∆S_max(T)] and rescaling the temperature axis as

where $TR1$ and $TR2$ are the two reference temperatures corresponding to the half maximum of ∆S(⁠$TR1$⁠) = ∆S(⁠$TR2$⁠) = ∆S(T_C)/2. It is interesting to note that, using this phenomenological construction of the universal curve, all the normalized ∆S′(T) curves under electric fields from 333 kV cm⁻¹ to 1000 kV cm⁻¹ collapse onto a single universal curve as shown in Fig. 6(b), once again confirming that the phase transition is second-order in nature for the BCZT/LSMO heterostructure.

Finally, we use the theoretical approach proposed by Law et al.⁶² for the quantitative determination of the order of FE phase transition in the BCZT/LSMO heterostructure. In this process, we determine the model-independent exponent (n) from the field dependence of the isothermal entropy change |∆S|⁶², which is represented as a power law of the field E as

with an exponent n that is dependent on field and temperature. It can be locally calculated as

If the above analyses yields n > 2, then it confirms a first-order phase transition, and if it yields n < 2, it implies a second-order phase transition. Figure 6(c) shows the three-dimensional plots of field and temperature dependence of the entropy change (|∆S|) as obtained from the polarization measurements for the BCZT/LSMO heterostructure in a temperature range from 300 K to 505 K under varying applied electric fields from 0 kV cm⁻¹ to 1000 kV cm⁻¹. From Fig. 6(c), it is observed that the EC surface exhibits a second-order phase transition with a “caret-type” behavior,⁶² showing that the temperature evolution of |∆S| is gradual for all electric field values and likewise there is no abrupt change in the field dependence for any of the isotherms. Figure 6(d) shows the field and temperature dependence of the values of the exponent n for BCZT range from 300 K to 505 K under varying applied electric fields from 0 kV cm⁻¹ to 1000 kV cm⁻¹. For the entire range of field and temperature, the maximum value of n is 1.14 (i.e., n < 2), which confirms the second-order phase transition in BCZT/LSMO because for the first-order phase transition the value of n > 2. The above analysis further confirms the conjecture that the phase transition in BCZT/LSMO is, indeed, second-order in nature.

Finally, an efficient prototype EC material should exhibit not only large values of the maximum ΔT but also large values of the full width half maximum of the ΔT(T) curves (δT_FWHM) achieving significant relative cooling capacity [RCP(T)] defined as^1,2

Based on the broad ∆T(T) curves as shown in Fig. 5(d) along with the record high value of maximum ∆T = 13.5 K, the calculated RCP(T) for the BCZT/LSMO heterostructure [RCP(T) = 1901 K² at T_C = 430 K for ∆T = 13.5 K under ∆E = 1000 kV cm⁻¹] using Eq. (8) was found to be much higher than those reported for EC materials in the literature (see Table I).⁶³ In fact, the RCP(T) value of the BCZT/LSMO heterostructure is comparable to the celebrated EC material PLZT [RCP(T) = 1912 K² at T = 400 K for ∆T = 25 K under ∆E = 990 kV cm⁻¹], which makes the BCZT/LSMO system highly attractive as a lead-free material system for EC applications. The enhanced RCP(T) observed in BCZT/LSMO is plausibly due to the enhanced polarization arising from the large tetragonal distortion of the BCZT layer and the second-order phase transition in the system as have been discussed in detail. It is noted that, in the future, we will investigate the effect of thicknesses of BCZT and LSMO layers on the electrocaloric properties and the order of phase transition in BCZT/LSMO heterostructures.

In summary, EC effects were investigated in an epitaxial BCZT thin film heterostructure fabricated using LSMO top and bottom electrodes on a single-crystal STO (100) substrate by using the PLD technique. In-depth x-ray analyses and HRTEM images revealed the single-crystalline nature, cube-on-cube growth morphology, and the defect-free interfaces in the BCZT/LSMO heterostructure. A broad second-order-type electro-structural phase transition near T_C = 430 K was observed in the BCZT/LSMO heterostructure from both temperature dependent x-ray diffraction and dielectric measurements. Thermodynamic analyses of the polarization and electric field revealed a large EC effect (|∆T| = 13.5 K and |∆S| = 17 J kg⁻¹ K⁻¹ under |∆E| = 1000 kV cm⁻¹ at T_C = 430 K) in the BCZT/LSMO heterostructure. Extremely broad adiabatic temperature change ΔT(T) curves over a wide range of temperatures (330 K < T < 480 K) resulted in enhanced relative cooling powers in the heterostructure, which are higher than those reported so far in most electrocaloric materials. Theoretical modeling of the experimental data confirmed the second-order nature of the phase transition in the BCZT/LSMO heterostructure, unlike the first-order transition observed in bulk BCZT materials. We propose that an interfacial strain-induced enhanced tetragonal distortion of the BCZT layer gives rise to these large EC effects in the BCZT/LSMO heterostructure system. The work adds on to the development of EC thin film heterostructures in lead-free materials for potential applications in eco-friendly solid-state cooling devices.

D.M. and A.D. conceptualized the work and designed the synthesis. A.B., S.C., C.O., D.M., and A.D. performed the experiments and collected the data. Measurement equipment in S.K.-N. lab was used for the dielectric and ferroelectric measurements. Y.Y.T. and N.B. performed the cross-sectional microscopy analyses. A.B. and S.C. equally contributed to this work. All co-authors contributed to the writing of this manuscript.

See the supplementary material for the structure refinement, compositional analyses, temperature dependent XRD, and dielectric measurements of the BCZT target and XPS fitting parameters for BCZT/LSMO heterostructures.

D.M. acknowledges the funding from the Technical Research Center, Department of Science and Technology, Government of India (Grant No. AI/1/62/IACS/2015) and Science and Engineering Research Board (SERB) Starting Research Grant, Government of India (Grant No. SRG/2019/000387). A.D. acknowledges funding from SERB-Ramanujan Fellowship (Award No. SB/S2/RJN-057/2017). N.B. acknowledges funding from EPSRC through the New Investigator Award No. EP/S016430/1. S.K.-N. acknowledges funding from an European Research Council Starting Grant (No. ERC-2014-STG-639526, NANOGEN).

The authors declare no conflict of interest.

The data that support the findings of this study are openly available at https://doi.org/10.17028/rd.lboro.13580621.