High-Temperature Energy Storage Properties of Bi_0.5Na_0.5TiO₃-0.06BaTiO₃ Thin Films

Abstract

Bi_0.5Na_0.5TiO₃-0.06BaTiO₃ (BNT-BT) thin films were prepared via both chemical solution (CSD) and pulsed laser deposition (PLD). The structural, dielectric, and ferroelectric properties were investigated. High stability of the dielectric permittivity or TCC (∆ε/ε (150 °C) ≤ ±15%) over a wide temperature range from room temperature to 300 °C was obtained. Distinctly, the CSD film showed high TCC stability with variation of ±5% up to 250 °C. Furthermore, the CSD film showed an unsaturated ferroelectric hysteresis loop characteristic of the ergodic relaxor phase. However, the PLD one exhibited an almost saturated loop characteristic of the coexistence of both ergodic and non-ergodic states. The energy storage properties of the prepared films were determined using P–E loops obtained at different temperatures. The results show that these films exhibited a stable and improved energy storage density comparable to ceramic capacitors. Moreover, the CSD film exhibited more rigidity and better energy storage density, which exceeded 1.3 J/cm³ under a weak applied field of 317 kV/cm, as well as interesting efficiency in a large temperature range. The obtained results are very promising for energy storage capacitors operating at high temperatures.

1. Introduction

The continued increase in energy consumption around the world, coupled with significant advances in renewable energy sources, has resulted in an urgent need for efficient and reliable energy storage systems. Dielectric capacitors with exceptional power density represent a vital component within the realm of energy storage technology and are crucially employed to address the ongoing advancements within the power electronics industry. In particular, solid-state ferroic capacitors have emerged as a promising and efficient material, exhibiting the ability to rapidly store and release electrical charges in harsh conditions. In comparison to batteries and electrochemical capacitors, these capacitors offer distinct advantages for applications in high-power and pulse power electronic devices, including their high power density, rapid charge–discharge capabilities, thermal and chemical stability, and long lifespan.

Recently, considerable effort was directed toward the development of a new generation of dielectric capacitors operating at high temperatures to meet the needs of modern technologies [1,2,3,4,5,6].

It should be noted that when operating at high temperatures, the two standard capacitors, COG and X7R, called type I and type II, respectively, no longer meet the current technological requirements, which require more and more performance. Indeed, COG offers a good stable capacity in an extended range of temperatures from −55 °C to 125 °C, but it has a very low variation in the capacitance (ΔC/C = ±0.54%), which limits its application to resonant circuits. Likewise, the capacity of X7R drops at high temperatures, although it offers a high capacitance (ΔC/C = ±15%) compared to a type I capacitor [7]. The high charge–discharge ability of dielectric capacitors makes X7R more convenient for pulsed power applications. For instance, the automotive and aerospace sectors need power electronics that withstand very high temperatures. Lead-based relaxor ferroelectric and antiferroelectric materials such as Pb(Zr,Ti)O₃ (PZT) [8,9] and PMN-PT [10] can perfectly meet this demand, but unfortunately the lead is toxic and can harm human health and the environment [11]. Universal legislation like European Directive 2011/65/EU 2011 has restricted the use of these harmful compounds. As an alternative, lead-free materials have attracted the attention of the scientific community, with the challenge to shape a material that can meet the demands of modern technologies, especially for pulsed power applications and in power electronic circuits [12].

Among the potential candidates, materials based on Ba_0.5Na_0.5TiO₃ (BNT) have emerged as an ideal choice for electrostatic energy storage capacitors. This interest is motivated by the structural properties of BNT, in particular the presence of a modulated phase between 200 and 300 °C in bulk material, which is associated with the depolarization temperature (T_d), and a high Curie temperature around 325 °C. In fact, BNT is among the rare perovskites that exhibit A-site disorder with the simultaneous presence of Bi³⁺ and Na⁺. The local chemical order–disorder in the A-sites is reported to be responsible for this modulation. Similar to PMN, BNT is considered a non-ergodic relaxor material because an applied electric field induces ferroelectricity in its matrix [13,14]. A polarization of 38 μC/cm² has been reported for BNT ceramic as well as a good piezoelectric coefficient of 73 pC/N [15,16]. In addition, the observation of a double pinched P–E hysteresis loop with “Antiferroelectric- like (AFE) behavior” was found to be very promising for the configuration of high-temperature capacitors. BNT is considered to have cubic symmetry with rhombohedral R3c and tetragonal P4bm polar nanoregions, with a domination of R3c at RT. However, BNT is prone to high conductivity and large coercive fields that limit its application for energy storage capacitors. Therefore, the modulation of the composition in the BNT matrix was adopted as an appropriate way to overcome these drawbacks. In the (1-x) BNT-xBaSnO₃ system, Zhenhao et al. reported the low-temperature thermal evolution of R3c and P4bm polar nanoregions that promote relaxation behavior and lead to a slim P-E hysteresis loop [17]. Further, Verma et al., when studying a (1-x)(Bi_0.5Na_0.5)TiO₃–xAgTaO₃ solid solution, showed that the system undergoes a thermal stabilization of the capacity in a wide temperature range compatible with X7R behavior [18]. However, more interest has been devoted to the (1-x)(Bi_0.5Na_0.5)TiO₃-xBaTiO₃ system, which endorses an interesting morphotropic phase boundary zone (MPB) where the functional properties seem to be interesting [19,20].

In recent years, revived interest has been devoted to improving BNT’s properties for this important topic [5,6,7,21,22,23,24]. Several substitutions in both the A and/or B sites have been tried, and interesting room energy storage properties were reported for BNT-based systems. However, only a few works have been interested in high-temperature applications [23,24,25,26,27,28]. In our previous work, we showed that doping BNT ceramics with rare earth elements induced a high dielectric stability with low dielectric losses as well as improved energy storage at a high temperature that exceeded 200 °C [27,28].

Presently, the majority of studies have been focused on ceramic capacitors, and few studies have concerned thin films, especially for high-temperature capacitors [29,30].

In a recent study, we demonstrated that BNT-0.06BaTiO₃ (BNT-BT)-based thin films exhibited very interesting energy storage properties at room temperature and strong electric fields that exceeded by several orders of magnitude those obtained for solid-state ceramic capacitors [31].

The present study focuses on investigating the impact of the preparation method of lead-free BNT-BT thin films with a composition near the morphotropic phase boundary (MPB) on the structures, relaxor ferroelectric domain types, and leakage currents, which are key factors determining the ferroelectric and dielectric characteristics of a functional capacitor. Furthermore, our research aims to explore the potential of the studied BNT-BT thin films for achieving high electrostatic energy storage performance with weak electric fields and high temperatures, as well as to evaluate their temperature-dependent dielectric stability. These specific characteristics hold significance for incorporating lead-free films into miniaturized devices where compactness, low energy consumption, and high-temperature stability are key requirements.

2. Materials and Methods

Two methods were used for the preparation of the studied BNT-BT thin films. The first one was a chemical method using Bismuth acetate (III) (C₆H₉BiO₆), Sodium acetate (C₂H₃NaO₂), Barium acetate (Ba(C₂H₃O₂)₂), and titanium isopropoxide (C₁₂H₂₈O₄Ti) as starting precursors to prepare a 0.3 mol/L precursor solution. The first step was to dissolve Bi-, Ba- and Na-acetates in acetic acid with the appropriate stoichiometry. Then, Ti-isopropoxide was dissolved in 2-Methoxyethanol and stabilized with acetylacetone (AcAc). The two prepared solutions were mixed together and stirred at room temperature for 24 h to yield a stable solution. The final chemical solution was deposited layer by layer onto a Pt/SiN substrate using spin coating to yield thin film fabrication. The pyrolysis of each layer was carried out in a hotplate at 400 °C for 10 min. The crystallization of the prepared films was carried out in a tube furnace at 600 °C under an O₂ atmosphere for 30 min. The final thickness was estimated to be around 460 nm.

For the pulsed laser deposition (PLD) method, a 1 mol.% Manganese-enriched BNT-BT target was used. The films were grown at a temperature of 600 °C under an oxygen pressure of 0.3 mbar on a (111)Pt/Ti/SiO₂/Si substrate. A 248 nm KrF excimer laser was used. The laser beam was focused on the BNT-BT target surface at a 45° angle of incidence. The laser’s energy was adjusted in order to produce a fluence of 2 J cm⁻², and the laser’s pulse frequency was fixed at 2 Hz.

The microstructure of the studied specimens was checked using an Environmental Quanta 200 FEG, FEI microscopy (SEM). The structure and the phase purity were investigated via X-ray diffraction (XRD) with ω–2θ scans at room temperature using a four-circle high-resolution Advance D8 diffractometer (Bruker Discover) equipped with a copper anticathode. A double reflection channel-cut Ge monochromator was used to select the Cu Kα₁ radiation (1.5406 Å). Raman spectroscopy was performed in a back-scattering configuration using a micro-Raman Renishaw spectrometer under a green laser excitation of 514.5 nm. The laser power was kept below 20 mW to avoid sample heating. The dielectric measurements were performed as a function of the temperature (from room temperature to 350 °C) and in the frequency range of 100 Hz–1 MHz using a Solartron Impedance analyzer SI-12060 with a probing AC electric field amplitude of 100 mV. The ferroelectric investigations were performed by measuring the polarization–electric field (P-E) hysteresis loops at 10 kHz as a function of the temperature up to 200 °C using a TF Analyzer 3000, aix-ACCT system. A Keithley 2611 A source was used to measure the leakage current properties. All electrical measurements were performed in metal–dielectric–metal geometry using sputtered-Pt circular top electrodes with 250 µm diameters deposited through a shadow mask. The measurements, as a function of the temperature, were performed using a Linkam stage (LTS350, Linkam, (Scientific Instruments Ltd., Redhill, Unit 9, Perrywood Business Park, Honeycrock Lane, Salfords, Redhill, UK, Surrey RH1 5DZ ).

3. Results and Discussion

3.1. Microstructural and Structural Investigations

The microstructure analysis of the studied films is depicted in Figure 1. Both films were dense and crack-free, as can be observed in the SEM images. It seems that the average grain size decreased from 0.6 µm for the film elaborated with the chemical deposition method to about 0.3 µm for the sample grown via PLD. The EDX spectra of both films showed the expected starting elements of Bi, Na, Ba, Ti, and O. It has to be noted that for the PLD film the Manganese element was also detected since a Mn-enriched BNT-BT target was used for PLD film deposition.

The average thicknesses of the films, determined from the cross-sectional scanning electron microscopy measurements, were about 428 nm and 464 nm for PLD and the sol–gel process, respectively.

Figure 2 presents the X-ray diffractograms (XRD) of the studied samples. As reported in our previous work, the film prepared via the chemical method exhibited the coexistence of the rhombohedral R3c and tetragonal P4bm phases, as expected with the bulk BNT-0.06 BT composition in the vicinity of the MPB region. A more detailed discussion can be found in Ref. [23]. However, the film prepared via the PLD method was also polycrystalline with a dominant pseudo-cubic perovskite and a (001) preferential orientation. A small fraction of the pyrochlore Bi₂Ti₂O₇ secondary phase was observed, similar to the work of Yanjiang Xie et al. [32].

Complementary information about the structure of the investigated film was given by a Raman spectroscopy investigation. Figure 3 shows the room-temperature Raman spectra of the two studied BNT-BT thin films. The obtained Raman spectra exhibited relatively broad features, which were comparable to the previously reported Raman studies on BNT-BT bulk materials. This broadening could be attributed to the disorder on the A/B perovskite sites and the overlapping of Raman modes (unpolarized Raman spectra). As shown in the figure, the Raman modes in BNT-BT samples can be divided into four local vibrations involving A-site cation displacements (modes below 100 cm⁻¹), vibrations of both A- and B-site cations (in the range of 100–200 cm⁻¹), off-center shifts of B-site cations (100–200 cm⁻¹), and BO₆ octahedral tilts (400–1000 cm⁻¹) [33].

3.2. Dielectric Investigation and Thermal Permittivity Stability

Figure 4 shows the thermal evolution of the dielectric constant and of the dielectric loss for the studied BNT-BT-based thin films. No clear dielectric anomalies were observed for the investigated films. In contrast, a clear dielectric loss anomaly was observed around 120 °C at 10 kHz for the film prepared via the CSD method, which corresponds exactly to depolarization temperature localization. As shown in Figure 4b, this anomaly was frequency-dependent and asserted the relaxor behavior of this film [31]. Such an anomaly in the dielectric loss was not detected for the film prepared via the PLD method, probably because of the presence of a high leakage current and/or a rather low concentration of polar nanoclusters, which could not give rise to such an anomaly.

It is interesting to note that the non-observation of any dielectric anomaly until 350 °C demonstrated good dielectric temperature stability. The local stress between the substrate and the film can contribute to this behavior, as reported in the literature [34,35].

The thermal stability of the relative permittivity of the investigated samples was evaluated using the temperature coefficient of capacitance (TCC), given by the following equation [23]:

$$TCC=\frac{∆C}{C_{Base Temp}}=\frac{∆{\epsilon }_r}{{\epsilon }_{Base Temp}}=\frac{{\epsilon }_T-{\epsilon }_{Base Temp}}{{\epsilon }_{Base Temp}},$$

(1)

where (ε_T) is the dielectric permittivity at a given temperature and (ε_BaseTemp) is the dielectric permittivity at the base temperature, which is fixed at 150 °C in this work. The value given here for the base temperature agrees with the value used for high-temperature dielectric capacitors and has also been used in several works dealing with BNT-based dielectric materials [36,37].

The thermal evolution of the TCC for both thin films is presented at some working frequencies in Figure 5. The dashed lines denote the operational range of type II capacitors, which is limited to ±15%. As can be seen in the plots, all samples showed good permittivity stability over a large range of temperatures that exceeded 200 °C, which was better than the operating temperature region of X7R, especially for high temperatures. Interestingly, the film prepared via the CSD method showed a better TCC stability that did not exceed a variation of ±15% and had an extended working domain exceeding 300 °C throughout the entire frequency range under investigation. For instance, at 10 kHz, the TCC exhibited excellent stability, with a variation of ±5% up to 250 °C. In addition, the dielectric losses remained below 9% for the entire temperature range, with a maximum obtained at the T_d anomaly.

3.3. Leakage Current, Ferroelectric, and Energy Storage Investigations

The current density (J) vs. the electric field (E) is displayed for both films in Figure 6. The leakage current curves are fairly symmetrical around E = 0 with a high resistivity. However, the film prepared via the CSD method was more resistive than the one prepared via PLD, as it was characterized by lower leakage currents in a strong electric field. The obtained results corroborate the dielectric measurement where the CSD film exhibited very low dielectric loss compared to the PLD film.

Figure 7 shows the temperature dependence of P-E hysteresis loops for the investigated specimens. For the film prepared via the PLD method, clear ferroelectric hysteresis was observed at room temperature, although it was not well saturated because of the weak applied electric field. However, for the film prepared via the chemical method, the obtained P-E loops had a slim shape with paraelectric-like behavior. It is worth noting that the introduction of Ba in the BNT matrix induces different relaxation states ranging from a non-ergodic relaxor (NER) for the mother-phase BNT to a paraelectric-like state. It was reported, however, that for BNT-0.06 BT both the ergodic (ER) and non-ergodic states are present simultaneously because of the competition of the R3c and P4mb nonpolar regions, which is promising for energy storage properties. Based on the shape of the obtained hysteresis loop, we can surmise that for the film prepared via the PLD method both the NER and ER states coexist. Furthermore, the slim hysteresis observed for the CSD film confirms the achievement of the complete ergodic relaxor phase [28].

Surprisingly, the increase in the temperature to 200 °C did not seem to considerably affect the hysteresis loops, as depicted by the temperature dependence of the ferroelectric hysteresis loop in Figure 7. Contrary to normal ferroelectrics, we observed increases in polarization in both films with increasing temperatures up to 200 °C. This could be attributed to the presence of the ergodic and/or non-ergodic relaxor phases. Recall that it is essential to maintain high polarization at high temperatures, as it plays a crucial role in achieving high energy performance in such conditions.

The energy storage properties can be calculated from the ferroelectric hysteresis loops using the integration of the polarization (P) as a function of the electric field (E):

$$W_{rec}={\int }_{Pr}^{Pm}EdP,$$

(2)

where (W_rec) is the recoverable energy density.

It is interesting to mention that to obtain a high energy storage density, both the ΔP = Pm − Pr value and the breakdown electric field should be high.

The energy loss density (W_loss) and the energy efficiency (η) are interesting to evaluate the discharge property of the capacitor and are calculated using the following equations:

$$W_{loss}={\int }_0^{Pmax}EdP-W_{rec},$$

(3)

$$\eta =\frac{W_{rec}}{W_{rec}+W_{loss}}\times 100,$$

(4)

The thermal evolution of the recoverable energy density (W_rec) and energy efficiency (η) are presented in Figure 8 in the temperature range from RT to 250 °C.

As can be concluded from this figure, a quasi-constant evolution of the energy storage density as a function of temperature was observed for both films. For the PLD thin film, W_rec was around 0.8 J cm⁻³ and the maximum efficiency was 55% at RT, which dropped to 38% at a high temperature. Interestingly, the CSD film showed improved energy storage properties, with a W_rec around 1.4 J cm⁻³ and a high efficiency around 76% at RT, which decreased to 40% at 250 °C. Furthermore, at a temperature ≤ 180 °C the efficiency was still very interesting (η > 70%) and remained relatively stable. The obtained values evidence better performances than some lead-free materials reported to be promising for high-temperature applications [22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38]. It is worth mentioning that in harsh conditions, the ability to rapidly store and release electrical charges could lead to stress corrosion cracking. A systematic analysis of this type of corrosion is needed before any application consideration, such as in the case of metal and alloys [39,40] or even some ferroelectric materials [41].

4. Conclusions

In summary, BNT-0.06BT thin films were elaborated using both physical and chemical methods. SEM micrographs provided information about the homogeneity, density, and grain distribution in the samples. X-ray diffractograms of the investigated films confirmed the formation of BNT-BT pseudo-cubic perovskite, and this was corroborated with a Raman investigation. The temperature-dependent dielectric stability was evaluated for the investigated samples using the temperature coefficient of capacitance (TCC). Both films showed good stability of the TCC in a wide temperature range exceeding 200 °C. In particular, the film prepared via the CSD method showed a better TCC stability that did not exceed a variation of ±15%, with an extended working domain reaching 300 °C. These films also exhibited stable energy storage properties over a large temperature range. The film prepared via the chemical method had better performance, with a stable W_rec of 1.4 J cm⁻³ and a high efficiency over a wide temperature range. This interesting result can be attributed to the improvement in the CSD film resistivity, as it was highlighted by the obtained low leakage current density.

The low leakage currents obtained, along with the high energy storage and dielectric stability observed over a broad temperature range in the investigated BNT-BT sample prepared using the CSD method, hold significant potential for the BNT-BT system to replace lead-based materials in miniaturized devices that require low energy consumption and high-temperature stability. Furthermore, these findings pave the way for further investigations into this family of materials.