Introduction

The development of next-generation cooling technologies that are environmentally friendly and can replace traditional vapor-compression refrigeration systems, which significantly contribute to greenhouse gas emissions, has garnered increasing attention due to climate change and environmental sustainability concerns1. The electrocaloric (EC) effect in ferroelectrics presents a promising avenue for advanced solid-state cooling technologies, offering environmental compatibility, high energy efficiency, and potential for miniaturization2,3,4,5,6,7,8,9,10,11,12. Perovskite-type ferroelectrics (Fig. 1a) possess intrinsic spontaneous polarization, which can be reoriented in response to an applied electric field. This enables them not only to be suitable for applications such as piezoelectric actuators, dielectric energy storage, and ferroelectric memories13, but also solid-state cooling due to their EC effect. The EC effect manifests as either an isothermal entropy change (ΔS) or an adiabatic temperature change (ΔT) because of the polarization variation upon the application or removal of an electric field14, as illustrated in Fig. 1b, c.

Fig. 1: Design strategy for high intrinsic electrocaloric (EC) effect in ferroelectrics near room temperature.
Fig. 1: Design strategy for high intrinsic electrocaloric (EC) effect in ferroelectrics near room temperature.The alternative text for this image may have been generated using AI.
Full size image

a Schematic of perovskite-type ferroelectrics and the spontaneous polarization (Ps) vector. b Schematic illustration of the EC refrigeration cycle based on the EC effect. c EC effect involving entropy change and temperature change induced by polarization change. Schematic diagrams illustrating the local lattice structure, mesoscopic domain configuration, and the evolution of macroscopic permittivity, ferroelectric polarization, and EC effect across different phase transitions: d pure barium titanate (BT), e generally modified BT with a diffuse phase transition near room temperature, and f a newly modified BT that achieves a sharp phase transition, high polarization (change) and permittivity, and a giant intrinsic EC effect is expected near room temperature. g The density functional theory (DFT) theoretical estimation of lattice distortion after incorporating different modified ions at A (Ba2+) or B (Ti4+) sites in tetragonal BT. Here, the coordinate of the Ba element or the Ti element is assumed to be (x, y, z), the center coordinate of the oxygen octahedron is (x’, y’, z’), and Δy = y-y’ and Δz = z-z’. The Δy is perpendicular to the Ps direction, demonstrating the relative angle of lattice distortion, and Δz is parallel to the Ps direction, demonstrating the relative magnitude of lattice distortion.

The EC effect can be evaluated through ΔT and the electrocaloric strength (ΔT/ΔE). The Maxwell phenomenological expression for ΔT is given by15:

$$\Delta T=-\frac{T}{\rho }{\int }_{{E}_{1}}^{{E}_{2}}\frac{1}{{C}_{{{\rm{E}}}}}{\left(\frac{\partial P}{\partial T}\right)}_{{{\rm{E}}}}{{\rm{d}}}E,$$
(1)

where T represents temperature, typically near room temperature for EC cooling applications; P denotes polarization; E1 and E2 are the initial and final electric fields, respectively; ρ is the material density; and CE is the specific heat capacity, which shows minimal variation under a low electric field. Consequently, ΔT is primarily influenced by polarization and its rate of change with temperature ∂P/∂T, and is also positively correlated with the applied electric field. This implies that a high ΔT can be achieved under a substantial electric field, even if the intrinsic EC effect (i.e., ΔT/ΔE) is small. Since ΔT/ΔE quantifies the EC response per unit electric field, a higher value means the same cooling effect can be achieved with a lower field, enhancing both efficiency and practicality in EC cooling applications. ΔT can also be expressed as2:

$$\Delta T=\frac{1}{2{C}_{E}}{\alpha }_{0}T{P}^{2}$$
(2)

where α0 is a phenomenological coefficient. Hence, ΔT/ΔE can be derived using Eq. (2) and dP/dE = ε16:

$$\frac{\Delta T}{\Delta E}=\frac{1}{{C}_{{{\rm{E}}}}}{\alpha }_{0}{TP}({{\rm{d}}}P/{{\rm{d}}}E)=\frac{1}{{C}_{{{\rm{E}}}}}{\alpha }_{0}{TP}\varepsilon=\frac{1}{{C}_{{{\rm{E}}}}}{\alpha }_{0}{TP}{\varepsilon }_{0}{\varepsilon }_{{{\rm{r}}}}$$
(3)

where ε, ε0, and εr denote the absolute, vacuum, and relative dielectric permittivity (dielectric constant), respectively. It is evident that ΔT/ΔE is positively correlated with polarization and dielectric properties. Therefore, high P, ∂P/∂T, and εr are advantageous for obtaining high intrinsic EC properties in ferroelectrics. Among various ferroelectrics, ferroelectric films require very high driving electric fields14, and single crystals are prohibitively expensive. In contrast, ferroelectric ceramics or polymers offer a more practical and commercially viable solution for achieving the necessary cooling volumes, with ferroelectric ceramics being particularly promising6,7,8,9,10,17,18,19. Barium titanate (BaTiO3, BT)-based ferroelectric ceramics stand out as one of the most promising lead-free EC materials due to their high ferroelectric polarization, large dielectric constant, and environmental friendliness19,20,21,22,23,24.

Pure BT ceramic exhibits an impressive ΔT/ΔE ≥ 1.2 K·mm/kV near the Curie temperature (TC, ~120 °C)25, being attributed to the sharp phase transition induced by the highly ordered ferroelectric structure, which results in a steep change in polarization with temperature (∂P/∂T). The large ferroelectric domains (from hundreds of nanometers to more than one micrometer) observed well below TC contribute to this effect (Fig. 1d)20. However, its EC property is relatively low near room temperature due to a small polarization change rate (∂P/∂T) and limited dielectric response, induced by its stable ferroelectric structure at room temperature26,27,28. As depicted in Fig. 1e, previous studies to enhance room-temperature EC property generally involve shifting TC towards room temperature through chemical modification, such as high-content or aliovalent ions doping or substitution, to destabilize the ferroelectric phases by introducing lattice distortions with heterogeneous local structures at the atomic scale13,18,20. This will break the long-range ferroelectric orders and usually lead to the fine domains with the domain size decreasing to the nanoscale or even forming the polar nanoregions (below several nanometers)29, which in turn results in a smeared ferroelectric–paraelectric phase transition over a broad temperature range. Although this can improve the room-temperature EC property under a high electric field, it often results in a significant reduction in ferroelectric polarization and ∂P/∂T, leading to a suboptimal intrinsic EC effect. As illustrated in Fig. 1f, an ideal strategy to achieve a high intrinsic EC effect near room temperature involves not only shifting TC closer to room temperature, but also maintaining the well-ordered ferroelectric structure below TC, with moderate domain size and sharp phase transition to preserve high polarization and ∂P/∂T, respectively. This requires achieving a well-ordered local structure with minimal abnormal lattice distortions to maintain the relatively large domains after chemical modifications. This approach is expected to preserve the sharp phase transition while sustaining a giant dielectric constant.

To validate this strategy, we designed a ferroelectric ceramic system of Ba0.87Sr0.13TiO3-xBa0.87Sr0.13SnO3 (BST-xBSS) guided by density functional theory (DFT) calculations. This system shows well-ordered structural features with minimal lattice energy difference upon ion substitution. By adjusting the x, TC is shifted downward to near room temperature, accompanied by convergent ferroelectric phase transitions around a sharp TC dielectric peak. This preserves relatively high ferroelectric polarization and enables a giant dielectric constant, characterized by rapid value changes, thus contributing to a pronounced intrinsic EC effect near room temperature. Scanning transmission electron microscopy (STEM) observations reveal that the multiscale nanodomain configuration, along with a polarization distribution with gradual dipole rotation induced by the well-ordered local structure, is the key structure origin for the sharp TC dielectric peak, high ferroelectric polarization, drastic polarization change, and giant EC response.

Results

Chemical modification designed to achieve a well-ordered local structure

To achieve a well-ordered local structure while shifting TC downward to room temperature by chemical modifications, the modifying ions must have the same valence and similar ionic radii to those of Ba2+ (1.61 Å) and Ti4+ (0.605 Å) in the BaTiO3 perovskite structure. This ensures that significant lattice distortions or the introduction of vacancy defects into the local structure of BT are prevented, thus helping to preserve high intrinsic ferroelectric polarization and a pronounced phase-transition dielectric peak30. Additionally, the modification ions should create a small lattice energy difference to ensure minimal lattice distortions. Based on these assumptions, Sr2+ (1.44 Å) and Sn4+ (0.69 Å) are selected as the primary modification ions in comparison to other possible candidates. Figure 1g presents the theoretical estimation of lattice distortion in BT resulting from different ion substitutions at the A and B sites, while Supplementary Fig. S1 shows the lattice energy differences between adjacent ions occupying A or B sites, oriented either parallel or perpendicular to the Ps vector. The DFT results demonstrate that Sr2+ occupying A sites and Sn4+ occupying B sites introduce minimal disruption to BT’s ordered local structure compared to other substitutions. This low level of lattice distortion in BT supports a more uniform structural evolution, enabling a rapid and coherent phase transition across the ferroelectric matrix under thermal or electrical stimuli. Meanwhile, the minimal lattice distortion suggests a low ion positioning energy preference31, promoting a homogeneous distribution of the modification ions within the BT matrix. In contrast, dopants such as Na+/Bi3+ or Ga3+/Nb5+ on the Ba2+ or Ti4+ sites, respectively, exhibit significantly higher energy difference, either parallel or perpendicular to the Ps direction, leading to a selective ion positioning preference or even clustering of the ions. This accounts for a significant lattice distortion and the formation of local structural heterogeneity, ultimately disrupting long-range ferroelectric orders, leading to reduced ferroelectric polarization and a more diffuse phase transition27. Collectively, these findings highlight that Sr2+ and Sn4+ are effective modification ions for achieving a well-ordered local structure in BT with the desired effects.

Phase transitions and electrical properties

Figure 2a illustrates the temperature dependence of dielectric constant (εr–T) for BT and BST-xBSS ceramics. The primary goal of introducing Sr2+ and Sn4+ is to effectively shift the TC of BT toward room temperature, meanwhile preserving the well-ordered local structure and ferroelectric orders, due to the minimal lattice distortion induced by these ions (Fig. 1g and Supplementary Fig. S1). The TC of BST-xBSS decreases from approximately 125 °C (for pure BT) to room temperature as x increases from 0.04 to 0.10. This shift is attributed to a reduced Landau free energy difference between the ferroelectric and paraelectric phases, caused by the incorporation of the modification ions31, which facilitates the phase transition and results in the observed decrease in TC. Concurrently, the phase-transition temperatures of the rhombohedral-orthorhombic (R-O, TR-O) and orthorhombic-tetragonal (O-T, TO-T) phase transition approach the room temperature TC, i.e., this shift transforms the room-temperature phase structure from a tetragonal phase (x = 0.04) to a ferroelectric multiphase coexistence (x = 0.06 and 0.08), and eventually to a TC peak below room temperature (x = 0.10), as corroborated by the phase diagram (Supplementary Fig. S2), Rietveld refinement of X-ray diffraction data (Supplementary Fig. S3 and Table S1), and Raman spectra (Supplementary Fig. S4). Integrating the temperature-dependent Raman spectra presented in Fig. 2b and Supplementary Fig. S5, it is evident that the convergent and successive R-O and O-T ferroelectric phase transitions nearing the TC peak are observable at x = 0.08 near room temperature. The phase convergence contributes to a pronounced phase-transition dielectric peak due to the multiphase critical point effect32. The ferroelectric phase structure remains metastable as it approaches the TC peak. Despite this metastability, distinct ferroelectric phase transitions are observed at x = 0.08, unlike at x = 0.10, where transitions completely merge into the TC peak. This indicates that this metastable ferroelectric structure retains significant ferroelectric polarization at x = 0.08. Furthermore, the incorporation of Sr2+ or Sn4+ into the BT matrix leads to the formation of nanosized domains, enhancing local polarization and dielectric response33,34. Consequently, BST-xBSS exhibits a high dielectric constant and dielectric anomalies at ferroelectric–ferroelectric and ferroelectric–paraelectric phase transitions, where the sharp ferroelectric–paraelectric phase transition mirrors the rapid phase transition observed in pure BT.

Fig. 2: Phase transitions and electrical & electrocaloric properties.
Fig. 2: Phase transitions and electrical & electrocaloric properties.The alternative text for this image may have been generated using AI.
Full size image

a Temperature-dependent dielectric constant of BT and BST-xBSS ceramics. b Temperature-dependent Raman spectra for BST-0.08BSS. c Comparison of room-temperature maximum polarization (Pm), remnant polarization (Pr), coercive field (Ec), dielectric constant (εr), and longitudinal piezoelectric coefficient (d33) between BT and BST-0.08BSS. d Polarization-electric field (P–E) loops, e polarization change rate with temperature (∂P/∂T), f electrocaloric strength (ΔT/ΔE), and g electric field-induced heat flow curves (near room temperature) for BST-0.08BSS. h Comparison of ΔT/ΔE at room temperature between BT and BST-0.08BSS. i Comparison of ΔT/ΔE between this work and other lead-free and some typical lead-based ferroelectric ceramics near room temperature.

The Curie–Weiss temperature (TCW) and the temperature (Tm) at which maximum dielectric constant (εm) occurs can be determined from the 1/εr–T curves (Supplementary Fig. S6a). TCW denotes the temperature where the dielectric constant deviates from the Curie–Weiss law. Thus, the difference TCW–Tm serves as an indicator of the phase transition’s diffuseness35. Moreover, the diffuseness can be quantified using the diffuseness coefficient (γ) derived from the modified Curie–Weiss equation36:

$$1/{\varepsilon }_{{{\rm{r}}}}{{\rm{\hbox{ - }}}}1/{\varepsilon }_{{{\rm{m}}}}={(T-{T}_{{{\rm{m}}}})}^{{{\rm{\gamma }}}}/C (1\le {{\rm{\gamma }}}\le 2)$$
(4)

where C is the Curie constant. The value of γ is obtained from the slope of the ln(1/εr–1/εm) versus ln(T–Tm) plot (Supplementary Fig. S6b). Both γ and TCW–Tm exhibit a gradual increase with rising x (Supplementary Fig. S6d, e), indicating that the phase transition becomes diffuse. This increased diffuseness is attributed to the addition of Sr2+/Sn4+, and the formation of smaller ferroelectric domains following Sr2+/Sn4+ incorporation34,37. Despite the enhanced diffuseness, the dielectric peak remains sharp after Sr2+/Sn4+ incorporation. Based on the normalized εr–T curves (Supplementary Fig. S6c, f–h), two parameters, WR-L and WR-H, are employed to characterize the breadth of the dielectric peak at TC38. The low-temperature side of the TC peak broadens slightly, whereas the high-temperature side narrows significantly after introducing Sr2+/Sn4+, culminating in a narrower phase-transition temperature range (i.e., a sharper phase transition) in BST-xBSS. Consequently, BST-0.08BSS not only manifests a high dielectric peak with a giant dielectric constant but also exhibits a sharp phase transition with an obvious ferroelectric phase existing near room temperature. This swift ferroelectric–paraelectric phase transition, coupled with the giant dielectric constant and significant ferroelectric polarization, greatly benefits the intrinsic EC effect near room temperature.

The room-temperature polarization-electric field (P–E) loops, maximum polarization (Pm), remnant polarization (Pr), coercive field (Ec), εr, and longitudinal piezoelectric coefficient (d33) of the ceramics are displayed in Fig. 2c and Supplementary Fig. S7. Pure BT exhibits P–E loops characterized by high Pm, Pr, and Ec. In contrast, BST-xBSS displays slim P–E loops with reduced Pm and Pr, and markedly lower Ec. For instance, comparing BT and BST-0.08BSS, Pm/Pr decreases from 20.0/12.2 to 15.0/4.8 μC/cm2, while Ec declines sharply from 3.0 to 0.35 kV/cm. The significantly lower Ec suggests easier polarization rotation and domain switching in BST-0.08BSS. Despite the reduction in Pm and Pr, BST-0.08BSS exhibits a higher polarization change (ΔP = Pm – Pr) compared to pure BT, increasing from 7.8 to 10.2 μC/cm2, reflecting improved polarization variation efficiency. The combination of relatively high polarization, easy polarization change, and convergent ferroelectric phase transitions results in enhanced dielectric constant for BST-xBSS32,34. For example, a colossal εr of approximately 12,400 is achieved in BST-0.08BSS at room temperature, more than five folds greater than the εr (~2450) of pure BT. According to the phenomenology relationship of d33 ∝ Pr·εr39, d33 is directly proportional to the product of Pr and εr. Consequently, the colossal εr and high Pr culminate in a giant piezoelectric response, with d33 reaching approximately 800 pC/N in BST-0.08BSS. This high piezoelectric response further corroborates the robust ferroelectricity and easy polarization rotation in BST-0.08BSS.

Intrinsic electrocaloric property

The electrocaloric (EC) property of conventional ferroelectrics can be accessed via indirect and direct methodologies40. The indirect approach relies on Eq. (1) and the polarization change rate with temperature (∂P/∂T). To determine ∂P/∂T, the temperature dependence of P–E loops and polarization under various electric fields for BST-0.08BSS are measured, as depicted in Fig. 2d and Supplementary Fig. S8a. The polarization values are derived from the Pm in P–E loops, and the polarization at 0 kV/cm is inferred from the Pr value. According to polarization versus temperature (P–T) curves, polarization exhibits a decreasing trend with increasing temperature. That is, ∂P/∂T is consistently negative, and its evolution as a function of temperature and electric field is illustrated in Fig. 2e. The maximum absolute value of ∂P/∂T appears near TC, especially under low electric fields. At higher fields, polarization can be induced even above TC, causing the peak of ∂P/∂T to shift toward higher temperatures as the electric field increases. According to Eq. (1) and the obtained results (Fig. 2e and Supplementary Fig. S9), the EC temperature change (ΔT) and EC strength (ΔT/ΔE) can be derived, as illustrated in Fig. 2f and Supplementary Fig. S8b. ΔT/ΔE and ΔT also peak near TC regardless of the applied electric field. Like ∂P/∂T, the ΔT/ΔE and ΔT peaks also shift to higher temperatures with increasing electric field.

To further validate the EC performance, the direct method using differential scanning calorimetry (DSC) heat flow measurements is employed to estimate the EC response of BST-0.08BSS near room temperature. As illustrated in Fig. 2g, exothermic and endothermic heat peaks induced by the EC effect are observed upon applying and removing the electric field, respectively, and the EC performance is calculated from the areas of these heat peaks. The calculated ΔT/ΔE and ΔT values from both indirect and direct methods are consistent. As shown in Fig. 2h, the room-temperature ΔT/ΔE significantly increases from pure BT to BST-0.08BSS, with the highest value reaching approximately 1.0 K·mm/kV. This will lead to a high ΔT under a low electric field, for example, ~0.25–0.4 K can be obtained under 2.5–5 kV/cm (Supplementary Fig. S8b). Based on Eqs. (1)–(3), EC performance is directly related to polarization and dielectric constant. Therefore, the excellent intrinsic EC response in BST-0.08BSS can be attributed to the giant dielectric constant and high polarization, as well as the large polarization variation induced by the easy polarization rotation and rapid phase transition near room temperature. Ultimately, the remarkable EC effect observed in BST-0.08BSS surpasses not only actively studied BT-based ceramics, but also lead-free ferroelectrics such as (K, Na)NbO3 (KNN)- and (Bi0.5Na0.5)TiO3 (BNT)-based ceramics11,13,18,19,20,21,22,23,41,42,43,44,45,46, as well as lead-based PMN-10PT, PST, PLZT, and PMN ceramics2,47,48,49, as shown in Fig. 2i. Thus, this study successfully demonstrates a significant intrinsic EC performance by employing an innovative strategy that combines rapid phase transition with high polarization variation and giant dielectric constant near room temperature by local well-ordered structural engineering.

Discussion

Revealing the property/structure origins of the high intrinsic EC effect

From Eqs. (1)–(3), it is evident that the EC performance is intrinsically linked to ferroelectric polarization and dielectric constant. To further elucidate these relationships and uncover the underlying mechanism of the high intrinsic EC effect in this work, a phenomenological theory is employed. The one-dimensional Gibbs free energy (G) of ferroelectrics under stress-free boundary conditions can be approximated as an expansion in terms of polarization (P)50:

$${G}_{1}={G}_{0}+\frac{1}{2}{\alpha }_{0}(T-{T}_{{{\rm{CW}}}}){P}^{2}+\frac{1}{4}\beta {P}^{4}+\frac{1}{6}\gamma {P}^{6}$$
(5)

where α = α0(T – TCW), β, and γ are expansion coefficients. Upon applying an electric field (E), the equilibrium state of the system is given by:

$$\partial {G}_{1}/\partial P=E=\alpha P+\beta {P}^{3}+\gamma {P}^{5}$$
(6)

According to Eq. (6) and ∂E/∂P = 1/ε, the following equation can be derived:

$$1/\varepsilon={{\rm{d}}}E/{{\rm{d}}}P=\alpha+3\beta {P}^{2}+5\gamma {P}^{4}$$
(7)

Differentiating Eq. (7) with respect to T yields:

$$-\frac{\partial P}{\partial T}=\frac{\left(\alpha+3\beta {P}^{2}+5\gamma {P}^{4}\right)}{\left(6\beta \varepsilon P+20\gamma \varepsilon {P}^{3}\right)}\frac{\partial \varepsilon }{\partial T}+\frac{{\alpha }_{0}}{6\beta P+20\gamma {P}^{3}}$$
(8)

As the temperature approaches TC from the low-temperature side, ferroelectric polarization rapidly decreases (i.e., ∂P/∂T is negative), while dielectric constant significantly increases (i.e., ∂εr/∂T is positive). According to Eq. (8), −∂P/∂T is approximately positively correlated with ∂εr/∂T. Since the EC effect is positively correlated with −∂P/∂T according to Eq. (1), it is also positively influenced by ∂εr/∂T. The temperature-dependent εr of BST-xBSS with x = 0.04, 0.06, and 0.08, all of which have TC above room temperature, is illustrated in Fig. 3a and Supplementary Fig. S10. Among them, BST-0.08BSS shows a pronounced TC dielectric peak and an elevated εr below TC. Consequently, giant εr and sharp εr evolution lead to a rapid variation in ∂εr/∂T around the low-temperature side of TC. Ultimately, BST-0.08BSS exhibits the highest ∂εr/∂T and −∂P/∂T, contributing to a significant intrinsic EC response. As depicted in Fig. 3b, ∂εr/∂T, −∂P/∂T, and ΔT/ΔE exhibit similar trends for BST-0.08BSS, all peaking near TC (close to room temperature). This demonstrates that a sharp phase transition and a giant εr facilitate a large polarization change, thereby enhancing the intrinsic EC effect.

Fig. 3: Analyses of dielectric change rate, in-situ synchrotron X-ray diffraction, domain structure, and elements mapping.
Fig. 3: Analyses of dielectric change rate, in-situ synchrotron X-ray diffraction, domain structure, and elements mapping.The alternative text for this image may have been generated using AI.
Full size image

a Temperature dependence of dielectric change rate with temperature (∂εr/∂T) at the low-temperature side of the Curie temperature for BST-xBSS. b Comparison of ∂εr/∂T, −∂P/∂T, and ΔT/ΔE at the low-temperature side of the Curie temperature for BST-0.08BSS. c In-situ synchrotron X-ray diffraction patterns of (111) and (200) reflections under 0 kV/cm (unpoled state), applying the electric field of 3 times the Ec (poling state), and 0 kV/cm (poled state), and the evolution of patterns intensity for BST-0.08BSS. d Transmission electron microscopy (TEM) images showing multiscale domain structure with domain widths from several nanometers to hundreds of nanometers. e Statistics of the width of representative domains marked by solid white lines in the TEM images. f Microscopic element mappings of O, Ba, Ti, Sn, and Sr elements.

In addition to rapid dielectric and polarization changes, the facile response of polarization to an applied electric field is another crucial factor for achieving a high intrinsic EC effect. Figure 3c illustrates the intensity variation in in-situ synchrotron X-ray diffraction patterns for BST-0.08BSS before and after applying an electric field. A low electric field (3Ec≈1.0 kV/cm) can induce noticeable change in the XRD pattern of BST-0.08BSS, indicating highly responsive polarization behavior, with ferroelectric domains readily rotating and switching51. This ease of polarization rotation and domain switching is likely due to the presence of nanodomains arising from ferroelectric multiphase coexistence30. Unlike BT, which features micro-sized domains, BST-0.08BSS exhibits much smaller domains (Supplementary Fig. S11a). TEM analysis in Fig. 3d, e further reveals the multiscale domain configurations, including randomly oriented strip-type domains and multiscale nanodomains. These range from larger nanodomains with widths of ~120–180 nm (colored TEM image I and line 1) to numerous smaller ones with widths from ~10 nm to several tens of nanometers (colored TEM images II and III, and lines 2 and 3). Of particular interest is that these nanodomains exhibit regular shapes, distinguishing them from irregular polar nanoregions and indicating a ferroelectric structure with strong local polarization. The high density of domain walls in the multiscale configuration contributes to a low domain wall energy, facilitating easy polarization rotation and domain switching. As demonstrated in Supplementary Fig. S11b, BST-0.08BSS requires a significantly lower switching voltage compared to BT, enabling substantial polarization change even under a low electric field and enhancing the dielectric response30,52,53. Therefore, the multiscale nanodomain configuration is the key microscopic structural origin behind the high dielectric constant, large polarization change rate, and exceptional intrinsic EC effect observed in BST-0.08BSS ceramic.

Atomic-resolution structure and local polarization

Figure 3f displays energy-dispersive X-ray spectroscopy (EDS) mappings of BST-0.08BSS, revealing the microscopic elements distribution within the ceramic. The mapping shows that the matrix Ba, Ti, and O elements are uniformly distributed. Particularly, Sr and Sn elements also exhibit a homogeneous distribution, indicating Sr2+ and Sn4+ have been effectively integrated into the ceramic matrix, and homogeneously occupy the A or B sites as expected. This result is well consistent with the previous DFT calculation results. The homogeneous elements distribution supports well-ordered local structure, which preserves robust intrinsic ferroelectric polarization and facilitates rapid, synchronized structural transitions in response to thermal or electric field stimuli31.

Further validation of the structural homogeneity is provided by atomic-resolution HAADF-STEM images and local polarization analysis. Figure 4a, b illustrates the relative intensities of A-site Ba2+/Sr2+ and B-site Ti4+/Sn4+ in BST-0.08BSS, as derived from HAADF-STEM lattice imaging along the [100] zone axis (Supplementary Fig. S12). The color contrast in the ionic columns reflects the relative intensity of each ion site, with differences predominantly attributable to varying atomic numbers and lattice distortions30,54. The intensity color mapping reveals a homogeneous local distribution of ion intensities for both A-site and B-site ions, demonstrating the small lattice distortions in BST-0.08BSS. To quantitatively assess the homogeneity of local ions, intensity statistics for A-site and B-site ions are conducted, as depicted in Fig. 4c, d. The ion intensities fall within a narrow range, with a notably small standard deviation of 3.40 for A-site ions relative to their average intensity (90.7), and a similarly small standard deviation of 3.21 for B-site ions relative to their average intensity (62.4)55. These findings collectively confirm the homogeneous distribution of the modification ions and the minimal lattice distortions, both contribute to the well-ordered local structure.

Fig. 4: Atomic-scale structure and local dipole polarization distribution obtained by scanning transmission electron microscopy along the [100] zone axis for BST-0.08BSS.
Fig. 4: Atomic-scale structure and local dipole polarization distribution obtained by scanning transmission electron microscopy along the [100] zone axis for BST-0.08BSS.The alternative text for this image may have been generated using AI.
Full size image

a Relative intensity of the A-site ions. b Relative intensity of the B-site ions. c Statistics of intensity distribution for the A-site ions. d Statistics of intensity distribution for the B-site ions. e Calculated dipole vectors. f Distribution of dipole vectors for the amplified region corresponding to the green frame zone (The color difference of dipoles indicates the relative dipole intensity or angle). g Statistic mapping of relative dipole intensity (based on the Ti-O displacement) corresponding to the green frame zone in Supplementary Fig. S13. h Statistic distribution of relative dipole intensity corresponding to Supplementary Fig. S13. i Statistic mapping of relative dipole angle corresponding to the region shown in Fig. 4f. j Statistic distribution of relative dipole angle corresponding to the Supplementary Fig. S15.

The distribution of local polarization, influenced by the homogeneous distribution of modification ions in BST-0.08BSS, is illustrated in Fig. 4e. The distribution maps of the dipolar vectors, including relative dipole intensity and angles, have also been calculated and presented in Fig. 4f and Supplementary Figs. S13–15. To further analyze the polarization distribution, quantitative mappings of the relative intensity and angle are provided in Fig. 4g, i, respectively. Figure 4h showcases the statistical distribution of relative dipole intensity. The variation of the dipole intensity is likely due to the chemical modifications. However, the polarization fluctuation in BST-0.08BSS is relatively small and randomly distributed at the local scale, which promotes a sharp macroscopic phase transition. This behavior likely stems from the well-ordered local structure induced by the minimal lattice energy differences in the material. Figure 4j displays the statistical distribution of dipole angle relative to the [001] zone axis. Unlike typical ferroelectrics (e.g., BT), which exhibit a regular long-range ferroelectric dipole distribution with micrometer-sized domains54, BST-0.08BSS displays a more randomized dipole distribution with gradual dipole rotational variation, attributed to its multiphase coexistence near room temperature. This structural complexity contributes to the formation of multiscale nanodomains, as observed in Fig. 3d, e. The dipole angle distribution is predominantly concentrated between the [001] and [010] directions, both characteristic of the T phase. However, contributions from the O and R phases—with polarization orientation along {011} and {111} directions, respectively—are also evident. This suggests that the net polarization primarily arises from the spontaneous polarization of the T phase, with additional contributions from the O phase and R phase13. This finding aligns with the previously established phase structure analysis of three-phase coexistence. This type of local polarization distribution can induce a remarkably low energy barrier for polarization rotation and domain switching, significantly accelerating dynamic polarization change under an external electric field30,56. Consequently, this leads to a high polarization variation even with a small electric field stimulus, which in turn results in the observed high dielectric constant and large polarization, together with a sharp phase transition, accounting for the giant intrinsic EC effect in BST-0.08BSS.

In summary, an exceptional intrinsic electrocaloric effect of ΔT/ΔE~1.0 K·mm/kV is achieved in a carefully engineered BST-0.08BSS ferroelectric ceramic through chemical modification that promotes a well-ordered local structure. This strategy decreases the TC to room temperature while preserving a sharp phase transition, enabling a giant dielectric constant, strong polarization, and rapid polarization/dielectric change rate. The gradual dipole rotation, along with the formation of multiscale nanodomains ranging from ~10 nm to >100 nm, further enhances polarization change by lowering the energy barrier for polarization rotation and domain switching, ultimately inducing a high intrinsic EC response. This work presents an effective microstructural strategy for enhancing room-temperature intrinsic EC effects in ferroelectrics, providing valuable insights into the microstructural mechanism underlying the high intrinsic EC effects, and offering guidance for future material design.

Methods

Sample preparation

(1-x)Ba0.87Sr0.13TiO3-xBa0.87Sr0.13SnO3 (BST-xBSS) (x = 0.04, 0.06, 0.08, and 0.10) ceramics were prepared through the conventional solid-state method. BaCO3 (99%), SrCO3 (99%), SnO2 (99%), and TiO2 (98%) were weighed stoichiometrically and then ball-mixed in ethanol for 24 h. The dried powders were calcined at 1280 °C (3 h), then pressed into pellets using 8% polyvinyl alcohol (PVA) solution as a binder. After burning out PVA, the pellets were sintered at 1450 °C (3 h). The sintered pellets were coated with a silver electrode for electrical properties measurements.

Structure characterizations

X-ray diffraction (XRD, Empyrean, Panalytical) is used to identify phase structure. Raman spectra are conducted by a Horiba Aramis Raman spectrometer (Horiba Scientific) with excitation sources of 532 nm. In-situ synchrotron XRD under different applied electric fields was performed using Source BL02U2 (photon energy 18 keV) beamline in Shanghai Synchrotron Radiation Facility. Domain structure analysis and local domain switching experiments were performed by a piezoresponse force microscopy (PFM, MFP-3D, Asylum Research, Goleta, CA). Transmission electron microscopy (TEM), high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM), and energy-dispersive X-ray spectroscopy (EDS) studies were conducted using an instrument (FEI Titan G2 60-300, USA) to analyze domain structure, element mapping, and atomic-scale structure.

Properties measurements

Temperature-dependent dielectric constant (εr–T) curves were measured using an LCR meter (4980 A, Keysight, USA). The longitudinal piezoelectric coefficient (d33) was measured using a commercial Berlincourt-type d33 meter (ZJ-3A, China) for the poled ceramics. Temperature-dependent ferroelectric hysteresis (P–E) loops were investigated by the ferroelectric tester (aixACCT, TF Analyzer 2000E, Germany). The direct electrocaloric effect was measured using a modified DSC (Q2000, TA Instruments, New Castle, DE), and the DC electric field was applied utilizing a dielectric strength tester (ET2671A, ENTAI, Nanjin, China).

Density functional theory (DFT) calculation

The structures of Sr, Sn, Bi-Na, and Ga-Nb doped BaTiO3 were studied by first-principles calculations based on density functional theory (DFT) using the Vienna Ab initio Simulation Package (VASP)57,58. The generalized gradient approximation (GGA) with the PBEsol59 exchange-correlation function was employed in the calculations. The wave function is represented as a plane wave expansion that is truncated at a cut-off energy of 500 eV. For lattice energy difference calculation, Γ-centered k-point meshes with a grid of spacing 0.04 × 2π Å−1 for Brillouin zone sampling were chosen, and a 2 × 2 × 8 perovskite supercell was used to conduct calculations. For lattice distortion calculation, Γ-centered k-point meshes with a grid of spacing 0.122 × 2π Å−1 for Brillouin zone sampling were chosen, and a 1 × 5 × 5 perovskite supercell was used to conduct calculations. The initial pure BaTiO3 structure is set as tetragonal symmetry with parallel and perpendicular polarization, respectively. After doping, the ionic positions, cell volume, and cell shape are allowed to change, and all of the geometry optimizations were completed when the residual force of each atom was less than 0.001 eV/Å.