Microstructure and Magnetic Properties of Grain Refined Pr₂Co₁₄B Melt-Spun Ribbons

Abstract

The correlation between the grain refining effect of TiC on the microstructure of Pr₂Co₁₄B melt-spun ribbons and the magnetic properties is presented in this study. TiC enabled greater control of microstructure both in the as-spun and heat treated Pr₂Co₁₄B, compared with the material without TiC. As a result, coercivity of the sample with TiC was nearly twice that of the sample without TiC. In addition to Pr₂Co₁₄B, two other phases were found in the sample with TiC: one rich in Co and the other having a composition near PrCo₂. TiC was found near the grain boundaries and at triple junctions. Also no Ti or C was found in the matrix phase indicating extreme low solubility of the elements when both are present with Pr₂Co₁₄B. As expected, both the samples with and without TiC have similar anisotropy field but the presence of room temperature non-ferromagnetic phases (TiC and PrCo₂), caused a small decrease in magnetization of the sample with TiC although the romance of the isotropic materials were comparable.

1. Introduction

The high intrinsic coercivity and outstanding maximum energy product of sintered Nd₂Fe₁₄B magnets [1,2] make them excellent candidates for many applications in electric motors, medical devices, generators, speakers, sensors, transducers and consumer electronics [3,4]. However, due to the relatively low Curie temperature (585 K), Nd₂Fe₁₄B magnets lack sufficient thermal stability above 380–450 K [5]. Studies have been performed to enhance temperature dependence of the magnetic properties of the Nd₂Fe₁₄B magnets by both composition and microstructural modifications [1,2,6,7,8]. Dy has been added to enhance coercivity and high temperature performance of Nd₂Fe₁₄B [4,9,10,11,12] but maximum operating temperature of commercial grade materials are still limited to ≤500 K [13,14]. Moreover, the forecasted long term criticality and high cost of Dy could become a major obstacle to the production of Dy-enhanced Nd₂Fe₁₄B magnets [3]. Therefore, it is of interest to advance research in developing alternative magnet materials with high Curie temperature, high coercivity and reasonable energy product. Although such alternative permanent magnets may lack in some magnetic performance, for example, energy product, compared to Nd₂Fe₁₄B, they can serve as substitutes where applicable. Even if the substitutes contain other materials deemed critical, e.g. cobalt, they will enable source diversification which is a good strategy for mitigating materials supplies problems. Such substitution approach is currently applied towards mitigating supply risks for cobalt in batteries [15,16].

Nd₂Co₁₄B and Nd₂Fe₁₄B can form a continuous solid solution in Nd₂Co_xFe_14−xB [17]. It is reported that Curie temperature increases monotonically with Co content in Nd₂Co_xFe_14−xB [2,18] and reaches 995 K at x = 14 [17] (i.e., when all Fe are replaced by Co). However the saturation magnetization and anisotropy field of Nd₂Co₁₄B are much lower than those of Nd₂Fe₁₄B [19]. By substituting Nd with Pr to form Pr₂Co₁₄B, an anisotropy field (~75 T at 4.2 K) can be obtained. This is more than twice that of Nd₂Co₁₄B (30 T at 4.2 K) obtained by extrapolation of magnetization data in aligned samples [20]. At room temperature, an anisotropy field higher than 12.6 T was reported for Pr₂Co₁₄B by Shimao et al. [21]. This high anisotropy field indicates the possibility of developing Pr₂Co₁₄B-based magnets with high coercivity. For example, coercivity of 15–25 kOe were reported by Fuerst et al. [22] and Christodoulou [19] in melt-spun ribbons. In the Nd-Fe-B system, it is known that addition of TiC refines grain sizes and leads to enhanced coercivity [23,24]. Such effect of refractory carbides on the microstructure of Pr₂Co₁₄B and the corresponding magnetic properties, is not well known. As a result, this study evaluates the relationship between microstructure and magnetic properties of Pr₂Co₁₄B due to grain refinement with TiC.

2. Materials and Method

Ingots of Pr₂Co₁₄B, with and without TiC, were made by arc melting of the constituent elements in Ar atmosphere. All samples were synthesized in the laboratory. Ribbons, with average thickness of ~30 μm, were made by melting the ingots contained in quartz crucibles under high purity He gas and ejecting the melt onto a single Cu wheel at a wheel speed of 30 m/s. The Pr₂Co₁₄B samples without TiC are hereafter referred to as PrCoB and the samples with 3 wt.% TiC are hereafter referred to as PrCoB-TiC. Some of the as-spun PrCoB and PrCoB-TiC were wrapped with Ta foil, sealed in a quartz tube filled with high purity Ar and annealed at 1020 K for 15 min. The samples were quenched in water to ambient conditions. Powders derived from grinding the ribbons were used for crystal structure analysis using Co K_α radiation for X-ray diffractometry (XRD). The phases were determined by comparing the XRD patterns with powder diffraction files (PDF cards). Microstructural characterization was performed on a probe aberration corrected FEI Titan Themis scanning transmission electron microscope (TEM) with Super-X energy dispersive X-ray (EDS) detector. The TEM samples were prepared by mechanical wedge polishing followed by argon ion milling with liquid N₂ cold stage.

Simultaneous thermal analyses (STA) including thermogravimetry analysis (TGA) and differential scanning calorimetry (DSC) were performed at a heating rate of 20 K/min to determine the Curie temperature and spin reorientation temperatures of the alloys. A pair of permanent magnets attached to the STA system creates a magnetic field gradient on the sample. Changes in the permeability reflecting changes in magnetic anisotropy at the spin reorientation temperature and magnetic order at the Curie temperature, result in an apparent mass changes [25]. Magnetization was measured using a vibrating sample magnetometer up to a maximum applied magnetic field of 90 kOe. In order to measure the anisotropy field, ingots were sealed in quartz tube, annealed at 1270 K for 7 days and slowly cooled in furnace to ambient conditions. The annealed ingots were pulverized by ball milling and sieved to grain sizes <20 µm. The powders were aligned in a 1.5 T magnetic field and cured in an epoxy resin with the magnetic field applied. Magnetization was measured with the applied field both parallel and perpendicular to the alignment direction in order to determine the anisotropy field. The anisotropy field H_A was derived from the intersection points of the extrapolated magnetization plots in parallel and perpendicular directions.

3. Results and Discussion

Figure 1a,b show the TGA and DSC plots of the as-spun PrCoB-TiC and PrCoB ribbons, respectively. In Figure 1a, the absence of a spin-reorientation peak in the TGA curve indicates that the Pr₂Co₁₄B phase is mostly in the amorphous state in the as-spun PrCoB-TiC. In Figure 1b, the peak corresponding to the spin-reorientation temperature is seen for the as-spun PrCoB at 670 K, indicating the formation of mostly crystalline materials. The crystallization activity of the PrCoB-TiC at ~930 K (Figure 1a) is less pronounced compared to that of PrCoB at ~920 K in Figure 1b. This further confirms that the sample PrCoB-TiC has more amorphous fraction compared PrCoB-TiC. The ~10 K difference in the crystallization temperatures can be attributed to the grain refinement effect of TiC leading to delayed crystallization.

From the TGA plots, Curie temperatures of 1038 K and 1013 K were obtained for the as-spun PrCoB-TiC and PrCoB samples, respectively. The obtained Curie temperature is higher than previously reported for the Pr-Co-B system and may be related to compositionally non-equilibrium state in the as-spun ribbons. This was verified by performing TGA measurement in magnetic field on as-spun ribbons of PrCoB annealed for seven days at 1270 K. The results (Figure 2) shows a Curie temperature of 989 K which is between the values reported by Pedziwiatr (986 K) [26] and Shimao (998 K) [21]. It also shows a spin-reorientation at 681 K.

Figure 3a,b show the XRD results of the as-spun ribbons of the PrCoB-TiC and PrCoB samples, respectively. As already seen from the thermal analyses data, the XRD pattern confirms that the fraction of the amorphous content is higher in the PrCoB-TiC, compared to PrCoB. The significantly higher amorphous content in the PrCoB-TiC indicates the microstructural refinement effect of TiC in Pr₂Co₁₄B, similar to reports on Nd-Fe-B based alloys [23,24].

The transmission electron micrographs of the as-spun samples are shown in Figure 4. Micrographs Figure 4a–c correspond to the microstructure of the as-spun PrCoB-TiC samples from wheel side to free side, while Figure 4d–f correspond to the as-spun PrCoB samples, also from wheel side to free side. The diffuse diffraction rings in the as-spun PrCoB-TiC samples indicate that they are predominantly amorphous. This agrees with the absence of a crystallization peak in Figure 1a and the XRD pattern in Figure 3a. The bright contrasts in the PrCoB-TiC samples are Ti-rich phase (TiC). In contrast, the as-spun PrCoB samples show obvious signs of crystallinity with grain size in the range of 20–200 nm, from wheel side to free side. The crystalline nature of PrCoB agrees with the observation of crystallization peak in Figure 1b and the sharp X-ray diffraction patterns in Figure 3b. The grain sizes are also non-uniform from the wheel to the free side of the melt-spun ribbons. These results show the ability to refine the microstructure of Pr-Co-B systems with TiC. The microstructural refinement is related to the reduced growth rate of the solidification front in the samples with TiC which provides the ability to control the microstructure and optimize magnetic properties. Similar effect has been previously reported for Nd-Fe-B systems [27,28]. In the Nd-Fe-B systems, it was shown that over quenched conditions can be achieved at slower wheel speeds with TiC addition but at higher wheel speeds without TiC [24].

In Figure 5, micrographs Figure 5a–c correspond to the microstructure of the heat treated PrCoB-TiC samples from wheel side to free side, while Figure 5d–f correspond to the heat treated PrCoB samples, also from wheel side to free side. It can be seen that TiC enables greater control of the microstructural development in large volume fraction of Pr₂Co₁₄B ribbons during heat treatment. PrCoB-TiC samples (Figure 5a–c) show relatively uniform grain sizes from wheel to free sides of the ribbon, with an average grain size of 100 nm. Under the same heat treatment conditions, the morphology of the grains in PrCoB samples (Figure 5d–f) changes significantly from the wheel to free side. Closer to wheel side, fine grain sizes (~25 nm) with high density of stacking faults were observed. On the free side of the ribbon however, the PrCoB samples show an average grain size of 200 nm with almost no stacking faults, compared with the wheel side of the ribbons. The TiC particles in PrCoB-TiC can be seen as bright contrasts in Figure 5a–c with grain size of ~20 nm. Most of TiC can be seen near grain boundaries and triple junctions. The effect of TiC on limiting grain growth can be easily seen by comparing Figure 5c,f.

The elemental mapping of the PrCoB-TiC in Figure 6 shows that the TiC and Pr₂Co₁₄B co-exist with, at least, two other phases in the microstructure. The phase labelled “rare-earth rich phase” has Pr to Co ratio of 1:1.7 which is closer in composition to PrCo₂. A phase with composition PrCo₂ was also observed closer to the wheel side in the PrCoB sample. The phase labelled “rare-earth lean phase” is predominantly Co with some Ti at a ratio of 8.4:1. Like in the Nd-Fe-B system [27], there was no Ti or C found in the matrix phase which indicates that with both elements present, they have significantly low solubility in Pr₂Co₁₄B. With only Ti, it was found that TiB₂ formed in addition to Nd-Fe-B [27]. The formation of extra phases in the present study may indicate that less than 3 wt.% of TiC is needed for optimum grain refinement. We intend to investigate the ability to use less refractory carbide in the future because such will be very useful in minimizing the reduction in magnetization due to presence of non-magnetic phases.

Figure 7 shows the room temperature magnetic hysteresis loops of the as-spun PrCoB-TiC and PrCoB samples. Typical of amorphous hard magnetic materials, the as-spun PrCoB-TiC sample showed a hysteresis loop corresponding to soft magnetic materials. This agrees with the observed microstructure and crystal structure of the sample. The as-spun PrCoB shows a hysteresis loop typical of hard magnetic materials containing significantly more crystalline than amorphous fraction, which also corresponds to previously discussed results. The observation that PrCoB-TiC approaches saturation faster than PrCoB is also indicative of higher amount of crystalline hard magnetic materials in the as-spun PrCoB. The higher magnetization at 30 kOe obtained for the as-spun PrCoB, compared to the as-spun PrCoB-TiC is likely due to the dilution of the magnetization of the Pr₂Co₁₄B by non-magnetic TiC phase.

The magnetic hysteresis loops of the samples annealed at 1020 K (above the crystallization temperature) for 15 min were measured up to 90 kOe and are shown in Figure 8. Hard magnetic properties developed upon annealing resulting in higher coercivity materials, compared to the as-spun samples. Nearly twice the coercivity of the annealed PrCoB was obtained for the annealed PrCoB-TiC samples. Since heat treatment is necessary to fully develop the hard magnetic microstructure, the grain refining effect of TiC provides a greater ability to control the microstructure during such heat treatments which leads to enhanced hard magnetic properties. TiC altered the growth kinetics and limited the formation of abnormal grain growth so that more uniform microstructure can be obtained over a volume of the materials. In Nd-Fe-B system, average grain size of ~75 nm and 200 nm were reported for samples with and without TiC, respectively [23,24]. In the present study, the average grain size was ~100 nm for the PrCoB-TiC sample. The finer grains provide greater pinning sites to magnetic domains which leads to the observed increase in coercivity. The difference in magnetization between both samples is greater in the annealed samples, compared to the as-spun. It is thus likely that the phases labelled as “rare-earth-rich phase” and “rare-earth-lean phase” in Figure 6 further reduced magnetization in the sample with TiC.

It can be seen in Figure 8 that the magnetization for both samples did not approach saturation at the maximum applied magnetic field of 90 kOe, indicative of high magnetocrystalline anisotropy in Pr-Co-B systems. An anisotropy field of ~158 kOe was estimated for the samples using the method described in the materials and method section. To verify the anisotropy field values, the first quadrant demagnetization curve of both samples in Figure 8 was applied to an approach used by McCallum [29] based on the Stoner-Wohlfarth model, which resulted in a comparable value of ~160 kOe. The observation that both samples have comparable anisotropy fields can also be seen from the similarity in the slopes of their high magnetic field magnetization plots in Figure 8. The anisotropy field obtained for these samples is higher than previously reported for materials based on Pr-Co-B (97–125 kOe) [21,30]. One may expect that such comparable anisotropy field would correspond to comparable coercivity for both PrCoB and PrCoB-TiC, which was not the case. This indicates that the higher coercivity obtained for the PrCoB-TiC is mainly due to microstructural refinement effect of TiC and highlights the fact that coercivity is an extrinsic magnetic property. The step near remanence in PrCoB-TiC is likely due to the magnetic properties of the predominantly cobalt phase seen in the TEM since the other secondary phase, PrCo₂, is paramagnetic above ~40 K [31].

Extrapolating the magnetization to saturation using the aforementioned model [30] yields an estimated saturation magnetization of 107 emu/g for PrCoB-TiC and 120 emu/g for PrCoB. This indicates that under anisotropic conditions, it is possible to develop permanent magnets with maximum energy product of ~30 MGOe with these alloys. It suggests the ability to develop permanent magnets comparable in properties with high performance Sm-Co based magnets. However Pr-Co-B has spin-reorientation at 680 K which is below its Curie temperature of 990 K. At spin reorientation, magnetization switches from easy axis to easy plane which diminishes magnetocrystalline anisotropy and coercivity. Substituting ~36% or more of the Co concentration with Fe results in materials without spin-reorientation with Curie temperature of ~942 K [32]. Such substitution also results in increased magnetization although coercivity is reduced [33]. Addition of TiC can therefore limit the decrease in coercivity due to Fe-substitution in Pr-Co-B whilst resulting in materials without spin-reorientation. If smaller amount of TiC is sufficient for optimum microstructural refinement, magnetization can further be increased by reducing the amount of TiC used; hence improving the potential energy product of the alloys.

4. Conclusions

This study shows that the microstructure of melt-spun Pr₂Co₁₄B is greatly refined by the addition of TiC. No obvious changes in the magnetic anisotropy field was observed, indicating that the observed significant increase in coercivity (nearly twice) is due to microstructural optimization effects of the TiC. Magnetization decreased due to the addition of non-magnetic TiC and the formation of low magnetic moment secondary PrCo₂ and Co-rich phases. However, the remanence is comparable for samples with and without TiC. Also, neither Ti nor C were found in the 2:14-1 phase. Formation of secondary phase was discussed as an indication of the opportunity to reduce the TiC content whilst achieving significant increase in coercivity and comparable values of remanence and saturation magnetization.