The XRD patterns of the (1-x)BT-xSFN (x = 0.00, 0.02, 0.05, 0.10, 0.15) ceramics at 2θ of 10°-70° are depicted in Fig. 1(a), corresponding to a polycrystalline perovskite structure. Notice that no impurity peaks (such as Fe2O3 and Nb2O5) are observed. The peaks of the prepared pure BT are indexed as the tetragonal structure of the P4mm space group (PDF#05-0626), which is consistent with the earlier report [28]. It is clearly observed from Fig. 1(b) that the diffraction peak (110) shifts toward higher angle with the addition of SFN at around 31.5° in XRD patterns, which confirms lattice distortion of the doped samples [29, 30]. The reduction of the lattice spacing (d) is calculated by the Bragg diffraction equation [31]: 2dsinθ = nλ (θ: the diffraction peak angle, λ: the wavelength of the X-ray). As illustrated in Fig. 1(c), the prepared pure BT has two diffraction peaks of (002) and (200) at around 45° in XRD. The intensity of peak (002) decrease for the x = 0.02 composition, and then (002) and (200) peaks gradually combine to form a single peak as the x above 0.02, which indicates a transformation of structure from tetragonal to cubic phase. Furthermore, in order to observe the structure more clearly, the lattice parameters of the as-synthesized samples are displayed in Fig. 2. On the whole, the lattice constant c tends to decrease with the x-value increasing, whereas the lattice constant a first increases until x = 0.02, and then becomes consistent with the lattice constant c as the x above 0.05. In addition, the decreased unit cell volumes are observed with the increase of SFN content in Fig. 2, which can be considered as the result of Ba2+ (r = 1.35 Å) being replaced by smaller Sr2+ (r = 1.18 Å). At the same time, the larger Fe3+ (r = 0.65 Å) and Nb5+ (r = 0.64 Å) substitute for Ti4+ (r = 0.61 Å) at the B-site, which is beneficial to reduce the excessive distortion of the lattice.
To further study the structural phase transformation of the synthesized solid solution, the RT Raman spectra of (1-x)BT-xSFN samples have been performed in the scope of 100–1000 cm− 1, as depicted in Fig. 3. All characteristic bands of pure BT are observed in the Raman spectrum. The vibration modes correspond to the A1(TO1) at 262 cm− 1, the E(TO) at 305 cm− 1, the A1(TO2) at 516 cm− 1, and the A1(LO2) at 717 cm− 1, respectively, which can be provided by previous reports [32, 33]. Any transition of the broad peak at 262 cm− 1 indicates the change in position and occupancy of Ti4+. Obviously, the peaks move gradually to lower frequencies and present a broadening with increasing SFN levels, which may be caused by the substitution of larger Fe3+ and Nb5+ ions for Ti4+ ions, resulting in the distortion of local structure at the B-site. The E(TO) mode observed at 305 cm− 1 exhibits that intensity of its sharp peak is weakened. Since the sharp peak characterises the tetragonal phase [33], it is confirmed that the structure of the materials changes from tetragonal phase to cubic phase with increase of SFN content. This is corresponding to the result of XRD measured. Broad character of the A1(TO2) and the A1(LO2) modes named are slightly offset to lower and higher frequencies, respectively, which can be attributed to the vibration of the BO6 octahedron. In addition, two new vibration modes 1 and 2 are observed at 185 cm− 1 and 648 cm− 1. The appearance of mode 1 is considered to be associated with the vibration of A-O, indicating the presence of clusters rich in Ba2+ and/or Sr2+ in the lattice, which can be explained by earlier literature [34]. The presence of mode 2 may be caused by the FeO6 and NbO6 octahedral vibrations due to addition of SFN doping, in which Fe3+/Nb5+ replace Ti4+ at the B-site. Similar results have also been reported earlier [35, 36].
The surface morphology of the prepared ceramics at 1250℃ is characterized by SEM, as presented in Fig. 4(a)-(e). The microstructure of pure BT ceramics displays clearly visible particles and no evident agglomeration of grains, which suggests that the pure BT prepared has favorable crystallization. However, with the increase of doping content, some pores appear among the grains, and the grains are obviously agglomerated. Further, compared with pure BT, the grain size exhibit uneven distribution for x = 0.02, that is, the large and small grains are close to each other, and a trend that the grain size decreases can be detected from the SEM images. However, for the x above 0.02, the grain size distribution is relatively uniform, and the grain size is outstanding different from that of the group divided into 0.00 and 0.02. As a whole, the grain size of SFN-doped BT has been significantly reduced, which reflects that the addition of the dopant restrains the growth of the grains. This may be caused by some stresses during calcination by solid solution reaction, which hinders the movement of the grain boundary and thus reduces the grain size [37, 38]. It is also possible that due to the incorporation of SFN, the mismatch between Fe, Nb and Ti with different ion radii at the B-site leads to the increase of lattice distortion, thus inhibiting the growth of grains [39, 40].
Figure 5(a) depicts the optical absorption spectra of the synthesized materials measured by UV-vis-NIR spectrophotometer. The absorption edge is about 400 nm for pure BT, while the absorption edge gradually widths and transfers to higher wavelength with the increase of doping for doped BT, suggesting that doped BT samples possess greater light absorption scope. It is well known that the variation of the absorption edge in the absorption spectrum is strongly related to the optical band gap (Eg) of the materials. In order to obtain Eg of (1-x)BT-xSFN, the light absorption spectra are derived by the Kubelka-Munk function [19, 41]: F(R) = (1-R)2/(2R) = α/s (R: the relative reflectivity of the materials; α: absorbance; s: scattering coefficient). The Eg are obtained by the tangent line of the Tauc plot, where the expression of Tauc plot is (αhν)n = C(hν -Eg) (h: Planck's constant; ν: frequency; C: proportional constant; n: 2 and 1/2, corresponding direct and indirect band gaps, respectively) [42, 43]. Thus, for materials synthesized on the basis of BT with a direct bandgap, the value of n is 2. Figure 5(b) display the Tauc plots of (1-x)BT-xSFN materials. Through the intersection point of Tauc tangent and the horizontal axis, it can be clearly observed that the Eg of pure BT is 3.21eV, which is consistent with earlier reports [44, 45]. The Eg of the doped BT (x = 0.02, 0.05, 0.10 and 0.15) are 3.15, 3.04, 2.95 and 2.66 eV, respectively. This reveals Eg of (1-x)BT-xSFN significantly lower with the increase of x, but not linearly. It can be divided into three parts: (i) 0 ≤ x < 0.05 (Δ Eg / Δ x = 3.4). (ii) 0.05 ≤ x < 0.10 (Δ Eg / Δ x = 1.8). (iii) 0.10 ≤ x ≤ 0.15 (Δ Eg / Δ x = 5.8), as illustrated in Fig. 5(c).
The above optical behaviors can be interpreted via the doping mediation mechanism. Many reports have proved that in pure BT, the conduction band and valence band are chiefly constituted by Ti-3d orbital and O-2p orbital, respectively. The conduction band and valence band are primarily formed from Fe-3d orbital and O-2p orbital respectively for the pure SFN, which is similar to the orbital arrangement of BaFe0.5Nb0.5O3 (BFNO) [46]. According to the relationship between the electronegativity of ions and the conduction band, the larger the electronegativity of ions, the lower the conduction band in terms of energy. In (1-x)BT-xSFN ceramics, since the electronegativity of the Fe3+ ion is larger than that of the Ti4+ ion, in terms of energy, the Fe-3d orbital is lower than Ti-3d orbital, which makes the conduction band of doped BT materials transfer into the band gap, leading to the narrowing of the band gap. Furthermore, the incorporation of SFN into BT can also be considered to the introduction of Fe3+/Nb5+ ions at the B-site and Sr2+ at the A-site into pure BT. However, the substitution of Ti4+ with Fe3+ and Nb5+ ions of different radius will introduce O vacancy defects and distort the crystal lattice. This directly changes the energy band structure of BT, introducing defects in the band gap, which may be the reason why the edge of the energy band moves into the bandgap. The changes of energy band described above are shown in Fig. 5(d). The results indicate that the Eg of BT can be adjusted in a small range by controlling the optimized doping amount of SFN and the absorption properties can be improved in the visible range.
To better understand the ferroelectric properties of (1-x)BT-xSFN with ions co-doping at the Ti-site, Fig. 6(a)-(e) present the RT polarization-electric field (P-E) hysteresis loops of all ceramics at frequency 1 kHz, where the applied electric field of the hysteresis loops of each sample corresponds to 10, 15 and 20 kV/cm, respectively. It is clearly observed that, for each sample, when the electric field changes from 10 to 20 kV/cm, the bigger and more saturated P-E hysteresis loops appear. The above phenomena may be caused by the enhancing of the stability of the ordered ferroelectric domains as the electric field increases [39]. For pure BT, roughly the standard hysteresis loops can be observed, which demonstrate the ferroelectric properties of ceramics [47, 48]. However, it is obvious that the hysteresis loops tend to be flatter and slimmer with the addition of higher doping. In the doped BT, when x = 0.05, the value of remnant polarization (Pr) is higher than that of x = 0.02, 0.10 and 0.15, indicating that the doped BT at this concentration has better ferroelectric properties. With the addition of SFN, the weakening of ferroelectric polarization is noticed, which may be due to the substitution of Fe3+ ions for Ti4+ ions at the B-site, introducing more oxygen vacancy defects and thus hindering the movement of ferroelectric domains, resulting in the decrease of Pr [49]. As can be seen from Fig. 6, there is a large opening, which may be due to the leakage current caused by O vacancies introduced into the samples during the calcination process. Therefore, from the analysis results, it is believed that enhancing the quality of as-prepared ceramics can reduce leakage current, thus improving the ferroelectric properties of the prepared samples.