Determination of Dy substitution site in Nd2−xDyxFe14B by HAADF-STEM and illustration of magnetic anisotropy of “g” and “f” sites, before and after substitution

Nd2Fe14B and Nd2−xDyxFe14B (x = 0.25, 0.50) particles were prepared by the modified co-precipitation followed by reduction–diffusion process. Bright field scanning transmission electron microscope (BF-STEM) image revealed the formation of Nd–Fe–B trigonal prisms in [− 101] viewing zone axis, confirming the formation of Nd2Fe14B/Nd2−xDyxFe14B. Accurate site for the Dy substitution in Nd2Fe14B crystal structure was determined as “f” site by using high-angle annular dark field scanning transmission electron microscope (HAADF-STEM). It was found that all the “g” sites are occupied by the Nd, meanwhile Dy occupied only the “f” site. Anti-ferromagnetic coupling at “f” site decreased the magnetic moment values for Nd1.75Dy0.25Fe14B (23.48 μB) and Nd1.5Dy0.5Fe14B (21.03 μB) as compared to Nd2Fe14B (25.50 μB). Reduction of magnetic moment increased the squareness ratio, coercivity and energy product. Analysis of magnetic anisotropy at constant magnetic field confirmed that “f” site substitution did not change the patterns of the anisotropy. Furthermore, magnetic moment of Nd2Fe14B, Nd2−xDyxFe14B, Nd (“f” site), Nd (“g” site) and Dy (“f” site) was recorded for all angles between 0° and 180°.

www.nature.com/scientificreports/ linear combination of pseudo-atomic-orbital (LCPAO) method, while Liu et al. used full potential plane-wave plus muffin-tin orbital method. Experimental evidences for Dy site in (Nd-Dy) 2 Fe 14 B crystal lattice was provided by Itakura et al. 22 Although they presented STEM-HAADF image but d-spacing value for the [001] facet deviated from the standard values. Absence of the stoichiometric ratio between Nd:Dy:Fe and distribution of Dy only near grain boundary raised more questions, hence it was interesting for the contemporary researchers to answer them. Saito et al. 23 used neutron diffraction technique and determined the quantitative distribution of Dy at substitution sites. They postulated that Dy substitutes at both the "f " and "g" sites, and population of Dy on each site depends on the annealing temperature of the experiment. This finding was different from the previous studies, those suggested that Dy only substitutes at "f " site. Furthermore, magnetic anisotropic properties of "f " and "g" sites in Nd 2−x Dy x Fe 14 B were not yet studied. In order to answer the questions discussed above and a comprehensive study of "f " and "g" sites in Nd 2−x Dy x Fe 14 B was required.
In this work, we prepared Nd 2 Fe 14 B and Nd 2−x Dy x Fe 14 B with optimized co-precipitation method followed by reduction-diffusion (R-D) process. This co-precipitation method is similar with the approach reported by Ma and Palaka et al. with minor modification 24,25 . Detail of this modification is explained in the "supplementary information". Site preference of Dy in Nd 1.5 RE 0.5 Fe 14 B particles was confirmed through the crystal structure determination by using HAADF-STEM. Effect of Dy substitution on the magnetic moment, magneto-crystalline anisotropy energy and coercivity of the Nd 2−x Dy x Fe 14 B particles is also studied. Nd 2−x Dy x Fe 14 B sample was rotated in the rotating angle range of 0° to 180° at the constant magnetic field to observe variation of anisotropic patterns after the substitution. Effect of the substitution on the anisotropic properties of "f " and "g" sites in the Nd 2−x Dy x Fe 14 B crystal is also studied.

Methods
Materials. All  Preparation of Nd 1.5 RE 0.5 Fe 14 B particles. Schematics for the preparation procedure is shown as the Fig. S1 and detailed process of synthesis is explained in the "supplementary information". In brief, all the metal chlorides of Nd, Fe and Dy were dissolved in deionized (D.I.) water under the stirring to obtain the clear solution. NaOH (3.5 M) was added to the solution in a drop wise until the pH approached to 13. The resultant solution was stirred continuously for 4 h. Then, products were washed twice with D.I. water and ethanol, and dried overnight at 353 K (80 °C) followed by annealing at 973.15 K (700 °C) for 30 min to convert all hydroxides to oxides. The product was mixed with boric acid and CaH 2 in a glove box and then pressed into pellet form. The pellet was undergone to R-D via annealing at 1273.15 K (1000 °C) for 3 h with Ar flowing in the furnace. The pellet after R-D was pulverized and washed with water to remove calcium oxide (CaO) completely and rinsed with acetone. Samples preparation for SEM and TEM. Nd 2 Fe 14 B and Nd 2−x Dy x Fe 14 B powder samples were placed on the movable lower ram and then the prepared solder pieces were placed on the sample. Electrically conductive polymer was poured to encapsulate the solder and powder samples, then heated at 180 °C for 6 min and then pressed under the pressure of 30 kN. Prepared sample was mounted on SEM holder, then mechanically polished with SiC paper, diamond suspension and colloidal silica, subsequently characterized with FE-SEM.
Specimens for TEM (transmission electron microscope) were prepared by focused ion beam (FIB-NX2000, Hitachi) using the lift-out technique. For TEM measurement, sample was treated as the same process reported by Kim et al. 26 and orientation of sample along [100] zone axis was confirmed by using electron backscatter diffraction (EBSD) (TEAM™ Pegasus, Ametek Co. Ltd. USA).
Characterization. Crystal structure and phases were determined by X-ray diffraction (XRD) patterns using a Rigaku Diffractometer (XRD, Rigaku). The morphology, size and elemental distribution were observed with field emission scanning electron microscope (JSM-7000F, JEOL), conventional transmission electron microscopy (TEM, JEM-2100F) and aberration corrected TEM (ARM-200F) with energy dispersive X-ray spectroscopy (EDS). TEM was operated at the accelerating voltage of 200 kV. Angular dependent magnetic properties were measured by magnetic property measurement system (MPMS3-Evercool) equipped with rotator. www.nature.com/scientificreports/

Results and discussion
Nd 2 Fe 14 B and Nd 2−x Dy x Fe 14 B magnetic particle were prepared by co-precipitation followed by reduction diffusion process. Chemical reactions and mechanisms during co-precipitation and R-D processes are explained in the "supporting information". R-D reaction follows the mechanism proposed by the Haider et al. 27 XRD patterns for Nd 2 Fe 14 B and Nd 2−x Dy x Fe 14 B particles are similar due to the almost same crystal structure (Fig. 1a). They have the Nd 2 Fe 14 B (JCPDS #36-1296) as main phase with additional peaks corresponding to the extra Nd phase. Nd substitution with Dy makes the peaks position be shifted to the right side (Fig. 1a). This is due to the different crystal lattice parameters of Nd 2−x Dy x Fe 14 B and to the smaller ionic radii of Dy (178 pm) as compared to the Nd (181 pm). Both "a "and "c" dimensions of the crystal lattice were decreased after the Dy substitution in the Nd 2 Fe 14 B crystal lattice (Fig. 1b). The decreased values of these Nd 1.5 RE 0.5 Fe 14 B crystal parameters are also evidence of Dy substitution in Nd 2 Fe 14 B crystal. In order to calculate the lattice parameters ("a" and "c"), at first d-spacing values were calculated from the XRD patterns ( Fig. 1). h, k and l values were determined from Nd 2 Fe 14 B JCPDS #36-1296. Finally, "c" and "a" value were calculated by the following equation as the same method reported by Rahimi 17 , Cullity 28 , and Charbel 29 , et al. www.nature.com/scientificreports/ SEM-BSE images in Fig. 1c-e revealed that the particles were in irregular shape and the size distribution is in the range of 0.3-10 μm. The Nd 2 Fe 14 B had the largest average particle size as 3.5 μm while Nd 1.5 Dy 0.5 Fe 14 B had the least average particle size as 0.8 μm. Nd 1.75 Dy 0.25 Fe 14 B had the average particle size of 1.7 μm. SEM-EDS confirmed that Nd and Dy are homogeneously distributed with Fe in the particles (Figs. S5, S6). The microstructure, elemental composition and distribution of Nd 1.5 Dy 0.5 Fe 14 B particles were evaluated with STEM as shown in Fig. 2. The elemental distribution of Nd, Dy, Fe, and O was investigated by using STEM-EDS, which confirmed that the Dy was substituted for Nd in the crystal structure and it was distributed inside the grain. Figure 2b is the line EDS taken from the blue circle of LAADF-STEM image of Nd 1.5 Dy 0.5 Fe 14 B. Figure 2c shows the EDS line profile of interface between two fused Nd 1.5 Dy 0.5 Fe 14 B particles, as marked with the blue circle in Fig. 2a. No oxygen was detected in EDS mapping because boundaries of the particles were not exposed to the water during washing process.
To evaluate the crystallinity of the specimen, SAED patterns (Fig. 3b) of the marked area with red circle in the TEM image (Fig. 3a) were obtained. It was confirmed that the particles produced were single crystalline. It was deduced by SADP of strong diffraction maxima that each grain was completely single crystalline. Figure 3d   www.nature.com/scientificreports/ HAADF-STEM image can be used to distinguish the Dy and Nd, and their positions ("f " or "g" site). Figure 4b is HAADF-STEM image, which confirms the same arrangement of atoms as the standard Nd 2 Fe 14 B [100] zone axis (Fig. 3a). Intensity histogram for red dotted panel in Fig. 4b was acquired. It is observed that the intensity of atoms (Dy) at "4f " column is higher than that of the atom (Nd) at "4 g" column. Higher peak intensity confirms that the substitution site of Dy is "4f " site because the atomic number of Dy (66) is larger than that of Nd (60), which leads to the higher intensity as compared to the Nd. The 'a' value of Nd 1.5 Dy 0.5 Fe 14 B crystal lattice is 8.78 Å (Fig. 3c), which is well consistent with the standard Nd 1.5 Dy 0.5 Fe 14 B value. Standard Nd 2 Fe 14 B have 'a' and 'b' values of lattice parameters as 8.80 Å 30 . Standard distance of atomic column between "4f " site and "4 g" site of Nd 2 Fe 14 B is 1.1 Å. In this study, the obtained distance is 1.09 Å as shown in histogram Fig. 4c. This is well matched with theoretical value. A slight error is due to a noise induced by the fine drift of the sample or the poisson noise in the STEM. Hence, site preference for the Dy in Nd 2 Fe 14 B is proved to be "4f " as the previously theoretically reported by Liu et al. 21 . It was found that 100% "g" sites are occupied by the Nd (Fig. 5) and Dy was substituted only at "f " site.
Fe is ferromagnetic with electronic configuration of [Ar] 3d 6 4s 2 . This electronic configuration shows that it has eight valance electrons. Arrangement of the electrons in the relevant orbitals is shown in the rigid band model as Fig. 6c. Density of the electrons was taken on X-axis and energy was taken on Y-axis. E f indicates the Fermi level of the rigid band. Energy level of 3-d electrons is similar to the 4 s electrons, hence, there is no movement of electrons between the 4s and 3d orbital. Four unpaired electrons will be in the spin up configuration. Presence of the unpaired electrons makes the Fe ferromagnetic.  www.nature.com/scientificreports/ Nd 2 Fe 14 B and Nd 2−x Dy x Fe 14 B are expected to be anisotropic, hence, magnetic moment and energy density are angular dependent magnetic properties. Closely packed particles of Nd 2 Fe 14 B and Nd 2−x Dy x Fe 14 B were aligned at 5 T with easy direction of magnetization, then magnetic field was reduced to the 200 Oe. Thereafter, sample was rotated in the angle range of 0°-180°. Figure 7a explains the preparation of the sample for the measurement of magnetic anisotropy. During the measurement at MPMS, the magnetic particles were closely packed, which stopped the rotation of the particles at low applied magnetic field (200 Oe). All processes were performed at room temperature so that the effect of thermal energy was neglected. Figure 7b shows the anisotropic character of the Nd 2 Fe 14 B and Nd 2−x Dy x Fe 14 B. When magnetic particles were rotated in the range of 0°-180° at constant applied magnetic field of 200 Oe, magnetic moment of the particles was changed significantly. Nd 2 Fe 14 B and Nd 2−x Dy x Fe 14 B have maximum magnetic moment along "c" crystal direction, with parallel/anti-parallel orientation to the applied field of (θ = 0°, 180°). This is because of the "c" crystal dimension is easy direction for magnetization. On the contrary, along "a and b" crystal direction magnetic moment approached to zero.  14 B 19-21 . This is also confirmed in this work (Fig. 4). Hence in case of Nd 1.75 Dy 0.25 Fe 14 B and Nd 1.5 Dy 0.50 Fe 14 B there is negligible change in the magnetic moment on the "g" site after substitution. In Nd 2 Fe 14 B formula unit, Nd is distributed equally among 50% "f " and 50% "g" sites. Magnetic moment on "f " and "g" sites of in Nd 2 Fe 14 B is determined as 2.69 and 2.67 μB, respectively.
In Nd 1.75 Dy 0.25 Fe 14 B and Nd 1.5 Dy 0.50 Fe 14 B, Nd occupies 75% and 50% "f " sites, simultaneously Dy occupies leftover 25% and 50% "f " sites respectively. Hence magnetic moment on one "f " sites in Nd 1.75 Dy 0.25 Fe 14 B and  , Nd (f) and Nd (g), anisotropic behavior of "f " and "g" sites was studied. Figure 7a explains the sample preparation for the measurement   Figure 7b graphically illustrates that the magnetic moment of Nd 2 Fe 14 B and Nd 2−x Dy x Fe 14 B as the function of rotating angle in the constant magnetic field. Figure 7c-e shows the variation of individual magnetic moments of "f " and "g" sites as the function of rotating angle.

Conclusions
Nd 2 Fe 14 B and Nd 2−x Dy x Fe 14 B (x = 0.25, 0.50) particles were successfully prepared by the modified co-precipitation method followed by reduction-diffusion process. Micro-structure analysis of the composition and distribution of elements confirmed the homogeneous distribution of Dy atoms in the crystal lattice of Nd 2 Fe 14 B and Dy substitution at "4-f " site. Orbital magnetic moment (L) direction of the unpaired valance electrons of Dy was opposite to the Fe and Nd which resulted in the anti-ferromagnetic coupling between them. Nd substitution with Dy on "f " site reduced the magnetic moment of Nd 2 Dy x Fe 14 B due to anti-ferromagnetic coupling with Nd and Fe, but enhanced the energy density, squareness ratio and coercivity. Furthermore, it was found that Nd 2 Fe 14 B and Nd 2−x Dy x Fe 14 B particles have maximum magnetic moment when they are aligned parallel or anti-parallel to the applied magnetic field and have the minimum energy density when they are rotated perpendicular to the applied magnetic field. Conclusively, "f " site substitution of Nd with Dy in Nd 2 Fe 14 B did not change the anisotropic patterns of Nd 2−x Dy x Fe 14 B.