**3. Results and Discussion**

Figure 1a shows the XRD patterns of Bi1-xSmxFe0.98Mn0.02O3 (x = 0, 0.02, 0.04, 0.06) films deposited on ITO/glass substrate. Figure 1b,c show the local magnified diffraction peaks of the (110) and (202) crystal planes, respectively. From the XRD pattern, the generated sample had a polycrystalline perovskite structure, and the films had good crystallinity, which matches well with the JCPDS (No. 86-1518) PDF standard card. However, the peak-splitting phenomenon of the rhombohedral structure was not observed in the BFSM thin films. At 2θ = 32◦, the BSFM thin film samples do not show (110)/(104) peak splitting, but preferentially grow along (110). This may be due to the structural phase transition in the BSFM films. Figure 1b,c show that with the increase in Sm doping amount, the diffraction peak of (110) and the diffraction peak of (202) gradually shift to a large angle, which may be because the radii of doped Sm3+ (0.96 Å) and Mn2+ (0.67Å) are smaller than that of Bi3+ (1.03 Å) and Fe3+ (0.64 Å) [33,34]. Lattice distortion occurred during doping, and the crystal plane spacing decreased. In addition, with the increase in Sm content, the relative strengths I = I(110)/I(012) of BSFMx = 0.02, BSFMx = 0.04, BSFMx = 0.06 were 2.03, 3.70, 3.02, respectively. When the content was 4 mol%, the relative intensity reached the

maximum value, and the intensity of diffraction peak (202) did not change significantly with doping amount.

**Figure 1.** (**a**) XRD patterns, (**b**) magnified patterns around 30.5–33◦, and (**c**) magnified patterns around 39–41◦ of the BSFM films with different Sm content.

The above data indicate that the crystal had structural distortion. To further study the changes in crystal structure, Rietveld refinement was conducted on all samples, and the results are shown in Figure 2a–d. All samples were used for two-phase refinement using R3c and Pnma cards from the International Crystallography database. The phase content and structural parameters of the samples are shown in Table 1. According to the experimental data, all samples have R3c and Pnma space groups. With the increase in Sm content, the content of R3c phase in the samples was 70.98%, 68.39%, 69.06%, and 67.06%. The addition of Sm reduced the content of R3c phase in the BSFM films. The lone electron pair 6S2 of Bi3+ in BiFeO3 is chemically active, and it is conducive to ferroelectric distortion [20]. The substitution of rare earth element Sm for Bi may reduce the chemical activity of the lone electron pair and reduce the rhombohedral distortion of the crystal. In all Sm-doped samples, the content of the R3c phase in BSFMx = 0.04 was the highest, and the content of the non-polar orthorhombic phase Pnma was the lowest [35,36].


**Table 1.** Rietveld refinement parameters of the BSFM films with different Sm content.

The surface differentiation features of the BSFMx (x = 0–0.06) films are shown in Figure 3a–d. The crystal grain size of the BSFMx (x = 0–0.06) films are shown in the inset. The figure shows that the grain distribution on the surface of the film without Sm doping is not uniform, and that there are many voids, which may be the reason for the volatilization of the organic solution. Furthermore, the surfaces of the Sm-doped films were uniform, compact, and well combined with the substrate, indicating that the annealing mechanism

was very suitable for the growth of the BSFM films on the ITO substrate. The average grain sizes of the BSFMx (x = 0–0.06) films were 62.29, 59.53, 48.93, and 60.98 nm, indicating that Sm doping can reduce grain size. Among them, the BSFMx = 0.04 film had the smallest grain size, suggesting that appropriate Sm doping accelerated the nucleation rate and decreased the grain size [37]. The cross-sectional image of the BSFMx = 0.06 film is shown in Figure 3e. As can be seen from the figure, the film has a clear interface with the substrate, and the cross-section thickness of the film is 585 nm.

**Figure 2.** Rietveld refined XRD patterns of the BSFM films with different Sm content: (**a**) x = 0 mol%; (**b**)x=2 mol%; (**c**)x=4 mol%; (**d**)x=6 mol%.

To further analyze the structure of Sm-doped BSFM thin films, Raman spectroscopy was used. Figure 4a shows the Raman spectra of BSFMx (x = 0−0.06) films with different Sm content in the wave number range of 50 cm<sup>−</sup>1–650 cm−1. The data for the crystal structure of the BSFM films (Pnma+R3c) according to XRD-refined parameters and developed by FullProf software are shown in Figure 4b,c. Figure 4d–g are the Raman spectrum fitting diagrams of each sample. Group theory analysis shows that the vibration modes of the BFO film with the R3c space group with a rhombohedral perovskite structure are ΓRaman,R3c = 4A1 + 9E [38]. Four A1 and nine E vibration modes analyzed by group theory were observed in the Raman spectra of the BSFM films. The Raman vibration modes extracted from the Raman fitting are listed in Table 2. From the Raman fitting diagram, the strength of mode A in the BSFM films significantly increased, which may be related to the change in Bi-O bonds caused by Sm3+ replacing Bi3+. Owing to the Jahn–Teller distortion effect [39], the strength of modes E-8 and E-9 improved. This is because changes in Bi-O bonds lead to the distortion of the ferrite octahedron, which further changes the Fe-O bond. In addition, when the Sm element was added, the A1-1 vibration mode shifted to a higher wave number (from 141.76 cm−<sup>1</sup> to 143.31 cm−1), because the frequency of the Raman vibration mode is related to the functions of ion mass and force [22]. Sm3+ (150.4 g) replaces Bi3+ (209.0 g), which leads to a blue shift in the A1-1 mode. These results show that Sm3+ doping causes lattice distortion, which is consistent with XRD analysis. In addition, all samples had vibration patterns near 200, 300, 400, 490, and 620 cm−1, which

were consistent with the Raman frequencies of the Pnma structure [40]. In Figure 4c–f, the vibration patterns are indicated by #. The results show that in addition to the R3c phase, the Pnma phase also existed in all samples, which was consistent with the XRD refinement results.

**Figure 3.** SEM images of the surface morphological features of the BSFM films: (**a**)x=0 mol%; (**b**)x=2 mol%; (**c**) x = 4 mol%; (**d**) x = 6 mol%, and (**e**) the cross-sectional image of the BSFMx = 0.06 film. The insets show the distribution of crystal grain size and the average grain diameter (Dave).

According to the defect equation (Formula (2)) [13], the Fe2+ content can affect the oxygen vacancy content. Thus, the Fe2p2/3 orbit of the BSFM films with different Sm doping amount was fitted, as shown in Figure 5a–d. The Fe2p2/3 peaks were fitted into two peaks corresponding to Fe2+ and Fe3+. From further calculations, the Fe3+:Fe2+ ratios of the BSFMx (x = 0–0.06) films were 2.33, 2.64, 2.94, and 2.77. The content of Fe2+ in the BSFMx = 0.04 film was the lowest, indicating that an appropriate amount of Sm doping can effectively inhibit the variation of Fe element. Generally, the content of Fe2+ will affect the generation of oxygen vacancies. The lower the content of Fe2+, the fewer oxygen vacancies will be generated, and the fewer defects will exist in the samples.

**Figure 4.** (**a**) Raman spectra of the BSFM films with different Sm content and crystal structure of the BSFM films: (**b**) Pnma; (**c**) R3c, Raman fitting spectra of the BSFM films with different Sm content: (**d**) x = 0 mol%; (**e**) x = 2 mol%; (**f**) x = 4 mol%; (**g**) x = 6 mol%. # refers to the Raman frequencies of the Pnma structure.

To further study the oxygen vacancy content in the BSFM films, the O1s orbital of the films was fitted, as shown in Figure 6a–d. In the figure, O1s is fitted into two peaks, which are lattice oxygen with binding energy of 529 eV (from metal) and oxygen vacancy of 531 eV (from defect) [41,42]. To calculate the oxygen vacancy ratio of the thin films, the two peaks were integrated to calculate the area, and the oxygen vacancy ratios of the four groups of samples were 0.24, 0.16, 0.14, and 0.15. According to the calculation results, the oxygen vacancy content and the Fe2p3/2 orbital Fe2+ content of the BSFMx = 0.04 film were the lowest. This further confirms the view that inhibiting the transformation of Fe3+ to Fe2+ reduces the oxygen vacancy content of the sample.


**Table 2.** The Raman modes of the BSFM films with different Sm content.

**Figure 5.** Fe2p3/2 orbital XPS fitting diagram of the BSFM films with different Sm content: (**a**)x=0 mol%; (**b**)x=2 mol%; (**c**)x=4 mol%; (**d**)x=6 mol%. The black lines refer to the original curve, the red lines refer to the fitting curve, and the blue lines refer to the background line.

Figure 7 shows the leakage current density curve (J–E) of the BSFM film with different Sm content. The test electric field was 350 kV/cm. The asymmetry of leakage current of positive and negative electric fields can be attributed to the different work functions of the upper and lower electrodes. With an increasing doping amount, the leakage current density decreased first and then increased. Under the same electric field, when the doping amount was 4 mol%, the leakage current density reached the minimum value. Under a 350 kV/cm electric field, the leakage current density of the BSFMx = 0.04 film reached 8.84 × <sup>10</sup>−<sup>5</sup> A/cm2. The leakage current density was about 0.5 orders of magnitude lower than that of the pure BiFe0.98Mn0.02O3 film. Under normal circumstances, the leakage current of the sample was affected by oxygen vacancy content, which is mainly related to the high-temperature volatilization of Bi3+ and the variation in Fe3+. It can be seen from XPS that the Fe2+ content and the oxygen vacancy content in the BSFMx = 0.04 film was the lowest, resulting in the minimum leakage current density. In addition, grain boundaries hinder electron migration and reduce leakage current density. According to the SEM results, the BSFMx = 0.04 film has the smallest grain size, indicating that it has more grain boundaries and greater resistance to electron migration [26,42].

**Figure 6.** O1s orbital XPS fitting diagram of the BSFM films with different Sm content: (**a**)x=0 mol%; (**b**) x = 2 mol%; (**c**) x = 4 mol%; (**d**) x = 6 mol%. The black lines refer to the original curve, the red lines refer to the fitting curve, and the blue lines refer to the background line.

**Figure 7.** Leakage current density curves of the BSFM films with different Sm content.

Figure 8a shows the hysteresis loops of the BSFM films with different Sm content with an electric field of 1410 kV/cm and frequency of 1 kHz. The remanent polarization intensities of BSFMx (x = 0−0.06) under the test electric field were 111.23, 83.30, 91.86, and 73.59 μC/cm2. The coercive field Ec was 835.22, 819.28, 830.95, and 987.55 kV/cm, respectively. When the content was less than or equal to 4 mol%, the change in the coercive field was small. The defects inhibit ferroelectric domain flipping. In all the Sm-doped

samples, the remanent polarization value increased first and then decreased with increasing doping content. When the doping content was 4 mol%, the remanent polarization value was the largest, and the ferroelectric property was the best, which was consistent with the XRD analysis results. The possible reasons are as follows: (1) The leakage current density of BSFMx = 0.04 has a minimum value compared with other samples. A smaller leakage current density can improve the voltage resistance of the sample and facilitate electric domain inversion [28]. (2) According to XPS experimental data, the BSFMx = 0.04 sample had the lowest oxygen vacancy content, and the reduction of oxygen vacancy will improve its ferroelectric performance. (3) Owing to its centrosymmetric characteristics, the Pnma phase shows paramagnetic properties without ferroelectric properties, while the R3c phase mainly affects the ferroelectric property of samples [43–45].

**Figure 8.** (**a**) P−E hysteresis loops, (**b**) dielectric permittivity, and (**c**) dielectric loss of the BSFM films with different Sm content.

Figure 8b,c show that the dielectric permittivity (ε) and dielectric loss (tanδ) of the BSFM films with different Sm contents vary with test frequency at room temperature in the range of 1 kHz–1 MHz. From the figure, the dielectric permittivity of the sample has little dependence on frequency, indicating that the sample has high intermediate frequency stability. The results show that the dielectric permittivity values of BSFMx (x = 0−0.06) at 10 kHz were 62, 89, 91, and 65. At the same frequency, the dielectric losses of BSFMx (x = 0−0.06) were 0.034, 0.025, 0.026, and 0.027. In the low frequency range (<40 kHz), Sm doping reduced the dielectric loss of the BSFM films. At the same time, Sm doping increased the dielectric permittivity of the BSFM films, and the maximum dielectric permittivity was obtained when the doping amount was 4 mol%. The existence of oxygen vacancies will distort the free volume used to replace Fe3+ in the Fe-O octahedron, which reduces the dielectric polarization and leads to a smaller dielectric permittivity [46]. As a result, the BSFMx = 0.04 film with the lowest oxygen vacancy content has the maximum dielectric permittivity.

BSFMx = 0.04 and BSFMx = 0.06 films were selected as representatives to explain the aging behavior of samples. Figure 9 shows the electrical hysteresis loop diagram of the BSFM films aged at room temperature for 110 d. The remanent polarization strength (2Pr) of the BSFMx = 0.04 and BSFMx = 0.06 films after aging treatment decreased by 24.9% and 41.3%, respectively, while the intensity of the coercive electric field (2Ec) decreased by 0.6% and 12.2%, respectively. This indicates that the two samples have different degrees of aging. Among them, the aging degree of the BSFMx = 0.04 film was smaller. Ren et al. proposed the symmetric short-range order principle of point defects and inferred that the aging effect could be triggered by ion doping [47–50]. Moreover, the migration of oxygen vacancies led to the formation of complex defect dipoles in the sample. The symmetrical short-range order of oxygen vacancies created conditions suitable for reversible domain switching. In addition, the internal electric field formed during the orderly arrangement of the defective dipoles will increase the offset of the coercive field, which directly explains the asymmetry of the coercive field strength of the sample shown in Figure 9. The domains within ferroelectrics can be switched under an applied electric field, exhibiting macroscopic polarization. However, when the applied electric field is removed, the domain structure gradually shifts to a random state, resulting in the degradation of properties over time, namely, sample aging. The movement of oxygen vacancies in the sample to the domain walls creates pin centers that provide resistance to the movement of the domain walls and reduce the mobility of the domain walls. This, in turn, affects the ferroelectric properties associated with the movement of the domain walls. According to the XPS data, compared with the BSFMx = 0.06 film, the oxygen vacancy content in the BSFMx = 0.04 film was less, which means that the defect dipole content of the sample was smaller. Thus, the resistance of the domain wall to movement was smaller, leading to weak aging effects and better retention of film performance. In addition, it can be seen from SEM images that the BSFMx = 0.04 film had a smaller grain size than the BSFMx = 0.06 film. For small-grain crystals, due to the small difference between lattice symmetry and defect symmetry, the thermodynamic force driving the symmetry matching between them is weak, and this hinders the kinetic migration of oxygen vacancies and affects reversible domain switching [47,51]. Therefore, the BSFMx = 0.04 film with small grains aged slowly.

**Figure 9.** Comparison of hysteresis loops of the BSFM films with different Sm content measured after an interval of 110 days: (**a**)x=4 mol%; (**b**)x=6 mol%.

### **4. Conclusions**

In conclusion, Sm-doped BSFM films were prepared on ITO/glass substrates by the sol-gel method. The effects of different Sm content on the leakage current density, dielectric properties, and aging properties of the BSFM films were systematically studied. Detailed explanations were made in terms of phase transition and oxygen vacancy. XRD and Raman analyses show that the samples all contained R3c and Pnma phases, and the samples with different Sm content had different phase composition. Sm doping led to lattice structure distortion and decreases in the crystal plane spacing. XPS analysis showed that the BSFMx = 0.04 thin film sample had the lowest oxygen vacancy content, indicating that an appropriate amount of Sm doping can effectively inhibit the valence of the Fe element. The decrease in oxygen vacancy increased the dielectric permittivity and the leakage current density. The minimum leakage current density of the BSFMx = 0.04 film sample was 8.84 × <sup>10</sup>−<sup>5</sup> A/cm2 in a 350 kV/cm electric field. The remanent polarization intensity of the BSFMx = 0.04 film was 91.86 μC/cm2, owing to the formation of avnonpolar

orthorhombic Pnma phase during the doping process. At the same time, the aging process of BSFMx = 0.04 sample was slow, and the performance of the sample was better preserved.

**Author Contributions:** Conceptualization, Y.W.; Data curation, Y.W.; Formal Analysis, Y.W.; Funding acquisition, F.Z.; Investigation, Y.W.; Methodology, Y.W., Z.M. and L.W.; Project administration, F.Z. and L.Z.; Resources, X.G. and Z.L.; Software, Y.W.; Supervision, L.Z.; Validation, Y.L., B.Y.; Visualization, Y.W.; Writing-original draft, Y.W.; Writing-review & editing, Y.W., L.W., Z.M. and F.Z. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by funding from the State Key Laboratory of Silicate Materials for Architectures (Wuhan University of Technology) (SYSJJ2020-05), and the Program of the Housing and Urban-Rural Construction Department of Shandong Province (2019-K7-10).

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.
