With the continuous consumption of non-renewable energy, renewable energy is constantly being explored by researchers. Solar energy, wind energy, tidal energy, etc., as valuable energy provided by nature to humans, are kinds of intermittent energy. To address this problem, energy storage devices have also been developed, such as lithium-ion batteries, sodium-ion batteries, aluminum-ion batteries, and supercapacitors (Kundu et al., 2015; Sun et al., 2015; Nayak et al., 2017; Ho and Lin, 2019). With the continuous development of science and technology, the requirements of 3C (Computers, Communication, and Consumer electronics) products for batteries are gradually increasing. Simultaneously, the energy storage equipment is required to have excellent performance, such as high energy density, small volume, and fast charge-discharge rates (Choi and Aurbach, 2016; Zhang et al., 2019).

Perovskite materials with a general formula of ABO₃ are used in energy storage materials such as solar cells, lithium ion batteries, fuel cells, and supercapacitors (Mefford et al., 2014; Saliba et al., 2016; Zhang et al., 2016; Zhou et al., 2016). In the previous work, the B site was widely studied as an active site. However, there is a few electrochemical processes that are linked with the ions from the A site. In recent work, we have found that the A-site with the largest atomic radius of the perovskite BFO can also act as an electrochemically active site (Ma et al., 2019). The results indicate that A site with heterogeneous ion doping would be a new perspective for improving the electrochemical ability in energy storage.

The inherent conductivity of transition metal oxides is low, which restricts the rapid charge—discharge process for high-performance electrode materials (Xia and Alshareef, 2015; Lukatskaya et al., 2017). In order to solve this problem, various strategies have been proposed to improve conductivity of materials, such as using a conductive carbon-based composites (Wang et al., 2016a,b). The synergistic effects of hybrid materials are widely used to improve conductivity and electrochemical property of a single-phase material. In order to essentially enhance the conductivity of the material, a strategy of heterogeneous ion doping is proposed to increase the material conductivity. For instance, Park et al. reported that the electrical conductivity of the perovskite oxide Pr_1−xSr_xCoO_3−δ for intermediate-temperature solid oxide fuel cells was enlarged with increasing strontium content (0.1 < x < 0.3). When the substitution amount x was >0.3, the conductivity of the material decreased. Thus, substitution of appropriate heterogeneous ions can improve the material conductivity. The electrochemical performance of Pr_1−xSr_xCoO_3−δ//Ni-GDC cell demonstrates that the maximum power density at 973 K was 1.19 W cm⁻² with x = 0.3 (Park et al., 2012). Li et al. found that a niobium and tantalum co-substituted perovskite SrCo_0.8Nb_0.1Ta_0.1O_3−δ exhibits high electroactivity. A symmetrical fuel cell consisting of SrCo_0.8Nb_0.1Ta_0.1O_3−δ positive electrode and Gd_0.1Ce_0.9O_1.95 negative electrode has peak power densities of 1.2 Wcm⁻² and 0.7 Wcm⁻² at 500 and 450°C, respectively. The high performance is attributed to an optimal balance of oxygen vacancies (Li et al., 2017). The electrical conductivity of La_0.5Ca_0.5CoO₃ was significantly increased and the specific capacitance reached 170 F g⁻¹ at current density of 1 A g⁻¹ (Mo et al., 2018).

As a typical ABO₃ perovskite structure, bismuth ferrite also faces the problem of unexpected conductivity. The potential strategy is proposed to raise the intrinsic conductivity of BFO by ion doping. The introduction of oxygen vacancies can effectively improve material electronic transport capabilities. In this article, this strategy is used to enhance the conductivity of BFO through the doping of Ca²⁺ ions. By comparing the electrochemical performance of different Ca²⁺ ion doping amounts, it is found that 10% Ca²⁺ doping (BFO-10%Ca) can offer 305.5 mAh g⁻¹ at 1 A g⁻¹. The assembled BFO-10%Ca//graphene hybrid electrochemical energy storage devices exhibit excellent cycle stability at 5 A g⁻¹ after 3,000 cycles.

A single-phase perovskite structure of BiFeO₃ was synthesized by sol-gel method. As presented in Figures 1A,C, the XRD patterns of BFO are well consistent with the standard value reported in JCPDS card (no. 86-1518), which reveals the presence of a rhombohedral crystal structure (R3c space group). After calcium ions doping into BiFeO₃, the single-phase BiFeO₃ (R3c space group) gradually transforms into a new phase (P4mm space group), which is mainly attributed to the change of Bi-O bond and the Ca-O bond after the Bi³⁺ ions were partially replaced by the Ca²⁺ ions (Figure 1B). Ca²⁺ ion-doped BFO crystal structure introduces oxygen vacancies due to the substitution of divalent Ca²⁺ ions for trivalent Bi³⁺ ions. At the same time, the introduction of oxygen vacancies can improve the conductivity of the material. The XRD results of BFO and different Ca²⁺ ion doped BFO samples are shown in Figure 1C. The XRD of BFO is in good agreement with the standard spectrum JCPDS card (no. 86-1518), while the XRD patterns of three samples, BFO-5%Ca, BFO-10%Ca, and BFO-20%Ca, exhibit little difference in position and intensity from the shape. The partially magnified patterns at 2 theta (~32°) show that the diffraction peak shifted toward a higher angle with the increase of the doping amount of Ca²⁺ ions (Figure 1D). The lattice parameters (a and c) of the R3c phase decrease with increasing doping level (Figure 1D), which causes the structure of BFO distorts or changes from rhombohedral to tetragonal phase.

In order to compare the differences in the morphology of pure BFO and Ca²⁺ ion doped BFO samples, SEM was used to characterize the morphology of the particles. Figure 2a shows the SEM image of pure BFO. It can be seen that the size of BFO particles is different, and mainly distributed in the range of ~400–700 nm. As displayed in Figures 2b–d, the particle size of three samples (BFO-5%Ca, BFO-10%Ca, and BFO-20%Ca) is significantly smaller than that of the pure BFO, which exhibits a typical feature of nanoparticle cluster. The introduction of Ca²⁺ ions into BFO has a great influence on the size of the particles due to the smaller crystal unit cell volume of the tetragonal phase. The SEM images exhibit that the surface of the BFO sample after Ca²⁺ doping is smoother than that of pure BFO. This can be attributed to small nucleation cores aggregating together into big grains (Costa et al., 2014).

HRTEM was utilized to characterize the microstructure of the neat BFO, BFO-5%Ca, BFO-10%Ca, and BFO-20%Ca samples, as shown in Figures 3a–d. In particular, the spacing of the adjacent lattice fringes in the neat BFO is about 0.284 and 0.399 nm (Figure 3a), in accordance with the (104) and (012) planes of pure BFO (Li et al., 2016). The TEM images of high-resolution images along [110] zone axis are shown to confirm small changes in the crystal structure (Figures 3e–h) (Gao et al., 2018). As shown in the inset, the coexistence of rhombohedral and tetragonal phases in the calcium ion doped BFO samples are also found by the fast Fourier transform (FFT). The results of selected area electron diffraction (SAED) pattern show good single-crystal characteristics for pure BFO. The presence of 1/2{ooo} superlattice spots confirms the rhombohedral phase BFO using SAED pattern along the [110] zone axis direction (Figure 3i) (Cheng et al., 2009; Liu et al., 2018; Wang et al., 2018). This corresponds to the standard XRD card and demonstrates that a pure phase BFO sample can be obtained by sol-gel method (Figure 1C). Figures 3j–l shows that the new tetragonal phase is found in the BFO-5%Ca, BFO-10%Ca, and BFO-20%Ca three samples. 1/2{ooo} superlattice spots disappear gradually with increasing amount of calcium ions. This is attributed to the changes of value of c/a, when the Bi³⁺ ions are partially replaced by Ca²⁺ ions. The spacing of the adjacent lattice fringes is 0.279, 0.284, 0.296, and 0.396 nm in the BFO-5%Ca BFO-10%Ca and BFO-20%Ca responds to the [110], (104), and (012) crystal plane of rhombohedral BFO, [110] crystal plane of tetragonal phase BFO, respectively (Zeches et al., 2009; Han et al., 2016; Li et al., 2016). The tilt of the same plane for the two phases is about 3.8°, in agreement with the existence of two phases (Figure 3l) (Cheng et al., 2009; Guo et al., 2017).

The chemical bonds of the elements can be monitored in BFO samples and three different Ca²⁺ ion doped BFO samples were used to track the valence state of each element by XPS spectrum analysis, as displayed in Figure 4. The wide XPS survey spectrum of the BFO samples with different calcium ion doping amounts is shown in Figure S1. As illustrated, the narrow-scan Bi4f spectrum of the neat BFO reveals two peaks located at 158.7 and 164.0 eV (Figure 4A), identified as Bi4f 7/2 and Bi4f 5/2, respectively. As reported in previous literature, this can be identified as Bi³⁺ (Bai et al., 2019). The bonding energy of the element Bi4f was shifted to the lower angle with increasing the amount of doped Ca²⁺ ions, because the introduction of Ca²⁺ ions changes the chemical environment around the Bi³⁺ ions. At the same time, this further proves that the change of Ca-O bond and Bi-O bond length formed by Ca²⁺ ions (radius: 1.12 Å) partially instead of Bi³⁺ ions (radius: 1.17 Å) causes structural change. For the Ca2p spectrum, the main Ca2p peaks were situated at 346.3 and 349.8eV (Figure 4B), corresponding to Ca2p3/2 and Ca2p1/2, respectively. This is one typical characteristic and demonstrate that Ca²⁺ ions were doped into BFO samples (Zhang et al., 2018). As an approximately quantitative analysis, the increased integrated area of the calcium 2p peak indicates that the content of calcium ions in the doped BFO sample gradually increases (Zhao et al., 2007; Wang et al., 2019). The difference electronegativity values of Ca and Bi elements results in slight shift of Bi4f peaks toward the lower binding energy side with increasing Ca²⁺ content in BFO samples. In this case, the Fe2p spectrum also shows different valence changes due to the surrounding environment changing. In the track of elemental Fe, two typical Fe2p 1/2 and Fe2p 3/2 peaks could be observed, at 724.6 and 710.8 eV, respectively. The Fe2p spectrum reveals one satellite peaks (“sat.”) at 718 eV, which is the characteristic peak of Fe³⁺, as reported in the literature (Ma et al., 2013). The Fe2p1/2 and Fe2p3/2 peaks split into two peaks with the introduction of Ca²⁺ ions. At the same time, the peaks appeared at 725.6, 723.4, 712.0, and 710.0 eV, which also indicates that Fe³⁺ and Fe²⁺ ions coexist in calcium-doped BFO samples (Ye et al., 2017; Spivakov et al., 2019). The formation of tetragonal BFO is accompanied by transformation of the FeO₆ octahedra into FeO₅ pyramids (Rossell et al., 2012). Thus, divalent and three irons coexist in three samples doped with Ca²⁺ ions. The high-resolution O1s spectrum can feed back the information including oxygen-metal bonding and oxygen vacancies. O1s spectra can be separated into three peaks three peaks at 529.3, 530.6, and 532.6 eV, which correspond to Fe-O, Bi-O bonds, and the surface-adsorbed water (Soltani and Lee, 2016; Gu et al., 2019). The bond energy was shifted from 530.6 to 530.2 eV, which indicates the presence of oxygen vacancies with different doping amounts of Ca²⁺ ions (Ibrahim et al., 2007; Jaafar et al., 2019). The formation of these oxygen vacancy defects may be attributed the Ca²⁺ ion substitution at Bi³⁺ ions of BFO to maintain charge neutrality (Ramachandran et al., 2012).

A three-electrode system was used to accurately characterize the electrochemical properties of the materials. Figure 5A displays the CV of BFO at various scan rates. A pair of obvious redox peaks exhibited in the CV diagram, which is mainly attributed to the redox reaction of Bi³⁺ ions located at the A site of the perovskite structure at different potentials. Due to the unexpected conductivity of pure BFO, the BFO (the inside electron conductance) does not match well with the external ions as the current increased, which will cause the distortion of the CV curves (Figure 5A). In order to improve the conductivity of the BFO material, calcium ions partially replace the Bi³⁺ ions to form more oxygen vacancies to improve the conductivity of the neat material (Mo et al., 2018). By measuring the electrochemical properties of different Ca²⁺ doping amounts, a suitable degree of substitution is finally selected. By comparing the CV plots of the different amounts of Ca²⁺ ion doping, it can be found that the degree of distortion in the CV curves is gradually reduced at the scanning rate 2, 5, 10, 20, 40, and 60 mV s⁻¹ and finally the 10% calcium-doped BFO is determined to be a suitable degree of substitution (Figures 5B–D). This also shows that doping of heterogeneous ions is an effective way to improve the electrical conductivity of materials.

Testing of the electrode material was completed through a three-electrode test system. Figure 6A shows a neat BFO charge—discharge curves. The charge—discharge test line does not exhibit good symmetry due to the poor conductivity of the material. However, the charge—discharge curves of Ca²⁺ ion doped samples with different doping levels are almost symmetrical, indicating that these materials have good reversible capacity and excellent coulombic efficiency (Figures 6B–D). The charge-discharge diagrams of pure BFO and calcium-doped BFO samples can also correspond to the cyclic voltammogram obtained at different scan rates (Figure 5). The specific capacity calculated at different current densities is shown in Figure 6E. Comparing different doping amounts of Ca²⁺ ions, 10% of the samples (BFO-10%Ca) shows the best specific capacity 305.5, 182.0, 102.1, 76.2, and 74.8 mAh g⁻¹ at diverse discharge current densities of 1, 3, 10, 20, and 30 A g⁻¹, respectively. The performance improvement of BFO-10%Ca is mainly attributed to the better electrical conductivity with appropriate increasing calcium content and on the other hand the improved performance of calcium-modified BFO is higher than the capacity loss in the samples substituted by Ca²⁺ ions. This energy storage mechanism of BFO with different doping amount of Ca²⁺ ions can be expressed by the following equation (Ma et al., 2019):

The electrochemical impedance spectroscopy (EIS) spectra of pristine BFO, BFO-5%Ca, BFO-10%Ca, and BFO-20%Ca are shown in Figure 6F. In contrast to the neat BFO sample, three samples of Ca²⁺ ion doped electrodes shows smaller semicircle diameter, manifesting a lower electron transfer resistance at the electrode interface and higher electrochemical conductivity after the introduction of Ca²⁺ ions. Noticeably, the results of the ohmic impedance at higher frequency suggest that the values were measured to be 0.95, 1.02, and 1.06 Ohm in the BFO-5%Ca, BFO-10%Ca, and BFO-20%Ca, respectively. These values are apparently lower than that of the pristine BFO (1.17 Ohm), which also indicates that the doping strategy could reduce the ohmic impedance, favorable for boosting the charge and discharge capabilities.

Calcium ions were doped at the A site in the BFO samples. The trivalent Bi³⁺ ions were replaced by the bivalent Ca²⁺ ions, which will cause the uneven charge distribution. Such doping effect would compensate Ca²⁺ ion acceptors in terms of generating oxygen vacancies as donor impurities (Masó and West, 2012). This conductor-insulator transition can be accepted in terms of the competition between a mobility edge and the Fermi energy through band-filling control. In a proper amount of calcium ion-doped BFO samples, the conductivity of the doped samples is enhanced by band filling (Park et al., 2012; Yang et al., 2012).

Figure 7 shows the electrochemical performance test using the three electrodes and two electrodes. A HEESD consists of BFO-10%Ca as the positive electrode and commercial graphene as the negative electrode. The cyclic voltammogram is obtained through a three-electrode system, which shows a rectangle with increasing scan rate. In order to characterize the electrochemical storage capacity of the graphene, a galvanostatic charge-discharge test was performed through a three-electrode system with graphene as the working electrode. This figure shows an isosceles triangle shape, indicating that graphene stores charges by means of an electric double layer (Figure 7B). The specific capacitances were 99.2, 74.6, 68.4, 60.8, 55.8, and 52 mAh g⁻¹ at 1, 3, 5,10, 15, and 20 A g⁻¹ (Figure 7C), respectively.

The HEESDs were fabricated using the BFO-10%Ca as the positive electrode and the graphene as the negative electrode. Electrochemical performance testing of the HEESDs was achieved using a two-electrode test system. The voltage range of 0–1.4 V was selected in accordance with the absence of polarization in the 6 mol L⁻¹ KOH aqueous solution by test results of CV plots at various potential windows (Figure S2). The CV diagram shows two typical electrochemical storage mechanisms, that is, the electric double layer characteristics in the 0–0.5 V voltage range and the Faraday reaction in the 0.5–1.4 V range. The discharge curve of the HEESDs is almost symmetrical with the charging curve, showing an isosceles triangle, as exhibited in Figure 7E. The specific capacity at different current densities can be obtained by calculation of the discharge curve. The specific capacitances were calculated to be 37.7 and 21.3 mAh g⁻¹ at 0.5 and 20 A g⁻¹ (Figure 7F), respectively. Cyclic performance, as an important performance indicator, was performed through 3,000 consecutive charge—discharge tests. The assembled HEESDs exhibited excellent cycle stability at 5 Ag⁻¹ after 3,000 cycles. Compared to the capacity after 3,000 cycles, very limited capacity loss was observed, showing that the assembled HEESDs presented good reversible capacity (Figure 7G). The capacity fading is related with carrier concentration loss and internal impedance growth in the HEESDs. This is mainly due to the charge transfer and diffusion resistances, caused by the formation of new phases after repeated charge and discharge cycles of BFO-10% Ca. It can be also proved by comparing the change of EIS diagrams before and after the different cycle numbers (Schmitt et al., 2017; Upadhyay et al., 2017; Yu et al., 2018). The EIS spectra of the HEESDs were shown in Figure 7H. The Nyquist diagram clearly includes the arc in the high frequency region and the straight line in the low frequency region. After 3,000 cycles, the semi-circular arc is larger in the high frequency region, indicating that the charge transfer resistance was significantly larger than that of the first cycle (Xu et al., 2017), which is in a good response to the decay of performance after multiple cycles.

In summary, the electrical conductivity is enhanced by introducing oxygen vacancies via partial replacement of Bi³⁺ ions by Ca²⁺ ions in the BFO perovskite materials. By tuning the doping amount, the electrochemical performance of BFO-10%Ca was significantly improved. The HEESDs were fabricated using the BFO-10%Ca as the positive electrode and the commercial graphene as the negative electrode, showing excellent cycle stability at 5 Ag⁻¹ after 3,000 cycles. The strategy by improving the intrinsic conductivity of materials by doping with heterogeneous ions is an effective method for improving the electrochemical performance of energy storage devices.

All datasets generated for this study are included in the article/Supplementary Material.

NL conceived and supervised the project. C-JM performed the experiment, characterized the materials, and wrote the manuscript supervised. All the authors discussed the results of the paper. W-LS helps correct grammatical errors.