3.1. Crystal structure and Rietveld refinement
Figure 1 presents the typical XRD patterns of pure YFO and YFMxO (0 ≤ x ≤ 0.1) powders at room temperature. The XRD patterns indicate a pure phase for all samples. Our study shows that YFO can be synthesized via hydrothermal method without having common impurity phases like Y2O2 and Fe2O3. As it is expected, no characteristic peaks related to Mn in the XRD patterns in comparison with the standard XRD patterns of pure YFO (orthorhombic Pnma: JCPDS file No. 01-086-0171), which is reasonable to suppose that all Mn ions have entered into the Fe-site of YFO. As is shown in the inset Fig. 1(a) that there is a slight shift of the major peak (200) to a bigger 2theta angle with the increase of Mn concentration. From the enlarged spectra in Fig. 1(b), other main peaks, such as (311) and (321) also have the same shifting trend. This shift in the diffraction angle might be ascribed to the unit cell contraction or the decrease in lattice constants because the ion radius of Mn3+ (rMn = 0.580 Å) is smaller than that of Fe3+ ion (rFe = 0.645 Å). However, it is worth noting that the intensity of the diffraction peaks reduced and merged partially to form broadened peaks after the Mn concentration further increased, especially for the sample with x = 0.1 (see in Fig. 1(a) and (b)), which is attributed to the presence of distortions, and a diminution in the crystalline size . From the crystallography point of view, the intensity of the peaks is usually related to the crystallinity, thus, the broadened width of the XRD peaks with the increase of Mn content indicates the decreasing crystallinity of YFMxO. The reduced crystallinity may be due to the reason that Mn3+ favors the creation of more nucleation sites, which in turn inhibits the growth of crystal grains.
To quantify the structure in detail and determine lattice parameters of the samples, an analysis of the XRD patterns by the Rietveld refinement was done using the Pnma space group in the orthorhombic unit cell. However, it is confirmed from the XRD results that the diffraction profiles belong to the orthoferrite structure for all samples. For such reason, only the samples with x = 0 and x = 0.1 were analyzed by Rietveld refinement, as shown in Fig. 2(a)-(b). Figure 2(c) shows the common crystal structure of the YFO. In this structure, Y3+ is surrounded by 12 O2− ions, and Fe3+ is surrounded by six O2− ions arranged in FeO6 octahedra. Table 1 in Fig. 2 shows the variation of unit cell parameters vs the content of Mn in YFMxO with x = 0 and x = 0.1. According to refined values, lattice parameters of YFO are a = 5.5944 Å, b = 7.6123 Å, and c = 5.2812 Å and they are comparable to those reported in works of literature [33, 34]. The goodness of fit χ2 is the conformity between an experimental result and theoretical expectation. R values are useful indicators for the evaluation of a refinement. As it is known, the Rietveld refinement results are reliable when the Rwp value is less than 10% and the χ2 in the range of 1.0 ~ 1.3. Small values of R factors indicate good consistency between observed and calculated results in XRD patterns, which means high reliability in our Rietveld refinement of YFMxO powders. As can be observed, the values of the lattice parameters, a, b and c, are decreased with increasing Mn content, leading to the contraction of the unit cell volume, because of the smaller ionic radius of Mn with respect to Fe, a similar phenomenon has been observed in Y doped BiFeO3 . The variations in the intensity of peaks and lattice parameters can be attributed to the incorporation of the dopant in the crystal . Thus, it is reasonable to believe that Mn3+ ions are introduced to the iron sites of YFO, which can also be further illustrated by EDX results.
3.2. Morphological evolution and composition
The morphology and phase structure of the pure YFO and YFMxO powders are investigated by SEM micrographs, as shown in Fig. 3(a)-(e). From these images, we can observe that the size average (ca. 10 µm) is nearly the same in all the particles with different shapes, except for the sample with x = 0.1. It is observed that the pure YFO exhibit a layered cuboid shape. When the Mn content x = 0.025, the multilayered cuboid is observed, and it is continuously layered with further doping (see Fig. 3(b)-(d)). In the hydrothermal crystallization processes of RFeO3, the addition of KOH could transfer R and Fe ions into amorphous hydroxides R(OH)3 and Fe(OH)3 for a very short time . The formation of RFeO3 can be described by the chemical reactions, as follows: R3+ + OH− = R(OH)3 (s); Fe3+ + OH− = Fe(OH)3 (s). The transition metal or rare-earth hydroxides usually form layered structures with ions inserted between the layers of metal hydroxide , which is in good agreement with the results observed by SEM. In contrast, when the Mn content reaches x = 0.1, the grain morphology changes to larger agglomerates shape with a remarkably reduced grain size. The larger agglomerates are composed of smaller particles. From the ionic radii point of view, smaller Mn ions can enter the Fe-site of YFO, which maintains the charge balance in the system. After Mn doping, mass transportation becomes weaker and the grain growth is inhibited. A similar reduction of particle size can be found in our previous work . Figure 3(f) presents schematic illustrations of the morphological evolution of YFMxO powders. This result indicates that the crystal shapes of YFMxO powders are strongly dependent on the concentration of Mn. The nucleation rate of the grains has changed when the Mn concentration exceeds a certain value, and this has resulted in a different grain morphology, from which it can be concluded that YFO is a suitable compound to study the shape-dependent physical properties. Elemental compositions, as determined by EDS analysis, are shown in Fig. 3(g)-(k). The obtained pure YFO (see Fig. 3(a)) reveals the existence of elemental Y, Fe, and O. The corresponding EDS patterns of YFMxO (see Fig. 3(h)-(k)) shows the characteristic peaks belong to the Y, Fe, Mn, and O, indicating the well doped of Mn in YFO. The spectra reveal that the higher Mn-doped molar ratio leads to the larger existence of Mn in YFO powders. To have a better comprehension of the structural evolution of YFMxO, we have performed XAFS analysis on all samples.
3.3. Fe and Y K-edge local electronic structure
Figure 4 shows the Fe K-edge XANES spectra of the pure YFO and YFMxO with Fe2O3 as a standard compound. As seen in Table 2 in Fig. 4, for all powder samples, the absorption edge energies were found at 7127 eV (less than 0.5 eV error) which is close to the absorption edge energy of reference Fe2O3 at ca. 7127.53 eV. The valence state of Fe2O3 is Fe3+, which means the Fe atoms in all YFMxO samples have the oxidation states of 3 + and all doping samples are dominated by the original orthorhombic Pnma phase without charge transfer from the Fe cations. Besides, the pre-edge peak position shifts towards higher energy with increasing oxidation state . In our samples, there is no pre-edge peak position shift, which further proves Fe is in 3 + valance state in all the samples. The XANES spectra of all these samples show an analogous pattern to each other, which substantiates the fact that Mn ions have occupied Fe-sites of YFO. The invisible pre-edge peaks are observed in the spectra of both pure YFO and YFMxO samples. The pre-edge peak is usually related to quadrupole transition from 1 s core state to 3d empty state, which is expected to be very weak for a Fe cation in an octahedral environment . It is well known that the pre-edge peak is a fingerprint of the octahedral coordination of Fe. Our spectra show almost no pre-edge shift as a function of x but their intensity is changing with x. In the enlarged XANES spectra in Fig. 4(a), it can be seen that, in the beginning, the intensity of the pre-edge peak is slightly increased. For 0.025 < x < 0.1, the intensity of the pre-edge peak begins to decrease and it is largest for the x = 0.1 sample. It is worth noting that for the compounds with x = 0.1 the pre-peak increases its intensity with x, which possibly indicates a decrease of the symmetry of the Fe environment. A similar phenomenon has been observed in the other perovskite ABO3 system [42, 43]. The increasing intensity in the pre-edge peak indicates the enhancement of the 1 s-3d electric dipole-forbidden transition while decreasing the intensity caused by the 1 s-4p dipole-allowed transition. These transitions have caused by Mn substitution, indicating a distortion of FeO6 octahedron. The two post-edge peaks are attributed to the transfer of 2p electrons in the oxygen 2p band to the Fe 3d orbital by a shakedown process . The intensity of these two post-edge peaks first increases then decreases when x = 0.1, as shown in Fig. 4(b). This indicates that the 3d-4p transition and charge transfer from the O 2p-Fe 3d is enhanced with both low and high doping contents of Mn due to the loss of inversion octahedral symmetry of the oxygens around the Fe atoms . These evolutions indicate that the local geometry and structure of Fe have changed.
In addition to the XANES data above, further analysis is carried out using the EXAFS. The Fourier transformation of the EXAFS is also shown. EXAFS features could provide useful information on both the short-range and the long-range orders (i.e., in the first shell and higher shell than the second). Figure 5 shows the variation of the observed k3-weighted EXAFS oscillation of the YFMxO powders. The error noise is observed above ca. 9 Å−1. Oscillations are still visible above ca. 12 Å−1, being less intense at the higher k, and show clear evolution as a function of Mn concentration, as given in Fig. 5(a). This phenomenon may be related to the less symmetric environment around Fe cations. The changes in the local structure could be better revealed in the Fourier transforms of the EXAFS oscillations providing real space information. The Fourier Transforms of the k3-weighted EXAFS spectra of the YFMxO samples are shown in Fig. 6 and the detailed coordination distances are listed in Table 3 in Fig. 6. The first and the second neighbor distributions in distance are easier to separate from the other shells in the Fourier transform. There are some characteristics peaks in spectra: (1) There are two strong amplitude peaks between 1 Å and 4 Å, with the first peak located at 1.54 Å, which is corresponding to the Fe-O coordinate peak due to the first oxygen coordination sphere of Fe ions. (2) The second strong peak is located at 3.28 Å, which is known as the Fe-Fe/Mn peak caused by the second rate nearby metal ions. (3) The small intensity of other peaks is not yet clear. They are probably due to the multiple scattering processes in the first coordination shell. Compared to the pure YFO, there is almost no shift of peak position (see Table 3 in Fig. 6) but the intensity of the Fe-O peak is decreased with the Mn content increases, as shown in Fig. 6(a). Moreover, with a close look at the Fe-O peak, we can observe that the intensity of this peak first decreases (x = 0.025) then slightly increases (x = 0.05) and then decreases again when x beyond 0.05. The reduction of the Fe-O peak intensity represents the loss of short-range order in the system. The intensity of the Fe-Fe/Mn peak also has the same trend (see Fig. 6(b)), which is due to the change of the local structure from Fe-Fe into Fe-Fe/Mn.
The normalized Y K-edge XANES spectra of the studied samples YFMxO, including the reference Y2O3 compound, are shown in Fig. 7. All the samples showed nearly similar near-edge features, indicating a similar local structure around Y ion in the first shell. The Y K-edge XANES spectra of all samples are similar to those of Fe K-edge XANES spectra without pre-edge peaks. Generally, the shift of the absorption edge energy in XANES is sensitive to the oxidation state of Y in the material. The edge position of the Y2O3 standard is approximately 17044.65 eV, while those of YFMxO samples are approximately 17044 eV (less than 0.5 eV error). These can be used simply as a fingerprint of phases and valence states, from which it can be seen that the edge positions of Mn-doped samples are quite similar to that of the standard sample of Y2O3. Thus, this result indicates that the Y ions in our samples are in the 3 + valence state. All samples show no pre-edge peaks, which is associated with the 1 s to 4d transition of Y. More specifically, this transition is partially allowed for the distortion of octahedral, only when p orbitals are mixed with d orbitals. The fact that this transition is not observed indicates a small distortion of the octahedral symmetry. From the examination of Table 4 in Fig. 7, it is clear that with Mn concentration increasing, there is no evident shift of the absorption edges in the whole series, but their intensity shows some difference. The two main peaks can be observed for all samples, which could be due to the transition from 1 s state to 5p state . For better clarity, the enlarged post-edge peak is shown in Fig. 7(a), from which it can be seen that the intensity of the first strong post-edge peak decreases as Mn content increases. As for the second strong post-edge peak, highest intensity can be found for x = 0.025 and x = 0.075 samples and lowest for x = 0.1 sample. The evolution of the post-edge peak intensity indicates that the local structure of Y has changed. Although Mn is doped in the Fe-site, the Y-site atom is also affected by the substitution.
Along the main edge, the XAFS spectra show a typical oscillation, which is caused by the scattering process of the electron wave near the nearest neighboring atoms. After the standard data processing, these waves yield of the χ(k)k3 function and its Fourier transformation data could provide valuable information about the coordination number, nearest neighbor distances, and the coordination geometry, etc. Among them, the radial distribution function is generated by backscattering along the R axis of the atom at different distances. This indicates the distance between the atom level and the central absorption atom. The analyzable k range of the EXFAS data is from 0 to 14 Å−1 and the spectra are weighted by k to amplify the oscillations at high k. The Y K-edge k3-weighted EXAFS curves of the YFMxO samples are given in Fig. 8. All the oscillations show similar patterns in the higher and lower k (see Fig. 8(a)), which indicates that the EXAFS functions of the Y atoms seem unchanged in all samples. The Fourier transforms of the k3-weighted EXAFS functions of the powder samples YFMxO are shown in Fig. 9. From the figure, it can be seen that the first and the second neighbor distributions are well separated from each other and other shells in the whole spectra. The primary features are two dominant peaks and other small peaks for all samples distributed at different distances: (1) The first shell has an R of 1.17 Å, corresponding to the Y-O peak caused by scattering of oxygen anions from the nearest neighboring Y atomic shell. (2) The second shell with an R of 2.54 Å, corresponding to the Y-Y peak, which can be explained by the scattering of oxygen anions from the next nearest neighboring Y atomic shell. The low second peak is a common feature in the case of EXAFS, which is often explained as the presence of high levels of disorder in the materials. (3) The other small peaks are probably due to a large number of multiple scattering in the first shell. More details are shown in Table 5 in Fig. 9, where the peak positions of YFMxO samples remain almost unchanged but the intensities of which are affected by Mn substitution. Compared to pure YFO, the intensity of the Y-O peak is increased then largely decreases for x = 0.1 sample. More specifically, the intensity of Y-O peak first increases with x = 0.025 then a little decrease when x = 0.05 and then increases again for x = 0.075, as shown in Fig. 9(a). Unlike the Y-O peak, the intensity of the Y-Y peak decreases from x = 0.025 to x = 0.1, as shown in Fig. 9(b). Although the changes in intensity can be explained by various reasons, such as the Debye-Waller factor, coordination number, and amplitude reduction factor, etc, the main reason for this reduction in intensity is most likely originated from the decrease in coordination number. These changes above indicated that the substitution of Mn ions not only affects the nearest neighbor atomic shell of Fe but also affects the nearest neighbor’s local structure of Y.
3.4. Optical measurement
Figure 10 demonstrates the transmission FT-IR spectra of the pure YFO and YFMxO powder samples. The ideal perovskite structure belongs to the space group of Oh1 symmetry. Based on group theory, the vibrations in the lattice are consist of 3F1ʋ + F2ʋ modes, of which the F2ʋ is inactive mode . Hence, the perovskite structure should have three absorption bands in IR spectra, which can be attributed to the B-O stretching vibration (F1:ʋ1), the B-O bending vibration (F1:ʋ2), and lattice vibration (F1:ʋ3). The expected energy order is ʋ1 > ʋ2 > ʋ3. Any deviation from the cubic symmetry will lead to the splitting of these three bands. At the first sight, our experimental result shows only two F1 bands in the IR spectra for all samples. They stem from the modes of Fe-O stretching vibration and O-Fe-O bending vibration, respectively, being characteristics of the octahedral FeO6 groups in the perovskite compounds . Based on further observation, it can be seen that the absorption peaks shift to a larger energy side with increasing Mn concentration, which implies the increase of the covalence of the B-O bond and the decrease of the iconicity, as given in Fig. 10(a). A similar phenomenon has been observed by the work of Cao et al and Kumar et al. [23, 49]. In Fig. 10(b), the bands around ca. 1382 cm− 1 and 3546 cm− 1 are representative of absorption of NO− 3 stretching vibrations from the small amount of trapped NO− 3 ions in the YFO and -OH from ambient moisture, during the experiment, respectively. However, the intensity of these two bands slightly increases with increasing Mn content, indicating the incorporation of Mn ions at Fe-site in YFO. Except for these peaks discussed above, no additional peaks occurred, which indicates the complete substitution of Mn. These are in consistent with the result of XRD and XAFS.
3.5. Magnetic property
As is commonly known, the magnetic properties in rare-earth orthoferrites originated from the super-exchange interaction of Fe3+-O2−-Fe3+, R3+-O2−-R3+, R3+-O2−-Fe3+. In the orthorhombic structure of YFO, there is no magnetic interaction between Y3+ and Fe3+. The Y3+ is diamagnetic as it has fully filled orbitals and each Fe3+ ion is surrounded by six O2− ions arranged in FeO6 octahedra, and the O2− is the common apex of the two adjacent octahedral, playing as a Fe3+-O2−-Fe3+ super-exchange interaction. The super-exchange interaction is an oxygen mediated exchange between transition metal ions based on virtual hopping processes of the oxygen 2p electrons. The O2− ions have fully occupied triple degenerate (2px, 2py, 2pz) orbitals with six electrons, while the 5d orbitals of Fe3+ split as triple degenerate t2g (dxy, dyz, dxz) and doubly degenerate eg (dx2−y2, dz2) orbitals. These eg orbitals of Fe3+ are along the crystal axes overlap with the 2p orbitals of O2−, which leads to the super-exchange interaction of Fe3+-O2−-Fe3+ through O2− ions at 180 oC. The GKA (Goodenough, Kanamori, and Anderson) rules were used to predict the super-exchange interaction and the anti-ferromagnetic nature of YFO . However, due to the Dzyaloshinski-Moriya antisymmetric exchange mechanism, each Fe3+ magnetic moment is not being exactly anti-parallel to the moments of the rest of the Fe3+ ions. This leads to the occurrence of weak ferromagnetism in YFO . The schematic representation of the magnetic structure of YFO is presented in Fig. 11(a) and (b), respectively. In this structure, the anti-ferromagnetism is along the a-axis with anti-parallel of Fe3+ spin with a small canted angle along the b-axis, as shown in Fig. 11(a). Figure 11(b) shows the weak ferromagnetism along the c-axis, the canted angle for all Fe3+ spin is parallel arrangement . Figure 12 shows the obvious magnetization hysteresis loops of pure YFO and YFMxO samples measured at room temperature. The saturation magnetization (Ms), remnant magnetization (Mr), and coercivity (Hc) values of the samples are listed in Table 6 in Fig. 12 for comparison. It is striking to note that all the samples behave as significant hysteresis loops, indicating the ferromagnetic properties of them. When an external field is up to 60 KOe, the magnetization has not reached saturation. In this structure, Y3+ is a diamagnetic cation and has a zero magnetic moment, thus, the observed ferromagnetism in our samples could be due to the spin canted of Fe3+ as the source of the magnetic moments . It means that the Fe3+ spins are not completely anti-parallel, but in reality, they may be canted. It is evident from Table 6 that when Mn content increased up to x = 0.075, Hc decreased while Ms and Mr increased compared to those of YFO. A sudden decrease of Ms and Mr values occurs for Mn content at x = 0.1. This phenomenon may be due to the lattice shrinking, resulting in enhanced internal stress with the increase of Mn substitution. From Fig. 12(a), from left to right, it can be seen that the magnetization of the samples strongly depends on the Mn concentration. Hysteretic loops of YFMxO have been narrowing with increasing Mn concentration, which reveals that the magnetic property changes from ferromagnetic to paramagnetic behavior. It can be accounted for several reasons as follows. First, it could be explained by the paramagnetic component and weak ferromagnetic component. Secondly, it could be due to the Fe3+-O2−-Fe3+ super-exchange interaction, which is the dominated magnetic interaction in the YFMxO system and may induce the weaker ferromagnetism. The Fe3+-O2−-Fe3+ super-exchange effect becomes stronger due to lattice change caused by different radius between Mn3+ and Fe3+ ions. Besides, the paramagnetic contribution from the Mn3+ ions cannot be ignored. A similar paramagnetic component caused by rare-earth ions has also been found in Bi1 − xGdxFeO3  and Y1 − xGdxFeO3 . The higher Mn concentration, the weaker ferromagnetic property, which is consistent with the change of the hysteretic loop.