Magnetic anisotropy and crystallographic alignment in Fe and NdH₂ during d-HDDR process of Nd-Fe-B-Ga-Nb powders

The hydrogen disproportionation desorption recombination (HDDR) treatment of Nd-Fe-B magnets is known to be an effective method for obtaining powders with good magnetic anisotropies and high coercivities.^1–3 The authors found that the degree of anisotropy can be changed by controlling the hydrogen pressure during HDDR treatment, and this method was called d(dynamic)-HDDR treatment.^4,5 The best anisotropy was obtained at 30 kPa of H₂ pressure during HD (P_HD), and the anisotropy decreased at higher P_HD.⁶ 

Various models of the induction mechanism of magnetic anisotropy during HDDR treatment have been reported to date.^7–16 The anisotropy in HDDR arises due to the crystallographic alignment of recombined Nd₂Fe₁₄B grains, and we proposed that the crystallographic alignment between the Fe and NdH₂ phases after the HD treatment and the initial and recombined Nd₂Fe₁₄B phases during the HDDR process is important.^17–19 

As supporting results of this hypothesis, the authors recently found that the degree of anisotropy (DOA) of d-HDDR-treated powder and the area fraction of fine lamellae observed after HD treatment decreased similarly with increasing HD treatment time (Fig. 1).²⁰ This decreasing tendency of the DOA is similar to that reported by Han.²¹ DOA was defined as DOA = (J_r - J_r^hard)/J_r.^21,22 The values J_r and J_r^hard represent residual magnetizations obtained by applying a magnetic field parallel or perpendicular to the magnetically aligned direction of the samples after d-HDDR, respectively. As for the fine lamellar structures, it is known that the Fe matrix and rod-shaped NdH₂ grains align with the same crystallographic orientation.^8–10,17 From these results, we concluded that the crystallographic alignment between Fe and NdH₂ in the fine lamellae formed after HD treatment resulted in the highly aligned Nd₂Fe₁₄B grains and resultant good anisotropy after DR.

However, several unclarified points remain. Although the fine lamellae disappeared after 10 h of HD treatment at a P_HD of 30 kPa (Fig. 1(b)), the DOA retained a relatively high value of 0.7 (Fig. 1(a)). This tendency cannot be explained only by the existence of fine lamellae. More recently, the authors found other structural difference depending on P_HD.²³ In the sample HD-treated at a P_HD of 30 kPa for 10 h, the characteristic structures consisted of thick rod-shaped NdH₂, and Fe matrices were mainly observed. These structures are believed to have been formed through the growth of fine lamellae, which are usually seen in the early stage of HD treatment. Thus, this structure is referred to as the “coarsened lamellar structure.” On the other hand, in the sample HD-treated at a P_HD of 100 kPa for 10 h, spherical NdH₂ grains in the Fe matrices were the main component, and we referred to this structure as the “spherical structure.” The details of the microstructures, their formation mechanisms, and the relationship with the DOA will be described elsewhere.²⁴ This result suggests that there is a relationship between the presence of the coarsened lamellae and the high DOA. Currently, unlike the fine lamellae, the crystallographic relationship between Fe and NdH₂ in the coarsened lamellae and spherical structures is not well understood. Therefore, the crystallographic orientation relationship between the Fe and NdH₂ grains in these two microstructures observed after HD treatment was investigated.

An Nd-Fe-B-based mother alloy with a composition of Nd_12.5Fe_balGa_0.3Nb_0.2B_6.2 (at%) was prepared using book mold casting and subsequent homogenization at 1140°C for 20 h under an Ar atmosphere. Hydrogen decrepitation was applied to the obtained alloy ingot under a hydrogen atmosphere at 25°C and 0.15 MPa for 2 h. The resultant material was sieved, and the obtained powders (<212 μm) were used as a starting material.

In this study, two HD-treated samples, A (P_HD = 30 kPa) and B (P_HD = 100 kPa), were prepared. For the HD treatment, the furnace chamber was filled with hydrogen after the sample powder was set, after which the temperature was raised to 820°C over 1 h. The temperature and pressure were held for 10 h, after which the furnace was cooled immediately to room temperature. An HD time of 3 h yields the best anisotropy. However, the larger grain sizes obtained after a longer HD time of 10 h were necessary for better phase detection in the t-EBSD measurement.

Microstructural observations were performed using a scanning electron microscope (SEM, JSM-7800F Prime, JEOL). To investigate the orientations of the crystal grains, transmission electron backscatter diffraction (t-EBSD, Hikari, TSL) analysis was conducted for the sample pieces (15 × 10 μm²) processed using a focused ion beam system (FIB, JIB-4600F Multi Beam System, JEOL). In the t-EBSD analysis, the measured area was divided into hexagonal cells with a step size (i.e., the distance between the centers of the hexagons) of 30 nm, and each hexagon corresponded to one measurement point.

Figures 2(a) and (b) show the SEM-BSE images for the inner and near surface regions of sample A (P_HD = 30 kPa), respectively. In the inner region, coarsened lamellae were mainly observed, whereas in the near surface region, spherical structures were dominant. Since coarsened lamellae extended three-dimensionally in various directions, it was difficult to determine the exact boundaries of these regions from the two-dimensional SEM-BSE images. However, in the region approximately 10 μm from the powder surface, even the coarsened lamellae that were extending in the in-plane direction were not observed. Therefore, there is a microstructural difference between the inner and near surface regions. In contrast, in the sample B (Fig. 2(c)), spherical structures were the main components, and coarsened lamellae were not observed in any parts of the sample. In both samples, it was difficult to distinguish Fe₂B particles by the contrast of BSE images.

Figures 3(a)–(f) show the results of t-EBSD analysis of a FIB-processed piece from an inner part of sample A. As explained above, coarsened lamellae were observed in the SEM-BSE image (Fig. 3(a)). In the inverse pole figure (IPF) map, since each measurement point is colorized with a color corresponding to the space group and the crystallographic orientation, we can evaluate the widths of regions with the same orientations in the sample. We refer to such a region as an “equi-oriented region.” From the IPF map of Fe (Fig. 3(b)), the average diameters of the equi-oriented regions were calculated to be approximately 1100 nm.

Since the space groups of Fe and NdH₂ are Im-3m (bcc) and Fm-3m (fcc), respectively, and they belong to the same cubic system, the correspondence of crystallographic orientation to a color in the IPF maps is also the same. Thus, by combining Figs. 3(b) and (c), the crystallographic orientation relationship between the Fe and NdH₂ grains could be visualized, as shown in Fig. 3(d). In this figure, the boundary at which the misorientation angle was larger than 5° is shown by red lines. The black areas in this figure correspond to Fe₂B grains or measurements points at which the determination of the crystallographic orientation was not successfully performed due to poor signal intensity, unsuitable surface conditions, and other factors. As shown in Figs. 3(b)–(d), the colors (i.e., crystallographic orientation) of the NdH₂ grains inside the equi-oriented region of Fe were the same as that of surrounding Fe. Finally, the boundaries of equi-oriented regions including Fe and NdH₂ could be estimated, as shown by the yellow lines in Fig. 3(e). In this figure, each yellow shaded region divided by yellow lines corresponds to one equi-oriented region. The total area fraction of these equi-oriented regions was approximately 70% of the area of the Fe and NdH₂ phases in the visual field (10 × 10 μm²). Figure 3(f) shows the distribution of the deviation angle of each measurement point from the averaged crystallographic orientation in a typical equi-oriented region. The deviation in the crystallographic orientation within each equi-oriented region was usually less than 2°. Thus, the value of 5° of the misorientation angle, used as a criterion for the boundary of the equi-oriented region, was adequate.

Similar to Fig. 3(e), the equi-oriented regions and their boundaries were estimated for the FIB-processed piece from a near surface region of the same sample, as shown in Fig. 3(g) by yellow shaded regions and lines, respectively. The bottom part of this figure shows the surface region of the sample powder, and spherical structures were mainly observed. Moreover, the distribution of equi-oriented regions was sparse in this region.

In sample B, a different behavior was observed. Figures 4(a)–(c) show an SEM-BSE image of the FIB-processed sample piece and IPF maps for Fe and NdH₂, respectively. Figure 4(a) shows the spherical morphologies of the structures. From Fig. 4(b), the average diameter of the equi-oriented regions of Fe was calculated to be approximately 530 nm. With the same procedure as that used for Fig. 3(e), the equi-oriented regions and their boundaries were estimated, as shown in Fig. 4(d). The total area fraction of equi-oriented regions in this sample was approximately 25% of the area of the Fe and NdH₂ phases. Moreover, these regions were distributed nearly uniformly within the sample.

These results suggest that the crystallographic orientations of Fe and NdH₂ remain aligned in coarsened lamellar structures, whereas the orientations become different in the spherical structures. In addition to the structural differences, there were large discrepancies in the amount and average diameter of the equi-oriented regions of Fe depending on P_HD. These differences are believed to have been induced by the differences in the rates for the reaction between hydrogen and the sample powder.

In the sample A, most of the equi-oriented regions comprised coarsened lamellae (Fig. 3(e)), and these structures were the result of morphological changes of the fine lamellae during the HD treatment.²⁵ In the fine lamellae, the Fe matrix and NdH₂ rods align in the same crystallographic orientation as mentioned earlier.^8–10,17 Typically, since such highly aligned structures are in an energetically unfavorable state, they grow into coarsened structures to reduce the interfacial energy and finally become spherical grains as the HD reaction proceeds. However, from the pressure-temperature diagram for the HD and DR reactions, a P_HD of 30 kPa is very close to the equilibrium curve at 820°C,¹⁸ and thus, the reaction rate was low. Therefore at 30 kPa, the aligned orientation relationship in the fine lamellae could be maintained even after the coarsening and relatively wider equi-oriented regions (1100 nm in diameter) formed with a large area fraction (70%). However, since the HD reaction initiated at the surfaces of sample particles, spheroidization of the microstructures and the resultant misalignment between the Fe and NdH₂ became significant at the near surface region, even at 30 kPa, as shown in Fig. 3(f).

However, at a P_HD of 100 kPa (sample B), the hydrogen reacted with the powder particles more rapidly than at 30 kPa. As a result, hydrogen diffused throughout the sample during the heating to 820°C, and the HD reaction initiated randomly throughout the powder particles.¹⁸ By the same reasoning, the morphologies of the microstructures easily become spherical in this sample. Therefore, the sizes of the equi-oriented regions were small (530 nm in diameter), and their area fraction was low (25%). The evenly distributed equi-oriented regions in Fig. 4(d) are thought to be the result of the random initiation of HD reactions, as mentioned above.

Finally, the tendencies of the DOA in Fig. 1(a) can be explained. At 30 kPa, even if the fine lamellar structures that resulted in highly aligned Nd₂Fe₁₄B grains disappeared after 3 h of the HD reaction, coarsened lamellar structures in which Fe and NdH₂ were aligned still remained as major components of the sample. Therefore, a relatively higher DOA (0.7) was obtained. In contrast, at 100 kPa, fine lamellar structures were observed only in the very early stage of HD treatment, after which the spherical structures became dominant and the DOA decreased.

From the results of the present study, a strong relationship between the microstructures, the presence of crystallographic alignment in Fe and NdH₂, and the high anisotropy was demonstrated. These results indicate that similar to the fine lamellae, the coarsened lamellae also maintain crystallographically aligned Fe and NdH₂, and at least either of these aligned structures is necessary for the induction of high anisotropy during d-HDDR treatment. However, in the spherical structures, the aligned orientation relationship is absent, and their contribution to anisotropy is less significant.

This work was partially supported by the Future Pioneering Projects/Development of Magnetic Material Technology for High-Efficiency Motors from NEDO and the Elemental Strategy Initiative Center for Magnetic Materials (ESICMM) under the outsourcing project of MEXT, Japan.