1 Introduction

Since the discovery of ferroelectrics, ferroelectric materials have been exploited in abundant applications, such as memories, energy harvestors, and so on [1, 2]. Recently, ferroelectrics, especially relaxor ferroelectrics, have been considered as the potential candidates for the electrocaloric (EC) solid-state cooling prototypes, which is a highly efficient and environmentally benign refrigeration technology [3,4,5,6]. Here, the electrocaloric effect (ECE) is defined as the adiabatic change in entropy (ΔS)/ temperature (ΔT) caused by the operation of an external electric field in ferroelectrics.

Aiming at achieving the high EC response with low voltage required and near room temperature, a great number of researches focus on the search of environmental friendly electrocalorics, such as lead-free BaTiO3-based ceramics [7, 8], and the investigation of the strategies for improving the EC response, for instance, maximizing the number of coexisting phases [9, 10]. It has been reported that the coexistence of multiple phases near an invariant critical point (ICP) would lead to a significant increase of the entropy and meanwhile broaden the effective temperature range of ECE in Zr and Sn hybrid doped BaTiO3 relaxor ferroelectrics [11]. In addition, modified BaTiO3 system with different isovalent substituting elements such as Ca2+, Sr2+ substituted Ba site [12, 13], or Hf4+, Ce4+ doped Ti site [14, 15], have also been widely reported to further improve the EC response. Furthermore, a little bit of attention has recently been paid to the heterovalent substitution which would introduce the oxygen or cation vacancies in the system. For example, a large EC response can be observed by the aliovalent substitution of Y and Mn into BaTiO3 [16]. Additionally, the co-doping of the donor Ce3+ into Ba site and the acceptor Mn3+ into Ti site would improve the EC response and broaden the operating temperature range, by regulating the oxygen vacancies and the grain size of ceramics [17]. These results illustrate that the ECE would be effectively enhanced and regulated through the aliovalent substitution in BaTiO3 system.

Since the introduction of the heterovalent element into Ba or Ti site would induce oxygen or cation vacancies, the dielectric permittivity and the polarization of this system would change subsequently [18]. The energy storage performance has been confirmed to be closely dependent on the ferroelectric behaviors [19], and thus the energy storage density and efficiency could also be tunable by the aliovalent dopants into Ba or Ti site of BaTiO3 ceramics.

In this work, the heterovalent-substitution dependent on the dielectric, ferroelectric, electrocaloric, and energy storage behaviors have been systematically investigated through acceptor Mn2+/Mn3+ or donor Nb5+ doping into Ti4+ site of Ba(Sn, Ti)O3 ceramics. It is found that both the electrocaloric and energy storage performances can be effectively regulated by the replacement of only 0.5 % Ti with aliovalent elements, and more significantly, these properties are obviously enhanced in 0.5 % Mn-doped BaTiO3 system.

2 Materials and methods

The BaSn0.11Ti0.89−xMxO3 (x = 0, 0.005, and 0.01, M = Mn or Nb) ceramics were prepared by the solid-state reaction method. High purity reagents BaCO3 (99 % pure), TiO2 (99 % pure), SnO2 (99 % pure), MnO (99 % pure), and Nb2O5 (99 % pure) powders were used. The stoichiometric precursors were grinded to achieve the uniform mixture and then calcined at 1100 oC for 2 h to stimulate the solid-state reaction. Followed by further milling, the powders were pressed into a disk shaped pellet of 12 mm diameter and 1 mm thickness, after mixing with 5 wt% polyvinyl alcohol (PVA) solution. The pellets were further sintered at 1400 oC~1450 oC for 2 h.

X-ray diffraction data of sintered powders were obtained using an X-ray diffractometer (Model X’Pert PRO, PANalytical, Almelo, Holland) with Cu Kα radiation. The surface morphology was observed by a scanning electron microscope (SEM, Model JSM6380-LV, JEOL, Tokyo, Japan). Prior to microstructural analysis, the ceramic samples were fine-polished and then thermal etched at 1300 oC for 1 h [20,21,22,23]. For electrical measurements, silver electrodes were used on the polished surfaces of the ceramics. The dielectric properties were measured by an E4980A impedance analyzer (Agilent, USA). Furthermore, the ferroelectric behavior was carried out using ferroelectric analyzer (TF2000, aix-ACCT, Aachen, Germany).

3 Results

3.1 Structural and morphological properties

The XRD patterns of BaSn0.11Ti0.89O3 (BST), BaSn0.11Ti0.885Mn0.005O3 (BSTMn0.005), BaSn0.11Ti0.88Mn0.01O3 (BSTMn0.01), BaSn0.11Ti0.885Nb0.005O3 (BSTNb0.005), and BaSn0.11Ti0.88Nb0.01O3 (BSTNb0.01) ceramics recorded at room temperature are displayed in Fig. 1. As these results indicated, all the samples have a pure perovskite structure without any secondary phase or impurities, and exhibit tetragonal phase with P4mn space group. With the introduction of Mn/Nb dopants, all the diffraction peaks shift to the lower angle and thus the lattice parameters increase in the Mn/Nb-modified BST system, which is caused by the larger-sized doping ions (Mn2+: 0.067 nm and Nb5+: 0.064 nm) than the equivalent radius of 0.0614 nm for B-site ions consisted of 89 %Ti4+ (0.0605 nm) and 11 %Sn4+ (0.069 nm). It means that the solid state reaction has occurred in the specified sintering condition and the doping Mn2+ and Nb5+ ions all enter into Ba(Sn, Ti)O3 lattice.

Fig. 1
figure 1

(color online) XRD patterns of BST, BSTMn0.005, BSTMn0.01, BSTNb0.005, and BSTNb0.01 ceramics recorded at room temperature

Full size image

Figure 2 shows the surface SEM images of BST, Mn or Nb-doped BST ceramics. It can be seen that all these samples are dense and the average grain sizes are around 1 μm for BST ceramics, 3 ~ 4 μm for BSTMn0.005 and BSTNb0.005, and 1 ~ 3 μm for BSTMn0.01 and BSTNb0.01. It is clear that the relatively larger grain size exists in 0.5 % Mn/Nb-doped BST ceramic samples. As we know, the Mn/Nb entering into the Ti site would cause the formation of the oxygen vacancies/cation vacancies. The vacant sites would be helpful for the mass transportation, which might be responsible for the larger grain size in 0.5 % Mn/Nb-doped BST system [17]. However, when the Mn/Nb content is increased up to 1 %, a part of Mn/Nb would accumulate at the grain boundaries, which would hinder the further motion of grain boundaries and the further growth of the grains [24].

Fig. 2
figure 2

The surface SEM images of BST (a), BSTMn0.005 (b), BSTMn0.01 (c), BSTNb0.005 (d), and BSTNb0.01 (e)

Full size image

3.2 Dielectric and ferroelectric properties

The temperature dependences of the dielectric permittivity at different frequencies for BST, and Mn/Nb doped BST are shown in Fig. 3. The temperature at which the dielectric constant reaches a maximum value and the maximum permittivity are assigned to be Tm and εm, respectively. The values of Tm and εm for all these ceramics are compared in Table 1. It is found that the maximum dielectric permittivity of 3889 for BST is lower than that reported in the previous studies [25, 26], since the dielectric behaviors are closely dependent on the grain sizes and microstructures [27,28,29], and SEM images display relatively small grain sizes in this system (1 μm for BST ceramics, 3 ~ 4 μm for BSTMn0.005 and BSTNb0.005, and 1 ~ 3 μm for BSTMn0.01 and BSTNb0.01). In addition, the highest εm exists in BSTMn0.005 ceramics, illustrating that the improved ECE and energy storage can be expected in the 0.5 %-doped BST system. The dielectric properties display a relaxor-like ferroelectric behavior, since all the samples display the broad peaks at Tm and the values of Tm shifts to higher temperature with the increasing frequency. The fact that the decreasing Tm with the increase of the Mn/Nb content further confirms that the Mn/Nb is successfully doped into Ti/Sn site [30]. It is found that Nb doping would increase the relaxor behaviors and broaden the transition temperature range [18, 31].

Table 1 The comparison of the dielectric, ferroelectric, electrocaloric, and energy storage performances of Mn/Nb-doped Ba(Sn, Ti)O3 ceramics
Full size table
Fig. 3
figure 3

(color online) The temperature dependence of the dielectric permittivity at different frequencies for BST (a), BSTMn0.005 (b), BSTMn0.01 (c), BSTNb0.005 (d), and BSTNb0.01 (e) ceramics. The comparison of the permittivity tested at 10 kHz for all these samples is shown in (f)

Full size image

Following the Curie-Weiss law described as

$$\frac{1}{{{\varepsilon _r}}}=\frac{{T - {T_0}}}{C}$$
(1)

the dielectric behaviors above Tm can be fitted. In Eq. (1), εr represents the dielectric constant, T0 is the Curie-Wiess temperature for the second phase transition, and C is the Curie constant. The fitting parameter C is 9.16⋅104 K for BST, 9.55⋅104 K for BSTMn0.005, 1.04⋅105 K for BSTMn0.01, 1.08⋅105 K for BSTNb0.005, and 1.28⋅105 K for BSTNb0.01, respectively. These fitting parameters will be useful for calculating the ECE of this system.

In order to further evaluate the EC response and energy storage performance, the polarization-electric field (P-E) hysteresis loops at various temperatures (300 K ~ 358 K) for these ceramics are tested under different electric fields and at an increment of 5 K, as shown in Fig. 4. The test frequency is 10 Hz for all of these measurements. The typical relaxor-ferroelectric hysteresis loops are achieved for all these samples, according with the dielectric results. However, the leakage exists around room temperature in BSTMn0.01 and BSTNb0.01 (see Fig. 4(c) and (e)), due to the ionic accumulation at the grain boundaries, and the more electrovalent mismatch and the charge fluctuation in samples with more heterovalent dopants. It is found that the shapes of P-E loops shrink and the intensities of the polarization decrease as the temperature increases. The polarization decreases slightly with the increasing temperature, confirming the relaxor behavior of this system.

Fig. 4
figure 4

(color online) The P-E hysteresis loops of BST, and Mn/Nb-doped BST samples, recorded at different temperatures

Full size image

The room-temperature polarizations of these ceramics under 80 kV cm− 1 are listed in Table 1. 0.5 % Mn substituted BST exhibits the enhanced polarization compared with other samples. Combined with the largest εm in BSTMn0.005, the dielectric and polarization results demonstrate that the EC response would be improved by introducing the heterovalent 0.5 % Mn into BST relaxors.

3.3 Electrocaloric effect

Based on P-E loops and dielectric behaviors, the ECE, i.e., the entropy change (ΔS) and the temperature change (ΔT) under applied electric fields, can be evaluated by the indirect methods from Maxwell equations and Landau-Ginzburg-Devonshire (LGD) phenomenological theory as follows [25, 32, 33]

$$\Delta S=\frac{1}{\rho }\int_{{{E_1}}}^{{{E_2}}} {{{\left( {\frac{{\partial P}}{{\partial T}}} \right)}_E}dE}$$
(2)
$$\Delta T= - \frac{1}{\rho }\int_{{{E_1}}}^{{{E_2}}} {\frac{T}{{{c_E}}}{{\left( {\frac{{\partial P}}{{\partial T}}} \right)}_E}dE}$$
(3)
$$\Delta S= - \frac{1}{2}[\beta (T){P^2}(E,T){\text{ }} - \beta \left( T \right){P^2}\left( {0,T} \right)]$$
(4)
$$\Delta T=\frac{T}{{2{c_E}}}[\beta \left( T \right){P^2}\left( {E,T} \right) - \beta \left( T \right){P^2}\left( {0,T} \right)]$$
(5)

where ρ is the density and cE is the specific heat capacity of the ceramics; T is the ambient temperature; E1 and E2 are the initial and final applied fields, respectively, where E1 = 0 and E2 are the maximum applied fields in this work. Here, β = (ε0C)−1, which is related to the Curie constant C, is considered as a temperature-independent coefficient.

Following Maxwell relations (Eqs. (2) and (3)) and LGD theory (Eqs. (4) and (5)), the changes in polarization and permittivity would reflect the corresponding changes in the entropy and the temperature as well. Figures 5 and 6 depict the variations of |ΔT|, deduced by Maxwell relations and LGD respectively, as a function of the measured temperature for three ceramic samples. As 1 % Mn/Nb doped BST samples exhibit electric leakage in P-E loops, the calculated |ΔT| of these two ceramics may not be accurate. Hence, the EC responses of BSTMn0.01 and BSTNb0.01 are not present for comparison. As expected, the EC response is enhanced by the increase in the magnitude of the applied electric field. As shown in Fig. 5, all the samples display maximum |ΔT|, which is denoted as |ΔT|m in Table 1. One can see that the obtained |ΔT| of 0.15 K at 20 kV/cm is much lower than that in the previous report [34], because the deduced ECE following indirect methods are strongly dependent on the dielectric and ferroelectric properties however the values of the dielectric permittivity and the ferroelectric polarization are suppressed by the relatively smaller grain size. In Fig. 5(d) we compare the temperature dependent EC response of BST and 0.5 % Mn/Nb doped BST, deduced by Maxwell relations. It is seen that in 0.5 % Mn-doped BST, |ΔT|m reaches 0.72 K, at 328 K and at an applied field of 70 kV cm− 1, while the values of |ΔT|m are 0.63 K for BST and 0.41 K for 0.5 % Nb-doped BST. Here, the EC strength (|ΔT|m/|ΔE|) of BSTMn0.005 achieves 0.1⋅10− 6 KmV− 1. It is interesting to note that the temperatures at which |ΔT|m appears are corresponding to the Tm obtained from dielectric properties. Based on the Maxwell relations, EC responses are closely associated with \({\left( {\frac{{\partial P}}{{\partial T}}} \right)_E}\), hence the |ΔT| would be sensitive to the changes of the polarization with the temperature and the variation around the transition temperature would be more obvious than that at other temperatures. Therefore, the optimum EC response over a broad temperature range (300 K ~ 358 K) is obtained in heterovalent-substituting BST with Mn content of 0.5 %, compared with that of BST and BSTNb0.005.

Fig. 5
figure 5

(color online) The temperature dependence of the electrocaloric response for BST and Mn/Nb-doped BST samples, deduced by Maxwell relations

Full size image
Fig. 6
figure 6

(color online) The temperature dependence of the electrocaloric response for BST and Mn/Nb-doped BST samples, evaluated using Landau-Ginzburg-Devonshire (LGD) phenomenological theory

Full size image

However, since the Maxwell relations are derived based on the reversible and ergodic thermodynamic process, the Maxwell relations would not be accurate if it is applied to relaxor ferroelectrics such as Sn, Mn/Nb-doped BaTiO3 system discussed here [32, 35,36,37]. Thus, the ECE deduced based on LGD thermodynamic theory, which is widely utilized to describe the macroscopic phenomena that occur in the polar materials, is shown in Fig. 6. One can see that both methods show the same trend with a maximum around the phase transition temperature (328 K ~ 333 K). In Fig. 6(d) we also compare the EC response deduced by LGD model for three ceramics, and it is found that a high cooling value of 1.31 K, and a large EC strength of 0.19 ⋅10− 6 KmV− 1, are achieved in 0.5 % Mn-doped BST, at the temperature of 328 K. Specifically, following LGD model, a wide temperature range (300 K ~ 358 K) EC response can also be observed in this sample, which is highly desirable for cooling devices.

3.4 Energy storage performance

In order to compare the energy storage performance of BaSn0.11Ti0.89O3 ceramics with different dopants, the energy storage density (W1) and energy storage efficiency (η) have been calculated by using the data generated from P-E loops, based on the following two formula

$${W_1}=\int_{{{P_r}}}^{{{P_{\hbox{max} }}}} {EdP}$$
(6)
$$\eta =\frac{{{W_1}}}{{{W_1}+{W_2}}} \times 100\%$$
(7)

where E is the applied electric field, P is the polarization, Pr and Pmax are the remnant polarization and the maximum polarization respectively, and W2 represents the energy loss density. The temperature dependence of the energy storage density and efficiency for 0.5 % Mn/Nb-doped BST ceramics is displayed in Fig. 7, and the values of W1 and η at room temperature are further compared in Table 1. Through the introduction of 0.5 % Mn, both the energy storage density and the efficiency are effectively improved over the whole tested temperature range. Only 0.5 % content of Mn in BST ceramics can increase the room-temperature energy storage density up to 29 % improvement, and the efficiency 32 % improvement, compared with the energy storage performance of BST. The maximum energy storage density of 0.24 J cm− 3 at 343 K and under 70 kV cm− 1 can be obtained in BSTMn0.005, and meanwhile the maximal efficiency reaches a level as high as 87.11 %, illustrating that the greater energy storage performance exhibits in Mn-doped system. However, as shown in Fig. 7(d), the partial replacement of Ti by Nb would suppress the energy storage density and efficiency as well.

Fig. 7
figure 7

(color online) The temperature dependence of the energy storage density and efficiency for BST and Mn/Nb-doped BST samples

Full size image

Table 2 compares the EC response and energy storage performances of the BST, BSTMn0.005, and BSTNb0.005 prepared in this study with those of the lead-free ferroelectric materials (especially BaTiO3-based ceramics). It can be seen that the acceptor-doped Ba(Sn, Ti)O3 in this work, i.e., BSTMn0.005, exhibits both the superior EC and energy storage behaviors: the fact that the maximal EC response of 1.31 K, energy storage density of 0.24 J cm− 3 and efficiency of 87.11 % can be induced by an applied electric field of 70 kV cm− 1, is comparable or even higher than those of other BaTiO3-based lead-free ceramics.

Table 2 Comparison of EC and energy storage performances of Mn/Nb modified-Ba(Sn, Ti)O3 ceramics with those of other lead-free ceramics (especially BaTiO3-based ceramics) obtained in the literature
Full size table

4 Discussion

As mentioned above, compared with BST and Nb-doped BST ceramics, the 0.5 % Mn-doped BST samples show the enhanced electrocaloric and energy storage performances. For Mn/Nb-doped BST samples, the element Mn, substituted for B-site Ti4+, would mainly exist as acceptor Mn2+ and Mn3+ [45], while Nb would be successfully incorporated as donor Nb5+ into Ti4+ site [24]. The small amount (0.5 %) of aliovalent doping such as Mn2+, Mn3+, or Nb5+, would induce the local internal stress due to the local size mismatch, and the local charge fluctuation due to the local charge imbalance. These local internal stress and local charge fluctuation would be beneficial for the polarization response, EC response and energy storage performance as well. While, the higher content (1 %) of aliovalent dopants could introduce the ionic accumulation, and the random electric fields and the random strain field, which would depress the polarization [46]. Thus, the P-E loops of BSTMn0.01 and BSTNb0.01 exhibit electric leakage around room temperature.

On the other hand, the introduction of acceptor Mn2+ and Mn3+ dopants would cause the formation of oxygen vacancies. The small ratio of oxygen vacancies could help the mass and cation transportation, which would ease the domain switching and subsequently raise the dielectric and ferroelectric responses [17]. It explains that the improved dielectric permittivity and ferroelectric polarization exhibit in tiny-amount (0.5 %) Mn-doped BST ceramics. Nevertheless, when the composition of Mn ions increases as well as the ratio of oxygen vacancies becomes higher, then the oxygen vacancies would pin the movement of the ferroelectric domain walls and subsequently reduce the dielectric and ferroelectric responses in 1 % Mn-doped BST samples [47,48,49]. Meanwhile, the donor Nb5+ doping induces the diffuse transition behavior and reduces the polarization [31].

Therefore, attributed to the two effects mentioned above, the enhanced dielectric, ferroelectric, electrocaloric, and energy storage behaviors can be observed in lower-content Mn-doped BST samples. While in the Nb-doped BST, these properties are depressed, because the donor doping could affect the polarization in a dominant way rather than the local stress and charge fluctuation.

5 Conclusions

In summary, the ECE of Mn or Nb doped Ba(Sn, Ti)O3 relaxor ferroelectrics has been evaluated using Maxwell relations and LGD thermodynamic theory. The effect of the heterovalent substitution such as acceptor doping and donor doping on the EC response and energy storage performance of this system has been clarified. The introduction of aliovalent Mn or Nb elements into Ti site would regulate both the EC and energy storage behaviors, and the enhanced EC response, recoverable energy storage density and efficiency can be observed in 0.5 % Mn-doped Ba(Sn, Ti)O3 ceramics. These results demonstrate that the tunable electrocaloric effect and energy storage performance can be achieved by the acceptor or the donor doping in Ba(Sn, Ti)O3 relaxor ferroelectrics.