Crystallographic alignment in the recombination stage in d-HDDR process of Nd-Fe-B-Ga-Nb powders

It is well known that by applying the hydrogen disproportionation desorption recombination (HDDR) treatment to Nd-Fe-B-based magnets, the initial Nd₂Fe₁₄B phase is disproportionated into α-Fe, NdH₂ and Fe₂B at HD process and they are recombined to Nd₂Fe₁₄B with a fine size of grains, which is approximately 300 nm at DR step.^1–3 The magnetic anisotropy of the resultant material can be controlled by applying a suitable H₂ pressure during the HD (P_HD) and DR processes, and from our previous study on the d(dynamic)-HDDR, the highest anisotropy can be obtained at 30 kPa of P_HD. This is because the reaction rate becomes suitable at this hydrogen pressure.^4–6 When high anisotropy is obtained, it is considered that the recombined Nd₂Fe₁₄B grains are crystallographically aligned.⁷ 

Along with efforts to obtain the higher degree of anisotropy, investigations of the mechanism that induces such a crystallographic alignment have been intensively performed to date.^7–16 Further, from the microstructural observation as well as the analysis of orientations of crystal grains, it is now recognized that the α-Fe and NdH₂ grains formed during the HD process tend to align with some type of crystallographic orientation.^8–14 Currently, there are several reports that describe the details of such an orientation observed at the state where the HD step was completed; however, not enough information is available on the alignment behavior during the DR process.

From these backgrounds, in the present study, the changes in the microstructure formed by the HD reaction and orientation relationship among α-Fe, NdH₂, and recombined Nd₂Fe₁₄B during the DR process were systematically investigated.

In the present study, Nd-Fe-B-based alloy with a composition of Nd_12.5Fe_balGa_0.3Nb_0.2B_6.2 was used, which was made using a book mold casting and subsequent homogenization by heating at 1140°C for 20 h under Ar atmosphere. Then the alloy was hydrogen decrepitated under 0.15 MPa of hydrogen atmosphere at room temperature for 2 h to convert it into powder form. For the d-HDDR treatment, the powder with a particle size less than 212 μm was used, and the treatment was carried out using a computer-controlled hydrogen furnace. As the HD step, the furnace chamber was firstly filled with 30 or 100 kPa of hydrogen atmosphere after the insertion of the sample powder, and then the temperature was increased to 820°C in 1 h. The temperature and P_HD was held for 90 min. As the DR step, the sample was subsequently heated at gently decreasing H₂ pressure for a specified time (T_DR). It took approximately 34 min to be 1 kPa from 100 kPa. When the completion of the DR treatment is required, the heating temperature was maintained for another 30 min in the vacuum.^5,6 However, in the present study, to investigate a T_DR dependence of sample properties, DR treatment with various T_DR (10, 20, and 30 min) was carried out, and then the furnace was cooled immediately to room temperature.

Microstructural observation of obtained samples was performed using scanning electron microscope (SEM, JSM-7800F, JEOL). The sample powder was fixed with a conductive epoxy resin and then polished so that the cross section of powder particles can be observed. For the investigation of orientations of crystal grains, electron diffraction using transmission electron microscope (TEM, JEM-2100, JEOL) as well as transmission electron backscatter diffraction (t-EBSD) analysis were carried out using samples processed by a focused ion beam. In the t-EBSD measurement, the orientation of the sample surface analyzed was not aligned to any specific crystallographic plane.

As mentioned in the introduction part and our previous reports,^5,6 the d-HDDR-treated sample with 30 kPa of P_HD exhibits excellent magnetic anisotropy. In contrast, with 100 kPa of P_HD the anisotropy becomes poor. Thus, in the following sections, the sample treated with 30 kPa of P_HD is mainly discussed.

Figure 1 represents SEM pictures showing the T_DR dependence of the microstructure of the samples (P_HD = 30 kPa). When T_DR was 0 min, i.e. when the HD process had just finished, NdH_2+x (bright part), α-Fe (gray part), and iron boride (dark part) phases were observed (Fig. 1(a)). Additionally, lamellar clusters (∼ 1 μm) consisting of fine rod-shaped NdH_2+x in an α-Fe matrix were also observed. Undisproportionated Nd₂Fe₁₄B phase was not found after 90 min of HD treatment in this study. This type of texture is commonly known as a hydrogen disproportionated state of Nd-Fe-B-based magnet.^8–12 It is a slightly difficult for the samples to avoiding absorption of the remaining hydrogen in the chamber during the cooling process of sample retrieval. Thus, the amount of hydrogen in the NdH_2+x is changed before cooling from the value at high temperature to room temperature. Similarly, the recombined Nd₂Fe₁₄B phase also forms Nd₂Fe₁₄BH_y in the obtained samples at room temperature.¹⁷ This behavior was confirmed by the changes in lattice parameters of these two phases calculated from the respective XRD patterns. The T_DR dependence of the lattice parameters was found to be very complicated, but it is not the main topic of the present paper, and thus the details are not discussed here.

In the DR process, along with the time elapsed, recombined Nd₂Fe₁₄BH_y appeared first at the rim parts of the NdH_2+x grains when T_DR was 10 min (Fig. 1(b)) as previously reported.^10,12 Simultaneously, it is found that almost lamellar clusters had already recombined to Nd₂Fe₁₄BH_y (X region in Fig. 1(b)). This change is reasonable when the reaction speed of the recombination is considered. The thickness of Nd₂Fe₁₄BH_y formed at the rim of NdH_2+x grains already reached approximately 100 nm, and this value is larger than the distance between rod-shaped NdH_2+x grains in the lamellar clusters (∼ 22 nm). When T_DR was 20 min, areas of NdH_2+x (bright part) and α-Fe (darker gray part) decreased, and the recombined Nd₂Fe₁₄BH_y phase (gray matrix) became dominant (Fig. 1(c)). Finally, only a little amount of NdH_2+x and α-Fe remained at 30 min of T_DR, suggesting a progress of the DR reaction (Fig. 1(d)). Although the shape of lamellar cluster after the HD step was different, above microstructural change during DR step was similar even when P_HD was 100 kPa.

Figure 2(a) shows a bright field image obtained from TEM observation of the sample treated with 30 kPa of P_HD and 10 min of T_DR. The electron diffraction patterns taken from the NdH_2+x grain, labeled B, and Nd₂Fe₁₄BH_y grains labeled F, G, and H, are also shown in Figs. 2(b) and 2(d)-2(f), respectively. As illustrated in Fig. 2(c), the diffraction patterns from [110] of NdH_2+x and [110] of adjacent α-Fe phase, labeled A, were observed simultaneously in Fig. 2(b), and the spots of (-220) of NdH_2+x and (-110) of α-Fe appeared in very similar positions. Moreover, all the diffraction patterns taken from other NdH_2+x grains (B, C, D, and E) also correspond to those taken from [110] zone axis of NdH_2+x. From these results, it is evident that the crystal orientation of NdH_2+x and α-Fe phases are aligned; this relationship can, therefore, be expressed as follows: [110]NdH_2+x // [110]α-Fe and (-220)NdH_2+x // (-110)α-Fe. The aligned crystallographic orientation between α-Fe and NdH₂ grains in the disproportionated stage was reported previously,^8–14 and from the present study, such an orientation relationship is found to be maintained even when the DR process is under progress. On the other hand, no relationship among recombined Nd₂Fe₁₄BH_y grains as well as between Nd₂Fe₁₄BH_y and α-Fe or NdH_2+x grains was observed at this stage. Figures 2(d)-2(f) are the superpositions of diffraction patterns from [110]α-Fe and Nd₂Fe₁₄BH_y. All the observed diffraction patterns of α-Fe (grain A) were the same as that in Fig. 2(b). However, there were no similar patterns amongst those from Nd₂Fe₁₄BH_y (grains F, G, and H) suggesting that the orientation of these grains is not aligned. The assignment of the indices of the zone axis to these patterns from Nd₂Fe₁₄BH_y was not well completed.

The sample treated with 30 kPa of P_HD and 20 min of T_DR showed characteristics similar to those shown by the sample with 10 min of T_DR. In this sample (T_DR = 20 min), the electron diffraction patterns were observed for an NdH_2+x grain, an adjacent α-Fe grain, and the recombined Nd₂Fe₁₄BH_y grains surrounding the NdH_2+x grain. The observed patterns confirmed that the orientation relationship between α-Fe and NdH₂ grains observed in the sample with 10 min of T_DR was retained, and the Nd₂Fe₁₄BH_y grains seemed to be oriented in random directions.

When the T_DR was prolonged to 30 min, the alignment of recombined Nd₂Fe₁₄BH_y grains could be observed clearly. Figure 3(a) shows a bright field image of TEM observation and Figs. 3(b)–3(e) show the electron diffraction patterns taken from four different grains I–L, respectively, in Fig. 3(a). Although a slight inclination was observed, it is evident that all the diffraction patterns were the same and the direction of the zone axis was [-121] of Nd₂Fe₁₄BH_y. This result indicates that these grains are aligned with the same orientation. The alignment and the orientation relationship among Nd₂Fe₁₄BH_y, NdH_2+x and α-Fe grains were confirmed by t-EBSD analysis of the sample. Figure 4(a) shows a t-EBSD color map taken from the area comprising Nd₂Fe₁₄BH_y grains. The majority of the grains in this area exhibited similar color, and the misorientation angle estimated from the (001) pole figure plot was approximated to within ±20°, as represented in Fig. 4(b). Moreover, as shown in Fig. 4(c), from the result of t-EBSD analysis at the area where all three kinds of phases (Nd₂Fe₁₄BH_y, NdH_2+x and α-Fe) were observed, the spots appeared in similar positions in the respective pole figures (Fig. 4(d)–4(f)) as indicated by arrows. This relationship can be described as follows: (110)NdH_2+x and (110)α-Fe // (001)Nd₂Fe₁₄BH_y. As for the sample treated with 100 kPa of P_HD and 30 min of T_DR, an orientation relationship was absent in the electron diffraction patterns taken from several recombined Nd₂Fe₁₄BH_y grains.

The abovementioned difference in the alignment induced by the change of P_HD was also observed in the demagnetization curves of these two samples (P_HD = 30 or 100 kPa, T_DR = 30 min) as shown in Fig. 5. For the measurement, the sample powder was aligned in the 1.6 MA/m of magnetic field and fixed in the cylindrical capsule using wax, and then magnetized at 4.0 MA/m in a pulsed field. When P_HD was 30 kPa, the shape of the curves was different by the measurement direction, namely, along parallel or perpendicular to the axis of magnetic alignment, suggesting that the magnetic anisotropy was induced by the aligned Nd₂Fe₁₄BH_y (Fig. 5(a)). On the other hand, at 100 kPa of P_HD, almost identical curves were obtained irrespective of the measurement direction (Fig. 5(b)). The sample was isotropic in this case because recombined Nd₂Fe₁₄BH_y was not aligned, as mentioned earlier. In both cases, the recombination process was not completely finished, therefore the coercivity was very low.

A key feature of d-HDDR treatment is the capability to maintain a suitable reaction rate throughout the disproportionation and recombination processes,^4,5 and this serves to control the retention of crystallographic orientation relationship and the resultant magnetic anisotropy.

The present study describes the transition of microstructure and crystallographic orientation relationship during the DR process. The behavior of the change in the microstructure seemed to be not very different by P_HD, while the aligned orientation relationship was found only in the sample treated with 30 kPa of P_HD. The alignment of α-Fe and NdH_2+x grains ([110]NdH_2+x // [110]α-Fe and (-220)NdH_2+x // (-110)α-Fe) developed after HD step was found to be retained until they almost disappear and recombine to form Nd₂Fe₁₄BH_y. The orientation relationship among the recombined Nd₂Fe₁₄BH_y and remained α-Fe and NdH_2+x ((110)NdH_2+x and (110)α-Fe // (001)Nd₂Fe₁₄BH_y) becomes more evident along with the progress of the DR process. There are some reports suggesting the possible role of Fe₂B as a memory site in the induction of the anisotropy,^11,14,16 however in the present study, the detection of iron boride phase was difficult, particularly in the DR step, and the role of B is still unclear. Considering that the amount of Fe₂B formed during the HDDR process is small, an adequate control of reaction speed that maintains the orientation relationship between α-Fe and NdH₂ seems rather important. Present observations are carried out only in the local region of the sample. The evaluation of the orientation relationship in a wider area of the sample is under investigation.

This work was supported by “Future Pioneering Projects/Development of Magnetic Material Technology for High-Efficiency Motors” from NEDO, Japan.

The authors thank Dr. K. Kobayashi for the help of TEM and t-EBSD measurements as well as valuable discussion.