betaFeedback

WorksPhysics and Astronomy

Origin of negative electrocaloric effect in Pnma-type antiferroelectric perovskites

Abstract

Anomalous electrocaloric effect (ECE) with decreasing temperature upon application of an electric field is known to occur in antiferroelectrics (AFEs), and previous understanding refers to the field-induced canting of electric dipoles if there is no phase transitions. Here, we use a first-principle-based method to study the ECE in Nd-substituted BiFeO3 (BNFO) perovskite solid solutions, which has the Pnma-type AFE ground state. We demonstrate another scenario to achieve and explain anomalous ECE, emphasizing that explicit consideration of octahedral tiltings is indispensable for a correct understanding. This mechanism may be general for AFEs for which the antipolar mode is not the primary order parameter. We also find that the negative ECE can reach a large magnitude in BNFO.

Electrocaloric effect (ECE) can make temperature change via adiabatic application (or removal) of an electric field, providing an efficient approach for cooling or heating [1, 2, 3]. While ferroelectric (FE) or relaxor materials typically have “normal” positive sign of ECE, i.e., the temperature increases by applying a voltage, antiferroelectrics (AFEs) are known to have anomalous ECE that can yield an opposite sign [4, 5, 6]. These two types of ECE can be utilized in combination to improve the performance of cooling/heating devices.

Such negative (or inverse) caloric effect is also known to occur in other occasions, e.g., magnetic Heusler alloys, transitions between FE phases of different polarization directions, and application of an electric field against the polarization of a FE phase without switching [7, 8, 9]; however, its origin in AFEs is less well understood. AFEs materials are characterized by anti-polar atomic distortions that can be switched to a FE state under an electric field, and two mechanisms to explain their negative ECE with no AFE-FE transition have been proposed: 1) the “dipole-canting” model that dipolar entropy increases by misaligning the anti-parallel dipoles upon application of the field [5]; 2) the perturbative theory based on the Maxwell relation that only temperature and electric field dependencies of polarization need to be considered [10]. Interestingly, all these mechanisms only take the electric degrees of freedom explicitly into account. In contrast, most of the known AFEs are neither proper type (that is, the AFE phase is rarely driven by an AFE soft mode [11]), nor systems with the anti-polar mode being the only significant order parameter. In fact, quite often, the AFE mode is secondary and coupled to other degrees of freedom, such as the octahedral tiltings in perovskites. For instance, PZO has a strong instability of anti-phase octahedral tilting [12], while in PnmaPnma-type perovskite, such as rare earth orthoferrites and CsPbI3_3, the anti-polar distortion arises from the condensation of both the in-plane anti-phase (ωR,x,y\omega_{R,x,y}) and out-of-plane in-phase tiltings (ωM\omega_M) via trilinear coupling [13]. Although these tilting modes are non-polar, they couple strongly with the polar and anti-polar modes so that they can be influenced by the electric field as well, and in consequence contribute to the ECE.

To get a deeper understanding of the (negative) ECE in AFE, analysis based on Landau models involving the most relevant degrees of freedom have been proved to be very useful [14, 15, 16, 17], and it may thus be necessary that all the important order parameters are taken into consideration. Furthermore, some previous phenomenological models are often over simplified, since only one dimension is assumed [4]. In reality, the direction of the applied field with respect to the crystallographic axis should have different effects regarding ECE. In this Letter, we take the antiferroelectric Nd substituted BiFeO3_3 (BNFO) solid solution as an example and demonstrate that the octahedral tiltings can have very important effect on the sign and magnitude of the ECE. We also construct a phenomenological model that allows us to rationalize the contributions of each degree of freedom. In particular, the dipoles alone are found to be insufficient to explain the negative ECE, while contributions from the in-phase and anti-phase tilting modes are indispensable. Moreover, BNFO is predicted to yield rather large negative ECE close to the AFE-to-FE transition.

BiFeO3_3 (BFO) stabilizes in a R3cR3c ground state, but rare-earth doping with composition larger than 20–30% is sufficient to alter it to the PnmaPnma structure [18]. Here, we adopt the effective Hamiltonian scheme of Ref. [19] to study the Bi0.6_{0.6}Nd0.4_{0.4}FeO3_3 solid solution under electric field at finite temperatures. With this composition, BNFO is stabilized in the AFE PnmaPnma phase at room temperature and can transform to a FE state under an electric field [18, 19, 20, 21, 22]. The solid solutions are simulated by a 12×12×1212 \times12 \times12 supercell (containing 8,640 atoms) using Monte-Carlo (MC) simulations, in which the Bi and Nd atoms are randomly distributed (see Supplemental Material (SM) Section S1 for details [23]).

Effect of electric field on the temperature dependence of the order parameters

Figure 1. Effect of E\mathcal{E} field on the temperature dependence of the order parameters. (a),(b) [1] field of 0.87 MV/cm (solid lines), in comparison with zero field (dashed lines), (c),(d) ibid for [11ˉ\bar{1}0] field of 0.87 MV/cm, (e),(f) ibid for [110] field of 0.61 MV/cm. The four relevant order parameters are anti-phase octahedral tilting ωR\omega_R, in-phase tilting ωM\omega_M, polarization PP, and antiferroelectric vector XX. Note that Px=−PyP_x=-P_y for [11ˉ\bar{1}0] field. The vertical dotted (or dashed) lines delimit different phases under finite (or zero) field.

First, let us check how the order parameters are influenced by the electric field E\mathcal{E} [24], using the effective Hamiltonian scheme. Note that the initial PnmaPnma structure has zero polarization, an in-plane anti-polar vector XX along the [110] direction, and an a−a−c+a^-a^-c^+ tilting in Glazer’s notation (which corresponds to finite xx- and yy-components of the antiphase tilting vector, ωR,x=ωR,y\omega_{R,x}=\omega_{R,y}, and finite zz-component of the in-phase tilting vector, ωM,z\omega_{M,z}). Three representative field directions are investigated (Fig. 1), together with the zero-field data (dashed curves) for comparison (for other $\mathcal{E}$ magnitudes, see SM Section S2 {citep}[23]). With no applied field, the $Pnma$ phase transforms to the paraelectric (PE) cubic phase at 1400 K (dashed lines in {xref}Fig. 1`).

Under [1] field, we consider a representative case with E=0.87\mathcal{E}=0.87 MV/cm, with which the AFE state transforms to the ferroelectric P4mmP4mm phase at 880 K, characterized by a large polarization PzP_z along [1] and no octahedral tiltings.

Effect of electric field on the temperature dependence of the order parameters

Figure 1 (PDF p. 2).

In the AFE state, one can see that the temperature dependence of ωM,z\omega_{M,z} does not differ much from the zero-field case, whereas PP and ωR,(x,y)\omega_{R,(x,y)} (ωR,x=ωR,y\omega_{R,x}=\omega_{R,y}) show apparent changes. The moderate field-induced change of XX can be understood to a good approximation via the change in ωR,(x,y)\omega_{R,(x,y)} since X(x,y)X_{(x,y)} should be proportional to the product ωR,(x,y)ωM,z\omega_{R,(x,y)}\omega_{M,z} – as a result of a trilinear coupling between X(x,y)X_{(x,y)}, ωR,(x,y)\omega_{R,(x,y)} and ωM,z\omega_{M,z} [25]. Finite PzP_z is induced by the field, whereas the field-induced suppression of ωR,(x,y)\omega_{R,(x,y)} is due to the competitive coupling with PzP_z.

Moreover, if the field is applied along [11ˉ\bar{1}0] (Figs. 1c and 1d), PP is induced along the same direction, i.e., a finite Px=−PyP_x=-P_y first develops, and a transition to a FE CcCc phase occurs at 790 K, a structure characterized by a polarization in ⟨uvw⟩\langle uvw\rangle direction (u>vu>v) and a−a−c−a^-a^-c^- octahedral tiltings (a>ca>c). The third direction is along [110], along which PP develops, while ωM,z\omega_{M,z} is much suppressed and ωR,(x,y)\omega_{R,(x,y)} is more or less unchanged within the AFE-based state [26]. Such AFE phase then transforms also into a CcCc phase at 780 K (Fig. 1e and 1f).

Let us now concentrate on the ECE coefficient α=∂T∂E∣S\alpha=\left.\frac{\partial T}{\partial\mathcal{E}}\right|_S with TT being temperature and SS being entropy, which can be calculated from the cumulant formula using outputs of the MC simulations[27, 28, 29, 30].

αMC=−Z∗αlattT{⟨∣u∣Etot⟩−⟨∣u∣⟩⟨Etot⟩⟨Etot2⟩−⟨Etot⟩2+2N(kBT)2}(1)\alpha_{\mathrm{MC}}=-Z^{*}\alpha_{\mathrm{latt}}T\left\{\frac{\langle|u|E_{\mathrm{tot}}\rangle-\langle|u|\rangle\langle E_{\mathrm{tot}}\rangle}{\langle E_{\mathrm{tot}}^{2}\rangle-\langle E_{\mathrm{tot}}\rangle^{2}+\frac{2N}{(k_{\mathrm{B}}T)^{2}}}\right\} \tag*{(1)}

where Z∗Z^{*} is the Born effective charge associated with the local mode, αlatt\alpha_{\mathrm{latt}} represents the lattice constant of the five-atom pseudo-cubic perovskite cell, TT is the simulation temperature, ∣u∣|u| is the supercell average of the magnitude of the local mode, EtotE_{\mathrm{tot}} is the total energy given by the effective Hamiltonian, kBk_{\mathrm{B}} is the Boltzmann constant, NN is the number of sites in the supercell, and ⟨⟩\langle\rangle denotes average over the MC sweeps at a given temperature.

For fields applied along the [1] direction (Figure 2a), similar to the case in PZO-based AFE [5, 6], α\alpha is negative in the AFE-based state, and its magnitude increases with temperature, which maximizes at the transition point where the AFE-based state disappears. Across the phase transition, α\alpha jumps to be positive in the FE state, then (slightly) increases with temperature, as such qualitative temperature dependence is known for ferroelectrics [31].

Temperature dependence of the calculated ECE coefficient, isothermal entropy change, and temperature change

Figure 2. Temperature dependence of the calculated ECE coefficient α\alpha, isothermal entropy change, and temperature change under (a)-(c) [1] (0.87 MV/cm), (d)-(f) [11ˉ\bar{1}0] (0.87 MV/cm), and (g),(i) [110] (0.61 MV/cm), respectively. Colors in ΔS\Delta S and ΔT\Delta T denote the total and individual contributions from each order parameter. αMC\alpha_{\mathrm{MC}} is computed from the cumulant approach, and αfit\alpha_{\mathrm{fit}} is a fitted curve based on the Landau model.

In order to have insightful analysis of the ECE, Figure 2b and Figure 2c also report two other quantities related to electro-caloric response, namely the total isothermal change in entropy ΔS\Delta S, and the adiabatic temperature change ΔT\Delta T, as well as their individual contributions. Practically, in order to be able to compute the total entropy change, we consider the following Landau model for PnmaPnma-type AFE [25].

F=F0+12aωR(T)ωR2+14bωRωR4+12aωM(T)ωM2+14bωMωM4+12aX(T)X2+14bXX4+12aP(T)P2+14bPP4−EP−cXωRωM+12d1P2ωR2+12d2P2ωM2.(2)\begin{aligned} F ={}& F_0 + \frac{1}{2}a_{\omega_R}(T)\omega_R^2 + \frac{1}{4}b_{\omega_R}\omega_R^4 + \frac{1}{2}a_{\omega_M}(T)\omega_M^2 + \frac{1}{4}b_{\omega_M}\omega_M^4 \\ &+ \frac{1}{2}a_X(T)X^2 + \frac{1}{4}b_X X^4 + \frac{1}{2}a_P(T)P^2 + \frac{1}{4}b_P P^4 \\ &- \mathcal{E}P - c_X\omega_R\omega_M + \frac{1}{2}d_1P^2\omega_R^2 + \frac{1}{2}d_2P^2\omega_M^2 . \tag*{(2)} \end{aligned}

Note that F0F_0 is the field-independent part of the free energy, only the quadratic coefficients aop=Aop(T−T0op)a_{\mathrm{op}}=A_{\mathrm{op}}(T-T_0^{\mathrm{op}}) have explicit temperature dependence, with “op” being the order parameter of ωR\omega_R, ωM\omega_M, PP, or XX, respectively, and T0opT_0^{\mathrm{op}} is a transition temperature for each order parameter. Also note that the tiltings are known to be the primary order parameters in many PnmaPnma-systems, while XX is secondary arising from the cXωRωMc_X\omega_R\omega_M trilinear coupling involving in-phase and anti-phase tiltings [32]. Recalling that S=−∂F∂T∣E,opS=-\left.\frac{\partial F}{\partial T}\right|_{\mathcal{E},\mathrm{op}}, the change of entropy associated with a change of E\mathcal{E} can be written as:

ΔS(T)=−AωR[ωR2(T,E2)−ωR2(T,E1)]−AωM[ωM2(T,E2)−ωM2(T,E1)]−AP[P2(T,E2)−P2(T,E1)]−AX[X2(T,E2)−X2(T,E1)]=ΔSωR+ΔSωM+ΔSP+ΔSX,(3)\begin{aligned} \Delta S(T) ={}& -A_{\omega_R}\left[\omega_R^2(T,\mathcal{E}_2)-\omega_R^2(T,\mathcal{E}_1)\right] \\ &-A_{\omega_M}\left[\omega_M^2(T,\mathcal{E}_2)-\omega_M^2(T,\mathcal{E}_1)\right] \\ &-A_P\left[P^2(T,\mathcal{E}_2)-P^2(T,\mathcal{E}_1)\right] \\ &-A_X\left[X^2(T,\mathcal{E}_2)-X^2(T,\mathcal{E}_1)\right] \\ ={}&\Delta S_{\omega_R}+\Delta S_{\omega_M}+\Delta S_P+\Delta S_X , \tag*{(3)} \end{aligned}

where E1\mathcal{E}_1 and E2\mathcal{E}_2 are the initial and final electric field, respectively, and the values of the order parameters are obtained from MC simulations. The adiabatic temperature change due to ECE can be further obtained via ΔT=−(T/CE)ΔS\Delta T=-(T/C_E)\Delta S, where CEC_E is the phonon specific heat [33] – which is calculated here by density functional theory (see SM Section S3 [23]). We thus have:

ΔT=−TCEΔS=ΔTωR+ΔTωM+ΔTP+ΔTX.(4)\Delta T=-\frac{T}{C_E}\Delta S=\Delta T_{\omega_R}+\Delta T_{\omega_M}+\Delta T_P+\Delta T_X . \tag*{(4)}

Equations (3) and (4) are used to evaluate the entropy and temperature change (as well as their individual contributions associated with ωR\omega_R, ωM\omega_M, PP and XX), in which each order parameter is directly obtained from MC simulations while the coefficients Aop=−aop0-K/T0opA_{\mathrm{op}}=-a_{\mathrm{op}}^{0\text{-}\mathrm{K}}/T_0^{\mathrm{op}} are extracted by the following two steps: (1) finding aop0-Ka_{\mathrm{op}}^{0\text{-}\mathrm{K}} from fitting the 0 K double-well energies based on our effective Hamiltonian to the Landau model (Eq. (2)); (2) determining T0opT_0^{\mathrm{op}} from fitting α\alpha calculated from the cumulant formula to that based on the Landau model with αfit=ΔT/(2ΔE)\alpha_{\mathrm{fit}}=\Delta T/(2\Delta\mathcal{E}) for ΔE=0.02\Delta\mathcal{E}=0.02 MV/cm (see SM Section S4 for details[23]). Note that αfit\alpha_{\mathrm{fit}} is shown in Figure 2a and agrees rather well (over a large temperature range) with the α\alpha directly obtained from the MC simulations in the AFE-based state, therefore demonstrating the relevance of the aforementioned Landau model. Also note that the obtained T0opT_0^{\mathrm{op}} reflects the depth of the energy wells. One can then realize from Eq. (4) two important features for each ΔTop\Delta T_{\mathrm{op}} associated with a specific order parameter (op), that are: (1) such change is directly proportional to the AopA_{\mathrm{op}} coefficient related to that order parameter; and (2) such change involves the difference in the square of the order parameter between two electric fields, e.g., [ωR2(T,E2)−ωR2(T,E1)]\left[\omega_R^2(T,\mathcal{E}_2)-\omega_R^2(T,\mathcal{E}_1)\right] for the anti-phase tilting. These four different individual changes in temperature are reported in Figure 2c for field along [1].

Here, we only focus on the ECE in the AFE state, since latent heat needs to be considered when first-order AFE-to-FE transition occurs – which is not included in Eq. 3. Therefore, ECE involving the FE and PE states are not investigated within the scope of the present study. ΔS\Delta S and ΔT\Delta T in Figure 2b and Figure 2c, respectively, are practically calculated from Eqs. (3) and (4) with E1=0\mathcal{E}_1=0 and E2=0.87\mathcal{E}_2=0.87 MV/cm being along [1]. We numerically find that ΔT\Delta T is negative and bears qualitatively similar TT dependence as that of α\alpha, i.e., its magnitude increases with increasing temperature and maximizes at the transition point. On the other hand, ΔS\Delta S has an opposite sign to that of ΔT\Delta T, as dictated by their relationship ΔT=−(T/CE)ΔS\Delta T=-(T/C_E)\Delta S. To understand how ECE is associated with each order parameter, we can focus on the different ΔTop\Delta T_{\mathrm{op}} displayed in Figure 2c. For this [1] field direction, ωR\omega_R, ωM\omega_M, and XX show negative contributions, with the magnitude from ωR\omega_R being the largest and that from ωM\omega_M being negligible, together with the rather large positive contribution from PP. According to Eq. 4, the negative (or positive) signs of the four different ΔTop\Delta T_{\mathrm{op}} can be understood by the fact that the corresponding order parameter is suppressed (or enhanced) with the application of the field. In fact, as suggested by Eqs. (3) and (4), a large magnitude of ΔTop\Delta T_{\mathrm{op}} relies on two factors, either those with large field-induced changes in their order parameter such as ωR\omega_R and PP here, or those with large coefficient AopA_{\mathrm{op}} such as XX here [34], whereas ΔTωM\Delta T_{\omega_M} for field along [1] is negligible since both AωMA_{\omega_M} and the field-induced change in ωM\omega_M are very small.

The same procedure is adopted to study the field directions along [1̄10] and [110]. For the [1̄10] field, as shown in Figure 2d–Figure 2f, a qualitatively similar ECE behavior is found compared with those from the [1] field (namely, a negative α\alpha having a magnitude increasing when approaching the AFE-to-FE transition), except that the predicted α\alpha, ΔS\Delta S and ΔT\Delta T are smaller in magnitude, e.g., the largest negative α=−1.0\alpha=-1.0 mK·kV/cm, ΔT=−9.2\Delta T=-9.2 K for the [1̄10] field, to be compared with −7.8-7.8 mK·kV/cm and −22.8-22.8 K for the [1] field, respectively; however, the relative contributions of the order parameters share qualitatively similar characteristics, i.e., ωR\omega_R and XX are associated to negative ΔTωR\Delta T_{\omega_R} and ΔTX\Delta T_X, respectively, with ωR\omega_R contributing the most, together with the opposite effect from PP (positive ΔTP\Delta T_P) and negligible contribution from ωM\omega_M (i.e., weak ΔTωM\Delta T_{\omega_M}). Note that α\alpha and ΔT\Delta T are also negative and increase in magnitude with TT within the AFE-based state for the [110] field (Figure 2g–Figure 2i), and their magnitude is about −2.5-2.5 mK·kV/cm and −3.9-3.9 K, respectively, at the transition point.

However, albeit the qualitative similarities from the sign of ΔTP\Delta T_P and ΔTX\Delta T_X for all three E\mathcal{E} directions considered in Figure 2, ΔTωR\Delta T_{\omega_R} becomes positive and ΔTωM\Delta T_{\omega_M} shows large negative contribution only for E\mathcal{E} along [110]. Such unique features are associated to the small enhancement of ωR\omega_R and large suppression of ωM\omega_M under [110] fields (see Fig. 1e). Further, the small ECE for [110] field can be well understood by the fact that the negative ΔTωM\Delta T_{\omega_M} and ΔTX\Delta T_X contributions are largely cancelled by the positive ΔTP\Delta T_P and ΔTωR\Delta T_{\omega_R} contributions. It is also worth mentioning that ΔTP\Delta T_P is always positive and ΔTX\Delta T_X is always negative, as PP increases while XX decreases upon application of the E\mathcal{E} field (see Eq. (4)). In other words, being consistent with Eq. (3), the applied field makes PP more ordered and XX more disordered.

It is interesting to note that the contrast of different ECE behaviors from various field directions are related to the relative orientation between E\mathcal{E} and XX; for instance, [1] and [1̄10] fields are perpendicular to the antipolar vector, whereas the [110] field is parallel to it. As detailed in SM Section S5 [23], via the calculated energy landscape, we show numerically that minimization of the total energy dictates that increasing PzP_z or P(x,yˉ)P_{(x,\bar{y})} in the PnmaPnma state would strongly suppress ωR\omega_R, whereas such suppression is much less for ωM\omega_M; and in contrast, the opposite is found for P(x,y)P_{(x,y)}, i.e., ωM\omega_M is strongly suppressed but ωR\omega_R is less correlated when P(x,y)P_{(x,y)} increases.

In fact, a continuous variation of ΔTop\Delta T_{\mathrm{op}} can be realized if the E\mathcal{E} direction changes from [110] towards [1]. In SM Section S6 [23], we include results for E\mathcal{E} lying along [111], [112], [113], and [114] directions. While the overall ΔT\Delta T remains negative and ΔTP\Delta T_P is always positive (because PP always increases with the field in the AFE-based phase), ΔTωR\Delta T_{\omega_R} becomes more negative while ΔTωM\Delta T_{\omega_M} becomes less negative, as the angle between E\mathcal{E} and XX changes from 0 to 90 degrees. Nonetheless, it is worth to note that the contributions associate to XX and PP also show opposite trend with such rotation of the field, i.e., ΔTP\Delta T_P decreases (less positive) and ΔTX\Delta T_X increases (less negative).

For each field direction, we can further change the field strength to tune the AFE transition temperature, where the maximum ECE occurs, thus manipulating the negative ECE. As depicted in Figure 3, for E\mathcal{E} along [1], [1̄10] and [110], the maximized negative ΔT\Delta T is gained by intermediate field, which allows the AFE-to-FE transition to occur at rather high temperature. More precisely, if E\mathcal{E} is too small, the FE state is absent; if E\mathcal{E} is too large, the AFE-to-FE transition happens at low temperature and ΔT\Delta T is relatively small. Remarkably, ΔT\Delta T is predicted to reach −29.0-29.0 K if E\mathcal{E} is along [1] with a magnitude of 0.70 MV/cm at 1100 K, therefore being very large compared with the previously reported value of −5.8-5.8 K in La-doped Pb(Zr,Ti)O3_3 [5]. However, if one insists in having a large negative ECE at room temperature (RT), a specific field direction and strength should be chosen, e.g., the largest negative ΔT\Delta T of −4.7-4.7 K is realized at RT with a [1] field having a magnitude of 1.13 MV/cm, with which the AFE-to-FE transition is close to RT. This value is comparable to that of PZO-based AFEs at RT [5, 6].

The ECE temperature change as a function of temperature under various electric field magnitudes and directions

Figure 3. The ECE temperature change as a function of temperature under various electric field magnitude and directions. (a) [001][001], (b) [1ˉ10][\bar{1}10], and (c) [110][110] field.

Finally, let us examine the two previously proposed mechanisms. Regarding the “dipole-canting” mechanism [5], in AFE BNFO, for all the investigated E\mathcal{E} directions, if only the dipole degrees of freedom (i.e., PP and XX) are considered, canting of dipoles indeed occurs in response to the field (Fig. 1). However, we numerically find that ΔTP\Delta T_P and ΔTX\Delta T_X have opposite signs, but the response of PP to E\mathcal{E} field is much greater than XX for all field directions. The corresponding change of temperature (ΔTP+ΔTX\Delta T_P+\Delta T_X) is thus always positive, instead of negative, which therefore can not explain the overall negative ECE. This observation strongly suggests that explicit consideration of octahedral degrees of freedom is mandatory to correctly account for the (negative) ECE in AFEs. On the other hand, the perturbative approach solely based on polarization (as functions of E\mathcal{E} field and temperature) [10] agrees reasonably well with our predictions at low fields (see SM Section S7 [23]), indicating that the Maxwell relation ∂S∂E∣T=∂P∂T∣E\left.\frac{\partial S}{\partial\mathcal{E}}\right|_T=\left.\frac{\partial P}{\partial T}\right|_{\mathcal{E}} holds for negative ECE systems with contributions from multiple degrees of freedom.

In summary, our quantitative method show that, as octahedral tiltings are often primary order parameters in AFE perovskites, they have to be taken explicitly into account to construe the negative electrocaloric effect, and mere inclusion of dipolar degrees of freedom is insufficient. We also predict that AFE BNFO solid solutions are very promising to achieve giant temperature decrease near the AFE transition upon application of an electric field. Our atomistic mechanism also suggests that enhanced negative ECE may be possible upon freezing or suppression of order parameter(s) with positive ΔTop\Delta T_{\mathrm{op}}, for instance via interfacial proximity effect.

Acknowledgments

This work is supported by the National Natural Science Foundation of China under Grant No. 12074277, Natural Science Foundation of Jiangsu Province (BK20201404), the startup fund from Soochow University and the support from Priority Academic Program Development (PAPD) of Jiangsu Higher Education Institutions. L.B. thanks the Office of Naval Research for the support under Grant No. N00014-21-1-2086.

References

References

  1. [1]T. Correlia and Q. Zhang, Electrocaloric materials: new generation of coolers (Springer, Berlin, 2014).
  2. [2]S. Fähler, U. K. Rößler, O. Kastner, J. Eckert, G. Eggeler, H. Emmerich, P. Entel, S. Müller, E. Quandt, and K. Albe, Adv. Eng. Mater. 14, 10 (2012).
  3. [3]X. Moya, S. Kar-Narayan, and N. D. Mathur, Nat. Mater. 13, 439 (2014).
  4. [4]R. Pirc, B. Rožič, J. Koruza, B. Malič, and Z. Kutnjak, Europhys. Lett. 107, 17002 (2014).
  5. [5]W. Geng, Y. Liu, X. Meng, L. Bellaiche, J. F. Scott, B. Dkhil, and A. Jiang, Adv. Mater. 27, 3165 (2015).
  6. [6]P. Vales-Castro, R. Faye, M. Velvlyi, Y. Nouchokwe, X. Perpiñà, J. M. Caicedo, X. Jordà, K. Roleder, D. Kajewski, A. Perez-Tomas, et al., Phys. Rev. B 103, 054112 (2021).
  7. [7]T. Gottschall, A. Gracia-Condal, M. Fries, A. Taubel, L. Pfeuffer, L. Manosa, A. Planes, K. P. Skokov, and O. Gutfleisch, Nat. Mater. 17, 929 (2018).
  8. [8]M. Marathe, D. Renggli, M. Sanliap, M. O. Karabasov, V. V. Shvartsman, D. C. Lupascu, A. Grünebohm, and C. Ederer, Phys. Rev. B 96, 014102 (2017).
  9. [9]A. Grünebohm, Y.-B. Ma, M. Marathe, B.-X. Xu, K. Albe, C. Kalcher, K.-C. Meyer, V. V. Shvartsman, D. C. Lupascu, and C. Ederer, Energy Technol. 6, 1491 (2018).
  10. [10]M. Graf and J. Íñiguez, Commun. Mater. 2, 1 (2021).
  11. [11]C. Milesi-Brault, C. Toulouse, E. Constable, H. Aramberri, V. Simonet, S. De Brion, H. Berger, L. Paolasini, A. Bosak, J. Íñiguez, et al., Phys. Rev. Lett. 124, 097603 (2020).
  12. [12]B. Xu, O. Hellman, and L. Bellaiche, Phys. Rev. B 100, 020102 (2019).
  13. [13]L. Bellaiche and J. Íñiguez, Phys. Rev. B 88, 014104 (2013).
  14. [14]A. Planes, T. Castan, and A. Saxena, Philos. Mag. 94, 1893 (2014).
  15. [15]J. Zhang, A. Heitmann, S. Alpay, and G. Rossetti Jr, Integr. Ferroelec. 125, 168 (2011).
  16. [16]Y.-B. Ma, N. Novak, J. Korzua, T. Yang, K. Albe, and B.-X. Xu, Phys. Rev. B 94, 100104 (2016).
  17. [17]A. Edström and C. Ederer, Phys. Rev. Lett. 124, 167201 (2020).
  18. [18]S. Karimi, I. Reaney, Y. Han, J. Pokorny, and I. Sterianou, J. Mater. Sci. 44, 5102 (2009).
  19. [19]B. Xu, D. Wang, J. Íñiguez, and L. Bellaiche, Adv. Funct. Mater. 25, 552 (2015).
  20. [20]S. Karimi, I. Reaney, I. Levin, and I. Sterianou, Appl. Phys. Lett. 94, 112903 (2009).
  21. [21]D. Kan, L. Pálová, V. Anbusathaiah, C. J. Cheng, S. Fujino, V. Nagarajan, K. M. Rabe, and I. Takeuchi, Adv. Funct. Mater. 20, 1108 (2010).
  22. [22]B. Xu, J. Íñiguez, and L. Bellaiche, Nat. Commun. 8, 15682 (2017).
  23. [23]See Supplemental Material for (1) Computational details, (2) Temperature dependencies of the order parameters for various fields, (3) DFT-calculated specific heat, (4) Determination of parameters in the Landau free-energy model, (5) Correlation between polarization and octahedral tiltings, (6) Additional E-field directions from [110] to [001] and (7) Comparison with a perturbative approach, which additionally includes Refs. [35–53].
  24. [24]BNFO is multiferroic and in principle the magnetic moments can also respond to the applied electric field. Here it is not considered since we focus on the structural order parameters for a general AFE and numerically the magnetic contribution to χ is found to the small (see Ref. [31]).
  25. [25]B. Xu, D. Wang, H. J. Zhao, J. Íñiguez, X. M. Chen, and L. Bellaiche, Adv. Funct. Mater. 25, 3626 (2015).
  26. [26]Note that by AFE-based state, we mean the AFE state with all its associated order parameters but on top of which polarization develops along the field direction.
  27. [27]Z. Jiang, S. Prokhorenko, S. Prosandeev, Y. Nahas, D. Wang, J. Íñiguez, E. Defay, and L. Bellaiche, Phys. Rev. B 96, 014114 (2017).
  28. [28]Z. Jiang, Y. Nahas, S. Prokhorenko, D. Wang, J. Íñiguez, and L. Bellaiche, Phys. Rev. B 97, 104110 (2018).
  29. [29]S. Bin-Omran, I. A. Kornev, and L. Bellaiche, Phys. Rev. B 93, 014104 (2016).
  30. [30]Z. Jiang, B. Xu, S. Prosandeev, Y. Nahas, S. Prokhorenko, J. Íñiguez, and L. Bellaiche, Phys. Rev. B 103, L100102 (2021).
  31. [31]Z. Jiang, B. Xu, S. Prosandeev, Y. Nahas, S. Prokhorenko, J. Íñiguez, and L. Bellaiche, Phys. Rev. B 103, L100102 (2021).
  32. [32]H. J. Zhao, J. Íñiguez, W. Ren, X. M. Chen, and L. Bellaiche, Phys. Rev. B 89, 174101 (2014).
  33. [33]Z. Kutnjak, B. Rožič, and R. Pirc, Wiley encyclopedia of electrical and electronics engineering , 1 (1999).
  34. [34]op with large coefficient Aₒₚ is expected to have strongly T-dependent change of the corresponding vibrational entropy.

Paper details

vlnc:W_u7mysgkp
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents