WorksPhysics and Astronomy
Atomistic origin and strain control of the finite-temperature dielectric response in BaTiO3
Abstract
The dielectric response of BaTiO3 (BTO) varies strongly with temperature, yet its local atomic origin remains unclear. Using electric-field-coupled molecular dynamics with a machine-learning force field, we show that the permittivity tracks the field-induced angular redistribution of local Ti-O off-centering rather than its magnitude or mean polar angle. Temperature and biaxial strain modify the local structure differently, but both responses follow a common relation with the same orientational descriptor, providing a real-space counterpart to soft-mode behavior.
Introduction
The dielectric response of ferroelectric materials varies strongly with temperature and often becomes exceptionally large near ferroelectric phase transitions [1, 2, 3]. This temperature dependence is central to ferroelectric physics and directly affects devices such as capacitors and piezoelectrics [4, 5]. A long-standing question is which local atomic degree of freedom carries this response in real space.
(BTO), whose ferroelectricity was discovered in the 1940s, has served as a prototypical ferroelectric for more than eight decades [6]. Upon heating, BTO undergoes successive rhombohedral (R), orthorhombic (O), tetragonal (T), and cubic (C) phase transitions, accompanied by pronounced changes in its dielectric response [1, 2, ?, ?]ref-7,ref-8. Within the framework represented by the Lyddane–Sachs–Teller relation, enhancement of the low-frequency permittivity is associated with a decrease in the frequency of a transverse optical soft mode {cite:p}ref-9. At the same time, an order–disorder picture has also been widely discussed for BTO, in which local Ti off-centering persists at finite temperature and phase transitions involve changes in the orientational ordering of these local polar distortions {cite:p}ref-10,ref-11,ref-12. BTO is therefore generally regarded as exhibiting characteristics of both displacive and order–disorder behavior {cite:p}ref-13,ref-14`.
Both pictures are collective. A phonon frequency or an averaged structure describes the system as a whole but does not directly specify how an individual local polar unit responds to an applied field. Importantly, the mean polar angle of a local distortion and the field-induced redistribution of its angular population contain distinct information: the mean polar angle can remain nearly unchanged even when the field response changes strongly. Previous atomistic simulations have reproduced the temperature-dependent permittivity of
BTO from equilibrium polarization fluctuations [?], but such approaches do not directly resolve the field-induced response of individual local polar units. Recent machine-learning force fields (MLFFs) enable larger and longer simulations [?, ?, ?, ?], while MACEField [?] predicts instantaneous Born effective charges and thereby enables electric-field-coupled molecular dynamics of dielectric response [?, ?, ?, ?, ?].
In this work, we use MACEField to identify the local real-space response underlying the dielectric behavior of tetragonal BTO. Across the tetragonal phase, the permittivity varies strongly with temperature, whereas neither the magnitude nor the mean polar angle of the local off-centering relative to the [001] polarization direction exhibits a corresponding systematic change. Instead, the field-induced redistribution of its angular population decreases markedly with increasing temperature. Biaxial in-plane strain modifies the dielectric response through a distinct microscopic route, changing both the mean polar angle and the field-induced angular response. Nevertheless, the temperature- and strain-dependent dielectric responses follow a common approximately linear relationship with the orientational response . These results connect the macroscopic permittivity to the field-induced angular response of local polar distortions and provide a real-space counterpart to the conventional soft-mode description.
Methods
Reference dataset and r²SCAN-based MACEField model
An r²SCAN-based MACEField model was used for the finite-temperature and electric-field-induced simulations. Its selection from LDA [?], PBEsol-1 [?], and r²SCAN-based [?] models based on phase-transition behavior is detailed in Figure S1 and Table SI.

Figure S1. Exchange-correlation-functional and supercell-size dependence of the finite-temperature phase transitions of BaTiO₃. Calculated R–O, O–T, and T–C transition temperatures obtained using the (a) LDA-, (b) PBEsol-, and (c) r²SCAN-based MACEField models as a function of supercell size. Solid lines represent the calculated transition temperatures, and horizontal dashed lines indicate the corresponding experimental values reported in Ref. [3]. The R–O, O–T, and T–C transitions are shown in brown, green, and orange, respectively. The LDA- and PBEsol-based models systematically underestimate the transition temperatures, whereas the r²SCAN-based model provides substantially improved agreement with experiment.
| Model | LDA | PBEsol | rSCAN |
| (K) | |||
| (K) | |||
| (K) | |||
| MAE (K) | 173.1 | 87.9 | 41.1 |
Table SI. Comparison of the finite-temperature phase-transition temperatures predicted by the LDA-, PBEsol-, and rSCAN-based MACEField models. Mean R–O, O–T, and T–C transition temperatures and their standard deviations are evaluated over the examined supercell sizes. The overall mean absolute error (MAE) is calculated relative to the corresponding experimental transition temperatures and is used as the primary criterion for model selection.
The r²SCAN reference dataset was constructed from density-functional-theory calculations performed using VASP [?, ?, ?, ?]. Forty-atom supercells representing the rhombohedral, orthorhombic, tetragonal, and cubic phases were employed, and finite-temperature configurations were sampled using the on-the-fly machine-learning framework implemented in VASP [?]. Reference energies and atomic forces were evaluated using the rSCAN meta-GGA functional.
Because density-functional perturbation theory for BECs is not currently available for meta-GGA functionals in VASP, the BEC tensors used to train the rSCAN-based MACEField model were evaluated using PBEsol-DFPT [?] for structures generated at the rSCAN level. The validity of this treatment was independently assessed using rSCAN finite-field calculations based on the perturbation-expansion-after-discretization (PEAD) method [?]. The comparison is presented in Figure S2.

Figure S2. Validation of the Born effective charges used for the rSCAN-based MACEField model. Comparison of the diagonal BEC tensor components calculated using PBEsol density-functional perturbation theory (DFPT) and those independently obtained using the finite-electric-field perturbation-expansion-after-discretization (PEAD) method with the rSCAN functional. The dashed line indicates . The mean absolute error between the two methods is 0.100.
The rSCAN-based MACEField model was fine-tuned from MACE-MP-0 large [?]. Prediction errors are summarized in Table SII, and phonon validation is shown in Figure S3.

Figure S3. Phonon dispersions of BaTiO3 in the (a) rhombohedral, (b) orthorhombic, (c) tetragonal, and (d) cubic phases. First-principles VASP results are shown by red solid lines, and predictions from the r2SCAN-based MACEField model are shown by blue dashed lines. The model reproduces the first-principles phonon dispersions across all four phases, with phonon-frequency mean absolute errors ranging from 0.13 to 0.22 THz.
Table SII. Prediction errors of the MACEField models constructed using different exchange-correlation functionals. The MAEs in energy, atomic forces, and BECs are evaluated using the test dataset. The MAEs in phonon frequency are additionally evaluated for the R, O, T, and C phases of BaTiO3.
Finite-temperature molecular dynamics and local Ti–O structure
Finite-temperature heating simulations were initiated from the rhombohedral phase after full relaxation of the atomic positions and lattice parameters. Variable-cell molecular dynamics was performed under zero external stress. The thermostat and barostat time constants were set to 25 and 75 fs, respectively, and a time step of 1 fs was used. The system was heated linearly from 1 to 500 K over 200 ps, and the atomic coordinates and lattice parameters were recorded every 1 ps.
To characterize the local polar structure, the Ti–O off-centering vector for each Ti atom was defined relative to the center of the cage formed by its six nearest-neighbor O atoms,
This quantity represents a relative Ti–O displacement and therefore naturally incorporates the motion of both the Ti and O sublattices. Its polar angle was defined with respect to the primary polarization direction, [001], as .
Phase-transition temperatures were identified using both the reduced lattice parameters and changes in the mean orientation of the Ti–O off-centering. A Gaussian filter was applied to the temperature-dependent orientation angles, and their temperature derivatives were evaluated to identify candidate transitions. Final transition temperatures were assigned only when consistent anomalies were observed in both the lattice parameters and multiple orientational components. The procedure is illustrated in Figure S4.

Figure S4. Identification of the phase-transition temperatures of BaTiO during continuous-heating molecular dynamics for an supercell described by the rSCAN-based MACEField model. (a) Temperature dependence of the reduced lattice parameters (blue), (orange), and (green). (b) Temperature derivatives of the Gaussian-filtered directional angles of the Ti–O off-centering vector relative to the , , and crystallographic planes. The blue, orange, and green curves represent , , and , respectively. Black vertical dotted lines indicate the R–O, O–T, and T–C transition temperatures determined from anomalies consistently observed in both the lattice parameters and the orientational derivatives. For this supercell, the transition temperatures are 162.5, 240, and 335 K, respectively.
Contact author: [email protected]
Electric-field-induced molecular dynamics and polarization
Electric-field-induced molecular dynamics was performed using the instantaneous BEC tensors predicted by MACEField. The external force acting on atom was evaluated as
where is the applied electric-field component and is the instantaneous BEC tensor.
The macroscopic polarization was evaluated relative to a nonpolar cubic reference structure using
where is the instantaneous cell volume and is the displacement of atom from its corresponding reference position.
The ferroelectric response of the tetragonal phase was independently validated at 270 K. The polarization–electric-field hysteresis loop and the corresponding local Ti–O switching dynamics are presented in Figure S5. Polarization switching in BTO has been extensively investigated in both experimental and computational studies [?, ?, ?, ?].

Figure S5. Ferroelectric polarization switching in tetragonal BaTiO at 270 K simulated using the rSCAN-based MACEField model. (a) Polarization–electric-field hysteresis loops obtained from the polarization components along the , , and crystallographic directions. , , and are shown in green, brown, and orange, respectively. (b) Evolution of the populations of the six crystallographic orientations of the local Ti–O off-centering vectors during polarization switching. The six crystallographic orientations , , , , , and are color-coded as indicated in the legend. Before complete reversal of the primary polarization, the local Ti–O polar distortions transiently reorient toward an approximately 90° direction, accompanied by an increase in the transverse polarization component in (a).
The relative dielectric constant along crystallographic direction was obtained from the low-field slope of the polarization response,
Hereafter, denotes the relative dielectric constant along the crystallographic direction unless otherwise specified.
Element-resolved and local orientational analyses
To decompose the dielectric response by atomic species, the summation in (3) was restricted to atoms belonging to species , yielding the species-resolved polarization . The corresponding polarization-response contribution was defined as
Because the vacuum contribution of unity in (4) cannot be assigned to an individual element, only the polarization-response term was decomposed. The fractional contribution from species was therefore evaluated as
To quantify the electric-field-induced reorientation of the local Ti–O polar structure, the orientational distribution was represented in spherical coordinates with the [001] direction as the polar axis. The polar angle measures the inclination of the Ti–O off-centering vector from [001], whereas the azimuthal angle describes rotation about [001]. In the pole figures, lies along the horizontal direction; therefore, a radial displacement corresponds to a change in , whereas an azimuthal displacement corresponds to a change in at fixed .
In these coordinates, the signed mean orientation along [100] is
where the average is evaluated at a given electric field . The field-induced orientational response was then defined as
Because depends on both and , a field-induced redistribution in the azimuthal coordinate can change even when the mean polar angle remains nearly unchanged. The same analysis was applied to both the temperature-dependent and in-plane-strain-dependent datasets.
For the strain-dependent calculations, the in-plane lattice axes were constrained to prescribed biaxial strains while the dielectric and local structural responses were evaluated at fixed temperature. Additional temperature-dependent simulations with the same in-plane lattice constraint were performed to verify that neither the characteristic temperature dependence nor the relationship between and is an artifact of the lattice constraint. These calculations are presented in Figure S6.

Figure S6. Temperature-dependent dielectric and orientational responses under constrained in-plane lattice conditions. (a) Temperature dependence of the relative dielectric constant under the constrained in-plane lattice condition used for the strain-dependent calculations. The characteristic decrease in dielectric response with increasing temperature within the tetragonal phase is retained under the lattice constraint. (b) Relationship between and the field-induced orientational response for the constrained-temperature dataset. For comparison, the unconstrained temperature-dependent and in-plane-strain-dependent datasets presented in the main text are also shown. The dashed line represents a linear regression to the constrained-temperature dataset, with .
Results and Discussion
Finite-temperature dielectric response and element-resolved contributions
The rSCAN-based MACEField reproduces the phase-transition behavior, BECs, phonon dispersions, and ferroelectric switching of BTO; validation and model selection are summarized in Figure S1–S5 and Tables SI–II. We therefore focus on the microscopic origin of the temperature-dependent dielectric response in the tetragonal phase.
Because the absolute phase-transition temperatures differ between calculation and experiment, the temperature within the tetragonal stability region was normalized according to
such that corresponds to the O–T transition and corresponds to the T–C transition, i.e., the Curie temperature.
Figure 1 (a) shows the relative dielectric constant along the crystallographic direction as a function of normalized temperature. Hereafter, denotes the relative dielectric constant along the direction. The calculated dielectric response closely reproduces the experimentally observed temperature dependence: decreases markedly with increasing temperature over the tetragonal region, while a sharp enhancement emerges near the T–C transition, corresponding to the Curie temperature [1, 2, 3].

Figure 1. Finite-temperature dielectric response of tetragonal BaTiO and its element-resolved contributions. (a) Relative dielectric constant as a function of the normalized tetragonal-phase temperature . The calculated results are shown by blue circles, while experimental data from Refs. [1, 2, 3] are shown by gray circles, squares, and triangles, respectively. and 1 correspond to the O–T and T–C phase transitions, respectively. (b) Element-resolved contributions of Ba, Ti, and O to the polarization-derived dielectric response. O dominates the response on the lower-temperature side of the tetragonal phase, whereas the relative contribution of Ti increases toward the Curie temperature. The contribution of Ba remains small over the temperature range examined.
This agreement allows us to use the trajectories to address the corresponding atomistic question. In the lattice-dynamical picture, the large low-frequency permittivity of BTO is associated with a soft transverse optical mode. While its frequency and
eigenvector characterize the collective response, they do not by themselves resolve how that response is distributed among local atomic motions at finite temperature. We therefore test this directly by decomposing the polarization-derived response into contributions from Ba, Ti, and O. As shown in Figure 1(b), although Ti off-centering is commonly discussed as a characteristic local polar distortion in BTO [?, ?], the O sublattice provides the dominant contribution to the dielectric response on the lower-temperature side of the tetragonal phase. The relative contribution of Ti remains smaller at lower temperatures but increases markedly as the temperature approaches the Curie temperature. In contrast, the contribution from Ba remains small throughout the temperature range examined. The
*Contact author: [email protected] temperature-dependent dielectric response is therefore governed predominantly by the Ti–O sublattice. The element-resolved analysis above provides a direct link to such a local description: the dielectric response is dominated by the relative motion of Ti and O rather than by Ti alone. This observation motivates the use of Ti–O off-centering as the relevant local structural coordinate in the following analysis. Because the off-centering is defined as the displacement of Ti relative to the center of the surrounding O cage rather than as the absolute displacement of Ti, it naturally incorporates the motion of both Ti and O—the two sublattices that dominate the dielectric response. At finite temperature, this local polar distortion can be characterized by distinct aspects, including its magnitude, its mean polar angle relative to the [001] polarization direction, and the field-induced redistribution of its angular population. We therefore next examine which of these local quantities changes together with the macroscopic permittivity.
Temperature dependence of the local Ti–O response
Figure 2 (a) and Figure 2 (b) show the mean polar angle with respect to [001] and the magnitude of Ti–O off-centering evaluated over structures sampled throughout the hysteresis loop. Despite the pronounced decrease in dielectric constant across the tetragonal phase, neither quantity shows a corresponding systematic temperature dependence. Thus, temperature does not substantially alter either the radial amplitude of the local polar distortion or its mean inclination from the primary polarization direction.

Figure 2. Temperature dependence of the local Ti–O polar structure and its electric-field-induced orientational response in tetragonal BaTiO3. (a) Temperature dependence of the mean polar angle of the Ti–O off-centering with respect to the [001] direction. (b) Temperature dependence of the Ti–O off-centering magnitude. In (a) and (b), symbols indicate mean values and shaded regions represent the corresponding standard deviations. (c) and (d) Pole-figure representations of the Ti–O off-centering projected along [001], where the radial coordinate is the polar angle from [001] and the azimuthal coordinate is the angle about [001]. The applied field [100] lies along the horizontal direction. (c) Schematic illustration of the differential orientational analysis. The differential distribution is defined as , where and denote the orientational probability distributions under opposite electric-field polarities. (d) Differential pole figures of the Ti–O off-centering viewed along the [001] direction at and 0.84. Red and blue indicate positive and negative values of , respectively.
The absence of a systematic change in the mean polar angle does not preclude field-induced redistribution of the full angular distribution. Figure 2 (d) shows the differential distributions at and , defined schematically in Figure 2 (c). At the lower temperature, the two lobes lie on opposite sides of the center along the field at similar radial positions, indicating predominantly azimuthal redistribution about [001] rather than a change in inclination. This redistribution becomes markedly weaker at the higher temperature.
The orientational response decreases from 0.0676 at 260 K to 0.0178 at 320 K, while the mean polar angle changes by less than 5° and the off-centering magnitude by less than 1%. Thus, heating leaves the local polar distortion largely intact but markedly weakens its field-induced angular response.
Strain modulation of the dielectric and local Ti–O responses
If is the quantity that carries the dielectric response, it must also track that response when the response is changed by something other than temperature. We therefore modulated the permittivity with biaxial in-plane strain at a fixed temperature of 290 K.
Figure 3 shows the relative dielectric constant as a function of in-plane strain. Under −0.3% compressive strain, is 386. It increases systematically as the strain becomes tensile and reaches 2559 at +0.3%. Thus, at a fixed temperature, the dielectric response can be varied over a wide range solely by modifying the in-plane lattice strain.

Figure 3. Strain modulation of the dielectric response in tetragonal BaTiO₃. Relative dielectric constant as a function of biaxial in-plane strain at 290 K. The dielectric response increases systematically from compressive to tensile strain, rising from 386 at −0.3% strain to 2559 at +0.3%.
To determine the microscopic origin of this strain-induced change in permittivity, we next examined the local Ti–O polar structure. Figure 4 (a) and Figure 4 (b) show the strain dependence of the mean polar angle with respect to [001] and the Ti–O off-centering magnitude, respectively. In contrast to the temperature-dependent case, the mean polar angle changes clearly with in-plane strain, particularly toward tensile strain, whereas the off-centering magnitude varies much less strongly. These results indicate that biaxial strain primarily modifies the mean inclination of the local Ti–O polar distortions from the [001] direction rather than strongly changing their radial amplitude.

Figure 4. Strain dependence of the local Ti–O polar structure and common relationship between the dielectric and orientational responses. (a) Strain dependence of the mean polar angle of the Ti–O off-centering with respect to [001]. (b) Strain dependence of the Ti–O off-centering magnitude. The quantities in (a) and (b) were evaluated from the local structures sampled throughout the entire hysteresis loop. Symbols indicate mean values and shaded regions represent the corresponding standard deviations. (c) Differential pole figures of the Ti–O off-centering viewed along the [001] direction at biaxial in-plane strains of −0.3% and +0.3%. The difference between the orientational probability distributions under and is shown. Red and blue indicate an increase and decrease, respectively, in occupation probability under relative to . (d) Relative dielectric constant plotted against the orientational response for the temperature-dependent and in-plane-strain-dependent datasets. Temperature- and strain-dependent data are shown by red and blue symbols, respectively. The dashed line represents a linear regression to the combined dataset, , with .
The differential pole figures in Figure 4 (c) further show how strain modifies the field-induced angular redistribution of the Ti–O off-centering. Under −0.3% compressive strain, the difference between the orientational distributions under + and − is relatively small. Under +0.3% tensile strain, a much more pronounced field-induced redistribution appears. Consistently, increases from 0.0091 at −0.3% strain to 0.0588 at +0.3%.
Temperature and strain therefore modify the local polar response through different microscopic routes: temperature strongly changes the field-induced angular redistribution while leaving the mean polar angle nearly unchanged, whereas strain modifies both the mean polar angle and the field-induced angular response. In both cases, the macroscopic dielectric response varies with .
Common angular descriptor and relation to the soft-mode description
*Contact author: [email protected]
The preceding results show that temperature and in-plane strain affect the local Ti–O polar structure in distinct ways. Temperature primarily changes the field-induced angular redistribution while leaving the mean polar angle relative to [001] nearly unchanged, whereas strain changes both. We therefore ask whether the dielectric responses arising from these distinct microscopic routes can nevertheless be described by a common local-response quantity.
Figure 4 (d) directly compares the dielectric response with for both the temperature-dependent and strain-dependent datasets. Despite originating from different external perturbations and distinct changes in the overall local structural distributions, the two datasets fall approximately on a common relationship. A linear regression to the combined dataset gives with . The mean polar angle does not collapse the two datasets: it changes only weakly across the temperature series despite the large variation in , whereas it increases with tensile strain together with . The off-centering magnitude is nearly constant throughout and cannot account for either trend. In contrast, tracks both datasets.
Additional temperature-dependent simulations performed with the in-plane lattice axes constrained in the same manner as in the strain calculations retain both the characteristic temperature dependence of the dielectric response and a similar relationship (Figure S6). The observed relationship is therefore not an artifact of the different lattice constraints used in the temperature- and strain-dependent datasets.
The collapse in Figure 4 (d) indicates that neither the radial amplitude nor the mean polar angle of the Ti–O off-centering alone tracks the macroscopic dielectric response. Instead, the common quantity is its field-induced angular redistribution, quantified by .
The finite-temperature dielectric response of BTO has been described for more than six decades within lattice-dynamical theory, in which the low-frequency permittivity is connected to the frequency of a transverse optical mode through the Lyddane–Sachs–Teller relation [?]. This framework has been quantitatively successful in describing the collective dielectric response of BTO [?].
A mode frequency, however, is a spectral description of collective dynamics and does not by itself identify how the corresponding softening is manifested in the real-space response of individual local polar units at finite temperature. This distinction is particularly relevant in tetragonal BTO, where the low-frequency soft mode associated with the -axis dielectric response is strongly overdamped [?].
Our results supply what the mode frequency leaves unspecified: the softening degree of freedom is angular. The local Ti–O distortion neither shrinks nor tilts as the permittivity falls; it stops turning. We regard this angular compliance not as an alternative to the soft-mode picture but as its real-space content. It also makes the order-disorder character of BTO concrete. The local dipoles survive, as that picture requires, and what changes with temperature is how freely a field can rotate them, a field-response quantity, not a static one.
Because temperature and strain reach the same – line from different structural starting points, angular compliance is a property that can be engineered as well as observed. Any perturbation that makes the local polar units easier to rotate about the polarization axis should raise the permittivity, whatever it does to their amplitude or mean tilt.
Conclusions
We connected the -axis dielectric response of tetragonal BaTiO to the field-induced response of its local Ti–O polar structure. Although temperature and biaxial strain modify the local structure through distinct microscopic routes, both dielectric responses follow a common relationship with the orientational response . Neither the amplitude of the local polar distortion nor its mean tilt from [001] tracks the permittivity. Its angular compliance does. This changes what one looks for. A large dielectric response does not require large or soft local dipoles, but dipoles that are easy to turn about the polarization axis. Amplitude and orientability are separate properties of the same unit, and strain tunes them independently here.
needs only trajectories under reversed field, and can be computed for any polar material. Applied across compositions and strain states, it would test whether high permittivity is better sought as rotational softness than as lattice instability.
Acknowledgments
This study was supported by the Ministry of Education, Culture, Sports, Science and Technology (MEXT, 26K22608, 26K01205, 25K24645), and New Energy and Industrial Technology Development Organization (NEDO). P.-Y.C. would like to acknowledge the support of JST SPRING (Grant Number JPMJSP2108). The computation was carried out using the computer resource offered by Research Institute for Information Technology, Kyushu University.
R.S. and P.-Y.C. conceived the idea and designed the research. R.S. performed the simulations, analyzed the data, and prepared the figures. R.S. wrote the original draft of the manuscript. T.M. supervised the project
*Contact author: [email protected] and acquired funding. All authors discussed the results and reviewed and edited the manuscript.
The authors declare no competing financial interest.
Data Availability
The trained machine-learning force field models, Born effective charge models, and simulation trajectories generated in this study are available from the corresponding author upon reasonable request.
Contact author: [email protected]
Supplemental Material for: Atomistic origin and strain control of the finite-temperature dielectric response in BaTiO
Ryotaro Sahashi; Po-Yen Chen; Teruyasu Mizoguchi
This Supplemental Material provides details of the selection and validation of the rSCAN-based MACEField model, including the exchange-correlation-functional dependence of the finite-temperature phase-transition behavior, validation of Born effective charges and phonon dispersions, the procedure used to identify the phase-transition temperatures, validation of ferroelectric polarization switching, and additional dielectric-response calculations under constrained in-plane lattice conditions.
Selection of the rSCAN-Based MACEField Model
Three exchange-correlation-functional-dependent models were considered to select the model used for the subsequent finite-temperature and electric-field-coupled molecular dynamics simulations. For the LDA-based model, the reference dataset reported in [1] was employed and the MACEField model was constructed in the present study. For PBEsol, the previously reported trained model of [2] was used directly without additional fine-tuning. In contrast, the rSCAN reference dataset and the corresponding rSCAN-based MACEField model were constructed in the present study.
Because the central purpose of the present work is to examine finite-temperature dielectric properties, the models were compared primarily in terms of their ability to reproduce the sequence and temperatures of the R–O, O–T, and T–C phase transitions of BTO. Figure S1 shows the calculated transition temperatures as a function of supercell size for the LDA-, PBEsol-, and rSCAN-based models. The corresponding experimental values reported in [3] are included for comparison.
Both the LDA- and PBEsol-based models systematically underestimate the experimental transition temperatures. In contrast, the r²SCAN-based model yields substantially improved transition temperatures and reproduces the experimentally observed R–O–T–C sequence over the supercell sizes examined. The transition temperatures averaged over the examined supercell sizes and their deviations from experiment are summarized in Table SⅠ. The r²SCAN-based model gives the smallest overall MAE, 41.1 K, and was therefore selected for the electric-field-coupled simulations and local-response analyses presented in the main text. To our knowledge, simultaneous quantitative reproduction of all three R–O, O–T, and T–C transition temperatures together with the temperature-dependent dielectric response has not been reported in previous atomistic MD studies of BTO [4].
Validation of the rSCAN-Based MACEField Model
The rSCAN-based model selected from the phase-transition analysis was further validated against first-principles reference data for energies, atomic forces, BECs, and lattice dynamics.
Because DFPT calculations of BECs are not currently implemented for meta-GGA functionals in VASP, the BEC tensors used in the rSCAN-based training dataset were calculated using PBEsol-DFPT for structures generated at the rSCAN level. To evaluate the validity of this treatment, BECs obtained using PBEsol-DFPT were compared with values independently calculated using the finite-electric-field PEAD method with the rSCAN functional.
As shown in Figure S2, the diagonal BEC components obtained using the two approaches exhibit close agreement. The MAE between the PBEsol-DFPT and rSCAN-PEAD diagonal components is 0.100, supporting the use of PBEsol-DFPT BECs as training targets for structures generated at the rSCAN level.
The quantitative prediction errors of the resulting rSCAN-based MACEField model are summarized in tbl:S2. The MAEs in energy, atomic forces, and BECs are 0.17 meV atom, 5.00 meV Å, and 0.0039, respectively. These values demonstrate that the model accurately reproduces both the interatomic energetics and the local polarization response represented in the reference dataset.
The ability of the model to describe lattice dynamics was further examined by comparing phonon dispersions predicted by MACEField with those obtained from first-principles calculations. Figure S3 shows the phonon dispersions of the R, O, T, and C phases. The rSCAN-based MACEField reproduces the first-principles dispersions over the full Brillouin zone, with phonon-frequency MAEs ranging from 0.13 to 0.22 THz depending on the crystal phase. This agreement confirms that the model reproduces the local curvature of the potential-energy surface around
Identification of the Phase-Transition Temperatures
The phase-transition temperatures were determined from both the lattice parameters and directional changes of the local Ti–O off-centering during continuous-heating molecular dynamics simulations.
These directional angles, used only for phase-transition identification, are distinct from the polar angle relative to [001] analyzed in Figure 2 (a) of the main text.
Figure S4 shows a representative analysis for an supercell described by the r2SCAN-based MACEField model. The reduced lattice parameters exhibit characteristic splitting and degeneracy associated with the R–O, O–T, and T–C phase transitions. Simultaneously, the mean directional angles of the Ti–O off-centering change across the corresponding temperature ranges.
To identify the transitions systematically, a Gaussian filter was applied to the temperature dependence of these directional angles, followed by evaluation of their temperature derivatives. Peaks in the derivatives were treated as transition candidates. Final transition temperatures were assigned only when anomalies were consistently observed in both the reduced lattice parameters and multiple orientational components, rather than from a single derivative extremum.
The black vertical dotted lines in Figure S4 indicate the transition temperatures identified using this procedure. For the supercell described by the r2SCAN-based model, the R–O, O–T, and T–C transition temperatures are 162.5, 240, and 335 K, respectively. These values represent apparent transition temperatures under the continuous-heating
*Contact author: [email protected] protocol employed here. The corresponding functional and supercell-size dependence is summarized in Figure S1 and tbl:SI.
Ferroelectric Hysteresis and Polarization-Switching Dynamics
The response of the rSCAN-based model to a large external electric field was further examined by calculating the polarization–electric-field hysteresis of tetragonal BTO at 270 K.
Figure S5 (a) shows the polarization components along the three crystallographic directions during the electric-field cycle. A characteristic ferroelectric hysteresis loop is obtained. The calculated remanent polarization is approximately 26 C cm, in good agreement with the experimentally reported value of approximately 21 C cm for BTO single crystals [5]. The calculated coercive field was 99 kV cm, substantially larger than the experimental value of 3.5 kV cm [5]. This discrepancy might be attributable to the ideal periodic crystal considered in the present simulations, which does not contain pre-existing defects, surfaces, or other local inhomogeneities that can facilitate polarization-reversal nucleation in real crystals. Recent atomistic simulations of BTO have likewise demonstrated that polarization reversal can involve nucleation and domain-wall formation, with the accessible switching pathway depending on temperature, mechanical conditions, and simulation conditions [6, 7]. The absence of such nucleation-facilitating features may therefore contribute to the larger coercive field obtained in the present simulations.
To examine the corresponding atomistic switching mechanism, the populations of the local Ti–O off-centering vectors along the six crystallographic directions were tracked throughout the polarization-reversal process. As shown in Figure S5 (b), the local polar distortions do not reverse directly from one polarization direction to its opposite. Instead, a substantial population transiently rotates toward an approximately 90° orientation before reaching the final reversed state. This intermediate reorientation is accompanied by a transient increase in the transverse component of the macroscopic polarization in Figure S5 (a).
The approximately 90°-mediated switching pathway is consistent with switching behavior reported in experimental studies of BTO single crystals and with theoretical analyses based on Landau–Ginzburg–Devonshire theory [5, 8]. This agreement provides an additional validation that the rSCAN-based MACEField captures not only macroscopic electric-field response but also the underlying local Ti–O polarization dynamics.
Robustness under Constrained In-Plane Lattice Conditions
In the strain-dependent calculations discussed in the main text, the in-plane lattice axes were constrained to maintain the prescribed biaxial strain. Because the temperature-dependent calculations in Figure 1 were performed without this constraint, we additionally evaluated the temperature dependence under the same in-plane lattice constraint used for the strain-dependent calculations. This comparison allows us to assess whether the different lattice constraints affect the temperature-dependent dielectric response or the relationship between and .
Figure S6 (a) shows the relative dielectric constant as a function of normalized temperature under the constrained in-plane lattice condition. The dielectric constant decreases from 2201 at 260 K to 731 at 320 K, demonstrating that the characteristic decrease in dielectric response with increasing temperature is retained even when the in-plane lattice degrees of freedom are constrained.
We further evaluated the field-induced orientational response under the same conditions. As shown in Figure S6 (b), the constrained-temperature dataset also exhibits a strong correlation between and . A linear regression gives with .
The slope is close to that of the combined temperature- and strain-dependent relationship presented in the main text. This result confirms that the relationship between the dielectric response and the field-induced angular response of the local Ti–O polar structure is retained even when the in-plane lattice degrees of freedom are constrained.
Contact author: [email protected]
*Contact author: [email protected]
*Contact author: [email protected]
References
- [1]R. Sahashi, P.-Y. Chen, and T. Mizoguchi, LDA-based machine learning force field for accurate electric-field-driven ferroelectric response in BaTiO3, J. Ceram. Soc. Jpn. 134, 431 (2026).DOI
- [2]P.-Y. Chen and T. Mizoguchi, Transition from homogeneous to domain-wall-mediated polarization switching in BaTiO₃: A machine-learning molecular dynamics study, Phys. Rev. B (to be published).
- [3]W. J. Merz, The electric and optical behavior of BaTiO₃ single-domain crystals, Physical Review 76, 1221 (1949).
- [4]L. Gigli, M. Veit, M. Kotiuga, G. Pizzi, N. Marzari, and M. Ceriotti, Thermodynamics and dielectric response of BaTiO₃ by data-driven modeling, npj Comput. Mater. 8, 209 (2022).DOI
- [5]B. Jiang, Y. Bai, W. Chu, Y. Su, and L. Qiao, Direct observation of two 90° steps of 180° domain switching in BaTiO₃ single crystal under an antiparallel electric field, Appl. Phys. Lett. 93, 152905 (2008).DOI
- [6]H. Azuma, T. Ogawa, S. Ogata, R. Kobayashi, M. Uranagase, T. Tsuzuki, and F. Wendler, Unique temperature-dependence of polarization switching paths in ferroelectric BaTiO₃: A molecular dynamics simulation study, Acta Mater. 296, 121216 (2025).
- [7]P. Chen and T. Mizoguchi, Effect of uniaxial compressive stress on polarization switching and domain wall formation in tetragonal phase BaTiO₃ via machine learning potential, Mater. Des. 265, 115851 (2026).
- [8]Y. Li, J. Wang, and F. Li, Intrinsic polarization switching in BaTiO₃ crystal under uniaxial electromechanical loading, Phys. Rev. B 94, 184108 (2016).DOI