Optical and magnetic properties of Sm-doped BiFeO₃ nanoparticles around the morphotropic phase boundary region

Multiferroics, which simultaneously possess ferromagnetic, ferroelectric, and ferroelastic ordering, have triggered enormous studies among various functional materials. The magnetoelectric (ME) effect in these materials, namely, the coupling of electric and magnetic ordering, can provide another degree of freedom in designing new functional devices. BiFeO₃ (BFO), with its Néel temperature (T_N, ∼643 K) and Curie temperature (T_C, ∼1103 K) far above room temperature, is one of the most highly promising single-phase multiferroic materials.¹ 

Element substitution for A sites of the ABO₃ structured perovskites has been proven to be a simple and efficient method for improving ferroelectric and piezoelectric properties of ABO₃, especially with regard to phase transition located around the morphotropic phase boundary (MPB) region.^2–4 The morphotropic phase boundary that was initially proposed in Pb(Zr,Ti)O₃ materials denoted a special boundary type with the coexistence of a ferroelectric rhombohedral phase and a tetragonal phase.⁵ Morphotropic phase boundaries (MPBs), which refer to the phase transition between the tetragonal and the rhombohedral ferroelectric phases, are mostly focused on the Pb(Zr,Ti)O₃ system; however, in the BiFeO₃ system, the morphotropic phase boundaries are located at the boundaries of the rhombohedral phase and the orthorhombic phase, as reported before.^6–8 Due to the similarities in their corresponding Gibbs free energies, the two individual phases could be switched to the other phase through a bridging phase under an external electric field, consequently harvesting a strong electromechanical coupling.⁸ In BiFeO₃ (BFO) based materials, the MPB is expected to couple with other parameters such as ferromagnetism. In our previous work, we prepared Sm³⁺ and Ti⁴⁺ co-doped BFO ceramics and found that the composition-driven phase transition from R3c to Pnma was simultaneously accompanied by ferroelectric–paraelectric transition as well as the evolution of ferromagnetism. It is also proved that the Sm element substitution in BFO films induced the structural translation to introduce the morphotropic phase boundary.^9,10 A lot of systematic work has been carried out on Sm-doped BFO. Ben et al. focused on the bulk dielectric properties of Sm-doped BFO,¹¹ and the visible light photocatalytic performance of Sm-doped BFO nanoparticles was limited to a substitution range between 0.01 and 0.1, which were studied as perovskite materials.^12,13 Even though phase transitions in this material system have been observed before, here, we deliberately introduced morphotropic phase boundaries into the BFO system and studied the optical and magnetic properties near the special phase diagram locations. Most studies are focused on the morphotropic phase boundary in BFO based ceramics and films;^14–16 however, there are few studies on the morphotropic phase boundary in nanoparticle systems. BFO nanoparticles have special applications such as the production of hydrogen under photocatalytic water splitting¹⁷ and degradation of organic pollutants under visible-light exposure^18,19 owing to their adjustable optical energy gap (2.2–2.8 eV).²⁰ The bandgap of BFO comes from the energy separation of the top of the Fe 3d–O 2p mixed-valance band and the bottom of the Fe 3d conduction band, which is related to the Fe–O octahedral structure. By constructing the morphotropic phase boundary, the modulation of the Fe–O octahedral structure leads to the evolution of bandgap and related properties. However, the morphotropic phase boundary with a special phase region combined with the size effect of nanoparticles and the related optical and ferromagnetic properties have not been systematically studied.

In this article, we investigated the MPB and the associated physical properties of Sm-doped BFO (Bi_1−xSm_xFeO₃, x = 0.05, 0.10, 0.15, 0.20) nanoparticles prepared using the sol–gel method. In addition to the improved photodegradation efficiency and ferromagnetism of BFO nanoparticles, the MPB has unique properties including the exchange bias effect and double magnetic hysteresis behavior.

Bi_1−xSm_xFeO₃ (x = 0.0, 0.05, 0.1, 0.15, and 0.2) nanoparticles, abbreviated as BiFeO₃ (BFO), Bi_0.95Sm_0.05FeO₃ (B5SFO), Bi_0.9Sm_0.1FeO₃ (B10SFO), Bi_0.85Sm_0.15FeO₃ (B15SFO), and Bi_0.80Sm_0.20FeO₃ (B20SFO), were fabricated using double-solvent sol–gel technology. Bi(NO₃)₃·5H₂O and Sm(NO₃)₃·6H₂O were dissolved in acetic acid (C₂H₄O₂) and ethylene glycol (C₂H₆O₂) and stirred for 90 min. Fe(NO₃)₃·9H₂O and tin (Sn) powders were dissolved in acetic acid separately. Both solutions were stirred for 3 h to obtain a colloidal solution or sol and then kept at 80 °C for 12 h to form dried gel powders. 3% excess of bismuth was added to compensate for bismuth loss. The dried gel powders were calcined in a furnace at 600 °C for 3 h.

The crystalline structure of these nanoparticles was characterized by x-ray diffraction (XRD) using a Bruker D8 Advance x-ray diffractometer with Cu Kα radiation (λ = 1.54 Å). The morphology of these nanoparticles was observed using scanning electron microscopy (SEM). Raman scattering spectroscopy (JY-HR800) was carried out using an Ar⁺ laser (514.5 nm) as the excitation line. The bandgap and UV–visible-light catalysis were determined using a UV–visible spectrophotometer (Shimadzu UV-3600 PLUS). DC magnetic hysteresis loops were measured using a superconducting quantum interference device (SQUID) magnetometer using the magnetic field ranging from 20 000 to −20 000 Oe at 300 K.

A 300 W Xe lamp was used as the simulated sun-light source, and a filter was added to obtain visible light. During the photodegradation reaction, water circulation was applied to control the reaction temperature to about 35 °C. Typically, 0.1 g of the as-prepared product was added to 100 ml of Methylene Orange (MEO) aqueous solution (10⁻⁵ mol l⁻¹). Before irradiation, the suspensions were magnetically stirred in the dark for 30 min to ensure the establishment of solid–liquid adsorption–desorption equilibrium between the photocatalyst and the MEO pollutant. 4 ml of Methylene Orange (MEO) aqueous solution was taken as the sample. The suspensions were placed under a light source, and their temperature was constantly maintained at 35 °C for 2 h. The samples were taken out every 30 min and were wrapped by foil for high-speed centrifugation, and the absorption value of the solution after centrifugation was measured by a Shimadzu UV-3600 PLUS spectrophotometer.

The XRD patterns of BFO, B5SFO, B10SFO, B15SFO, and B20SFO nanoparticles are shown in Fig. 1, indicating that most of the detected peaks are associated with the parent rhombohedral phase without any impurity or secondary phase. With increasing Sm content, two characteristic peaks at (104) and (110) are first clearly split in the BFO sample and then tend to merge in the B5SFO sample. For the B10SFO sample, the characteristic peaks of (104) and (110) are completely merged into a single peak. The characteristic peaks of (020) and (112) with an orthorhombic phase Pnma appear in the B15SFO sample and are intensified in the B20SFO sample. Similar phenomena have been reported previously.^7,8,21,22 These results imply that Sm substitution may enter the lattice and lead to the structure and phase transformation of BFO. For the BFO and B5SFO samples, their structure belongs to the orthorhombic R3c phase. An orthorhombic interim phase (Pna2₁), marked as the bridging phase,^6,7,11 is observed in B10SFO and B15SFO samples. The representative peaks located at (131), (111), and (212) belonging to the orthoferrite Pbnm phase^7,21,23,24 are intensified, and they evolved into a dominant phase in B15SFO and B20SFO samples. These results demonstrate that the B10SFO sample has coexisting phases consisting of a rhombohedral phase, orthorhombic interim phase (Pna2₁), and orthorhombic phase, known as the morphotropic phase boundary, which is consistent with the characteristics of its respective XRD peaks and the following Raman measurement.

Raman scattering spectra are used for detecting atomic displacements and structural translations. The Raman spectra of BFO, B5SFO, B10SFO, B15SFO, and B20SFO nanoparticles are described in Fig. 2. The BFO sample is reported to have 13 Raman active modes (4A₁ + 9E) corresponding to the rhombohedral R3c perovskite structure. The modes of A₁-1, A₁-2, and A₁-3 have strong scattering intensities located at 136, 168, and 211 cm⁻¹, respectively, which are related to the Bi–O covalent bonds. Six E modes are detected with medium scattering intensities at 275, 335, 365, 456, 549, and 597 cm⁻¹, which is consistent with the previous report.²⁵ With the increase in Sm substitution, a new mode at 310 cm⁻¹ gradually becomes pronounced, originating from A-site vibrations and the oxygen tilting of the Pbnm orthorhombic structure.²⁶ The scattering spectrum of the B5SFO sample is consistent with the rhombohedral R3c perovskite structure, while the B15SFO sample is mostly composed of the Pbnm orthorhombic structure. The coexisting phases of the B10SFO sample mean that the morphotropic phase boundary may be formed at this composition, which is consistent with the results of the XRD analysis. Similar phenomena have also been observed in other RE doped BFO nanoparticles and ceramics with the MPB.⁷ 

The surface morphology of BFO, B5SFO, B10SFO, B15SFO, and B20SFO nanoparticles is illustrated in Fig. 3. The BFO sample exhibits a well-defined cubic structure, with an average particle size of about 200 nm. However, Sm substitution clearly causes a decrease in the average particle size of these samples. Another characteristic is that particle agglomeration is particularly apparent in the B20SFO sample. As shown in Fig. 3, energy dispersive spectroscopy (EDS) is used to investigate the element distribution of Sm in the bulk sample. Some points are taken at the boundary and the center of a grain to analyze the element distribution. The EDS diagram of one point of the B10SFO sample is not shown in the paper, but a comprehensive comparison table is made as shown in Table I. The EDS mapping of the B10SFO sample is shown in Fig. 4. These results show that the percentage of the Bi atom on the boundary is 15.07% and 13.15% and the percentage of the Bi atom in the central region is 9.36% and 12.35%; the percentage of the Sm atom on the boundary is 3.53% and 2.85%, and the percentage of the Sm atom in the central region is 1.37% and 2.83%. EDS mapping results and specific distribution data together illustrate that Bi and Sm are slightly rich ini and Sm are slightly the boundary.

The UV–vis spectra of BFO, B5SFO, B15SFO, B10SFO, and B20SFO nanoparticles are depicted in Fig. 5(a). It indicates that the optical spectrum of the rhombohedral structure differs from that of the structure in BFO based nanoparticles.²⁷ The BFO nanoparticles reveal an absorption band edge at 603 nm, which is similar to previous reports.^19,20,28 The absorption peaks of B5SFO, B15SFO, B10SFO, and B20SFO are located at 599, 636, 631, and 762 nm, respectively. These results prove that the prepared samples can absorb visible light, which can be used as visible-light photocatalysts. As can be seen from Fig. 5(b), the linear extrapolation of (αhν)² to zero of the Tauc plot is used to determine the optical bandgap of BFO, B5SFO, B15SFO, B10SFO, and B20SFO around 2.14, 2.16, 2.09, 2.08, and 1.81 eV, respectively. In the BFO based materials, the bandgap represents the energy separation between the top of the O 2p–Fe 3d mixed-valence band and the bottom of the Fe 3d conduction band²⁹ although theoretical studies, especially the results of first-principles calculations, show that direct bandgaps and indirect bandgaps are different because of the calculation methods.² The experimental and theoretical results are in good agreement that BFO is a direct bandgap semiconductor with a bandgap width of 2.1 eV (with the gap between the top of the valence band near L and the bottom of the conduction band along L).¹⁰ The Sm substitution occupies the lattice location of Bi and impacts both the valance and conduction band edges, thus decreasing the bandgap of Sm-doped BFO samples.^30,31

Figure 5(c) shows the photodegradation efficiency of Methylene Orange (MEO) as a function of illumination time with different photocatalysts. All catalyst reactions are carried out under the condition of the same catalyst mass (0.1 g). In order to reach the adsorption–desorption equilibrium between the photocatalyst and MEO, the reactants were magnetically stirred in the dark for 2 h before the light-catalyzed reaction. The photocatalytic degradation of MEO reaches ∼72%, 80%, 72%, 16%, and 14% for the BFO, B5SFO, B10SFO, B15SFO, and B20SFO samples, respectively, indicating the B5SFO nanoparticles as the most effective photocatalyst. With the increase in Sm substitution, the photocatalytic efficiency first increases and then decreases. The related mechanism is illustrated as follows: Both particle size and specific surface area affect the photocatalytic properties. From Fig. 3, it can be seen that the B10SFO sample has an average grain size of ∼100 nm and its dispersibility is the best among all the Sm-doped nanoparticles, which provides a large surface area with more activity sites to improve the photocatalytic efficiency. Under the illumination of visible light, electrons are excited from the valence band into the conduction band, producing electron–hole pairs. These electron–hole pairs can transfer to the surface of the photocatalyst and then act as a redox source, which accelerates the reaction with adsorbed organic reactants. Finally, decomposition of the organic reactants occurs. Another interesting phenomenon is that the change in photocatalytic efficiency is not synchronous with the optical bandgap. The possible reason is that the spontaneous polarization can provide the depolarization field that contributes to the separation of the electron–hole pair.^32,33 The orthorhombic Pnma phase of the B15SFO and B20SFO samples is centrosymmetric paraelectric, while the rhombohedral R3c perovskite structure is noncentrosymmetric ferroelectric.³⁴ The main phase structure of BFO, B5SFO, and B10SFO is rhombohedral, has ferroelectric properties, and exhibits spontaneous polarization. The resulting internal electric field can separate photogenerated electrons and holes, reducing recombination of the charges. In turn, the photocatalytic efficiency is improved. Opposite to this, the main phase of B15SFO and B20SFO is orthorhombic and exhibits paraelectric properties. In the absence of an internal electric field, photogenerated electrons and holes easily recombine, severely decreasing the photocatalytic performance of B15SFO and B20SFO as compared to that of BFO, B5SFO, and B10SFO.

The ferromagnetic hysteresis loops of these samples, measured at room temperature, are illustrated in Figs. 6(a)–6(e). The remnant magnetization first increases and then decreases with increasing Sm substitution content, as shown in Fig. 6(f). Similar results have been reported in the literature.³⁵ The coexisting phases such as the R3c phase, the bridging phase, and the Pnma phase combined with grain size contribute to the magnetization in BFO based nanoparticles. The R3c phase has an anti-ferromagnetic feature, while the ferromagnetism of the bridging phase is stronger than that of the Pnma phase. Ferromagnetism reaches the optimal value around the MPB region with a Sm substitution content of x = 0.1.

Double ferromagnetic hysteresis loops are observed in the Sm-substituted BFO nanoparticles. With increasing Sm content, the double ferromagnetic hysteresis loop is obvious around the MPB region (x = 0.1). The wasp-waisted hysteresis loops are rarely reported in BFO. In general, the appearance of the wasp-waisted hysteresis loop is induced by the coexistence of two magnetic phases such as ferromagnetism and antiferromagnetism or soft magnetism and hard magnetism/permanent magnets.^36,37 In the Sm-substituted BFO nanoparticles, the two magnetic phases may be related to the antiferromagnetic R3C phase and the ferromagnetic Pnma phase. Furthermore, an asymmetric shift of the hysteresis loop along the magnetic field axis is observed in the inserted M–H loop. The phenomenon is called the exchange bias (EB) effect, which is calculated based on H_E = (H₋ + H₊), where H₋ and H₊ are the negative and the positive coercive field, respectively.³⁸ The coupling effect between the ferromagnetic surface and the antiferromagnetic core of the nanoparticles with a large surface-to-volume ratio can induce the EB effect. In this study, the EB effect is only observed in the B5SFO, B10SFO, and B15SFO samples, whereas it is negligible in the BFO and B20SFO nanoparticles. The grain size of B5SFO and B20SFO nanoparticles is smaller than that of B10SFO and B15SFO nanoparticles, as shown in SEM images. Here, the EB effect can be induced by the orthorhombic antiferromagnetism coupled with the orthoferrite ferromagnetism rather than the breaking of the periodic spin structure at the surface,^39–41 which arises from the exchange interaction of various magnetic phases around the MPB, that is, the intrinsic exchange effect.

In summary, Sm-doped Bi_1−xSm_xFeO₃ (x = 0.05, 0.10, 0.15, 0.20) nanoparticles were synthesized using a double-solvent sol–gel method. The effect of Sm substitution content on their photocatalytic activity and magnetic properties was investigated. A structural translation from rhombohedral R3c to orthorhombic Pnma is indicated by XRD and Raman spectroscopy results. Meanwhile, the MPB and the bridging phase are formed when the Sm content is 0.1. The morphology and grain size of these nanoparticles are influenced by Sm-substitution. The bandgap of Sm-doped BFO nanoparticles is successfully modulated from 2.05 to 1.63 eV by increasing the doping concentration of Sm. The photocatalytic and magnetic properties both reach optimum levels around the MPB region.

This work was supported by the Program for Innovative Research Team in Science and Technology in the University of Henan Province (Grant No. 19IRTSTHN019), the China Postdoctoral Science Foundation (Grant No. 2020M672207), the Program for Young Teachers of Higher School in Henan Province (Grant No. 2019GGJS197), the Program for Excellent Team of Spectrum Technology and Application of Henan Province (Grant No. 18024123007), and the Program for Cultivation of National Project in Luoyang Normal University (Grant No. 2014-PYJJ-006). We would like to thank Editage (www.editage.cn) for English language editing.

The authors declare that they have no conflict of interest.

The data that support the findings of this study are available from the corresponding authors upon reasonable request.