Processing of NdFeB magnetic material

Abstract

A method of processing NdFeB magnetic powder comprises: providing a source of hydrogenated NdFeB powder (101, 102, 103); feeding said powder into an inlet of a cyclone separator (104); separating the powder into an overflow enriched in Nd-rich grain boundary phase and an underflow enriched in Nd.sub.xFe.sub.yBH.sub.z matrix phase particles (106); optionally feeding the underflow back into the inlet of the cyclone separator whereby to further enrich the underflow in the Nd.sub.xFe.sub.yBH.sub.z matrix phase particles (108a); and collecting the underflow (108).

Claims

1. A method of processing NdFeB magnetic powder to form an NdFeB magnet, the method comprising: providing a source of sintered NdFeB magnet material comprising Nd-oxide; performing hydrogenation decrepitation on the sintered NdFeB magnet material to form hydrogenated NdFeB powder comprising Nd-oxide; feeding the hydrogenated NdFeB powder into an inlet of a cyclone separator; separating the powder into an overflow enriched in Nd-rich grain boundary phase and the Nd-oxide, and an underflow enriched in Nd.sub.xFe.sub.yBH.sub.z matrix phase particles; collecting the underflow; blending the underflow with fresh Nd particles to produce an Nd-enriched powder; and sintering and magnetising the Nd-enriched powder to form a NdFeB magnet.

2. The method of claim 1 wherein the underflow is fed into the inlet of the cyclone separator sufficient times until the underflow is at least 95 wt. % Nd.sub.xFe.sub.yBH.sub.z matrix phase particles.

3. The method of claim 1 wherein the Nd particles are elemental Nd, Nd-hydride or an alloy of Nd.

4. The method of claim 1 wherein the cyclone separator is a hydrocyclone separator and the hydrogenated NdFeB powder is formed into an aqueous slurry before being fed into the inlet of the separator.

5. The method of claim 4 wherein the collected underflow is dried.

6. The method of claim 1 further comprising feeding the underflow back into the inlet of the cyclone separator to further enrich the underflow in the Nd.sub.xFe.sub.yBH.sub.z matrix phase particles.

7. The method of claim 1 wherein the source of sintered NdFeB magnet is Hard Disk Drives (HDD).

8. The method of claim 1 wherein the hydrogenated NdFeB powder is demagnetized.

9. The method of claim 1 wherein the source of sintered NdFeB magnet comprises a matrix of Nd.sub.2Fe.sub.14B matrix surrounding an Nd-rich grain boundary.

10. The method of claim 9 wherein when performing the hydrogenation decrepitation on the sintered NdFeB magnet source, Nd.sub.2Fe.sub.14B.sub.x matrix phase particles and NdH.sub.2.7 grain boundary phase particles are formed, the Nd.sub.2Fe.sub.14B.sub.x matrix phase particles being larger than the NdH.sub.2.7 grain boundary phase particles.

11. The method of claim 10 wherein the separating the hydrogenated NdFeB powder separates the matrix phase particles from the grain boundary phase particles.

12. The method of claim 11 wherein the grain boundary phase particles have a higher oxygen content than the matrix phase particles.

13. The method of claim 1 wherein the hydrogenated NdFeB powder comprises NdH.sub.2.7, Nd.sub.2Fe.sub.14BH.sub.x, and the Nd-oxide.

14. The method of claim 13 wherein the NdFeB powder comprises Nd.sub.2Fe.sub.14BH.sub.x matrix phase particles, NdH.sub.2.7 grain boundary phase particles, and the Nd-oxide, wherein the NdH.sub.2.7 grain boundary phase particles and the Nd-oxide are smaller than the Nd.sub.2Fe.sub.14BH.sub.x matrix phase particles.

15. The method of claim 14 wherein the Nd.sub.2Fe.sub.14BH.sub.x matrix phase particles are greater than 5 microns in size and the NdH.sub.2.7 grain boundary phase particles, and the Nd-oxide are smaller than 1 micron in size.

16. The method of claim 1 wherein the Nd-rich grain boundary phase comprises oxygen-rich portions.

17. The method of claim 16 wherein the oxygen-rich portions are particles smaller than 1 micron.

18. The method of claim 1 wherein the oxygen-rich portions are removed using the cyclone separator and replaced with the fresh Nd particles to produce the Nd-enriched powder.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 is a flow diagram of a method for recycling NdFeB magnets according to an embodiment of the present invention.

DETAILED DESCRIPTION

(2) Referring to FIG. 1, the NdFeB magnet starting material is obtained by breaking apart waste electronics and electrical equipment 100. The starting material is subjected to hydrogen decrepitation 102 to form a powder comprising larger Nd.sub.xFe.sub.yBH.sub.z matrix phase particles and smaller Nd-rich grain boundary phase particles. The powder is separated from the other components by sieving 103.

(3) The decrepitated powder is fed into a cyclone separator 104, wherein the powder is separated 106 into an overflow comprising mostly Nd-rich grain boundary phase particles and an underflow comprising mostly Nd.sub.xFe.sub.yBH.sub.z matrix phase particles.

(4) The underflow is collected 108, and may be optionally fed back into the cyclone separator 108a in order to improve the separation yield. After one or more cycles through the cyclone, the underflow is dried 110 to obtain a powder consisting almost entirely of Nd.sub.xFe.sub.yBH.sub.z matrix phase particles. The Nd.sub.xFe.sub.yBH.sub.z powder is blended with fresh neodymium hydride 112, and sintered and magnetised 114 to form a recycled NdFeB magnet.

(5) The starting material was in the form of Ni coated arc segments obtained from scrap VCM magnets from the former Philips factory, in Southport UK. The composition, minor additions excluded, was Nd.sub.11.61Dy.sub.0.53Pr.sub.1.59Fe.sub.77.89Al.sub.0.75Co.sub.1.44Cu.sub.0.09B.sub.6.01 and was determined by inductively coupled plasma optical emission spectrometry (ICP-OES). The Ni coating layer was removed from the surface of the VCMs by scoring and peeling. The magnets were then HD processed at 4 bar for 1 h. The powder remained non-milled.

(6) In a specific embodiment of the present invention, Nd H.sub.2.7 particles are separated from matrix phase particles by use of a hydrocyclone separator. 1 kg of non-milled powder was mixed with water to form a slurry, at a 10 wt % ratio, and pumped tangentially into a cone-shaped separator, which creates a vortex flow as the water travels helically downward. Larger particles are forced radially outward to the wall of the vessel by centrifugal force and descend to the bottom of the vessel due to friction and gravity. Smaller particles tend to spiral upwards and exit at the top of the vessel. This produced a small particle sized overflow stream (OF) and a large particle size underflow stream (UF). A total of three hydrocyclone separation experiments were performed on each input feed. After the first hydrocyclone separation the UF was the input for the next hydrocyclone separation stage. After separation, both fractions were filtered and then dried at 80 C. in air. All produced fractions were assessed by x-ray fluorescence (XRF), performed in a Philips XRF PW2400; by ICP-OES, analysed in an ICP-OES Varian 720ES; and by x-ray diffraction (XRD), performed in a PANalytical Empyrean. The error associated with XRF, ICP-OES and XRD was, respectively, 0.10 wt %, 0.01 wt % and 0.50 wt %.

(7) The OF stream which exits at the top of the vessel comprises mainly grain boundary phase particles (NdH.sub.2.7 and Nd-oxides/hydroxides), while the UF stream which exits at the bottom of the vessel comprises mainly matrix phase particles (Nd.sub.2Fe.sub.14BH.sub.x).

(8) Fresh NdH.sub.2.7 was produced by roller ball milling for 20 h and sieving through a 45 m sieve. The Nd.sub.2.7 was blended with the HD VCM hydrocyclone separated powder at 5 at % by passing them together through a 45 m sieve.

(9) All sintered magnets produced during this work were made via the HD powder metallurgy route, at optimum conditions [R. S. Mottram, A. Kianvash, I. R. Harris, The use of metal hydrides in power blending for the production of NdFeB-type magnets, J. Alloy. Compd. 283 (1999) 282-288 and R. N. Faria, J. S. Abell, I. R. Harris, High coercivity sintered PrFeBCu magnets using the hydrogen decrepitation process, J. Alloy. Compd. 177 (1991) 311-321]. 15 g of powder of the specified sample was aligned at 9 T in a capacitor discharge pulse magnetiser prior to isostatic pressing at 60 MPa. The green compact was then sintered at 1060 C. for 1 h in vacuum. After sintering the samples were pulse magnetised at 9 T prior to measurement of the magnetic properties in a Permagraph von Dr Steingroever. Density measurements were performed using a standard Archimedes displacement method and calculated according to international standards [American Society for Testing and Materials B 962, Standard Test Methods for Density of Compacted or Sintered Powder Metallurgy (PM) Products Using Archimedes' Principle (2008)]. A JEOL 6060 scanning electron microscope (SEM) was used for back-scattered electron (BSE) micrographs of sintered NdFeB magnets. The errors associated with density () and relative density (.sub.rel) was 0.05 g cm.sup.3 and 0.1%. The errors of the coercivity (H.sub.c), remanence (B.sub.r) and maximum energy product ((BH).sub.max) were calculated to be 5 kA m.sup.1, 5 mT and 5 kJ m.sup.3 respectively.

(10) In another embodiment (not shown), an inert gas cyclone may be used instead of a hydrocyclone.

(11) Results

(12) BSE SEM images of UF and OF powder after cyclic hydrocyclone separation were obtained using non-milled hydrogen decrepitated (HD) VCM powder as input feed (not shown). It was immediately evident that in all cases there is a greater fraction of smaller particles <2 m in the OF stream. Using EDX analysis this was shown to be Nd-rich and is thought to be the grain boundary phase (GBP). However, some larger particles can still be observed in the OF which were shown by EDX analysis to relate to the Nd.sub.2Fe.sub.14B hydride matrix phase.

(13) The UF fraction contains particles in the range of 10-20 m, which relate to the Nd.sub.2Fe.sub.14B hydride matrix phase but with smaller particles on the surface. This is thought to be due to triboelectric charges that still exist between the GBP and the larger Nd.sub.2Fe.sub.14B hydride particles. It is interesting to note that the particle size is about the same size as the original grains of Nd.sub.2Fe.sub.14B in the sintered magnet. Therefore the hydrocyclone appears to be breaking apart the hydrogen processed powder into near single crystal particles.

(14) As the aim was to purify the Nd.sub.2Fe.sub.14B matrix phase particles, then only the results of the UF fraction are shown here. XRF, ICP-OES and XRD analysis are presented, respectively, in Tables 1, 2 and 3.

(15) TABLE-US-00001 TABLE 1 XRF results expressed in wt % from the UF after three cycles of hydrocyclone separation. Numbers refer to the cycle of hydrocyclone. Sample Fe Nd Dy REE Starting material 61.5 34.9 1.4 36.7 Non-milled Air 61.0 35.2 1.4 37.1 HD UF-1 61.6 34.7 1.4 36.6 UF-2 65.6 30.8 1.4 32.6 UF-3 66.5 30.0 1.4 31.7

(16) TABLE-US-00002 TABLE 2 ICP-OES results expressed in wt % from the UF after three cycles of hydrocyclone separation. Sample Fe Nd Dy REE Starting material 61.92 34.21 0.95 35.45 Non-milled Air 61.37 35.31 1.01 36.59 HD UF-1 63.27 33.59 0.97 34.85 UF-2 68.71 28.25 2.09 30.82 UF-3 68.89 28.06 2.09 30.63

(17) TABLE-US-00003 TABLE 3 XRD quantification results expressed in wt % from the UF after three cycles of hydrocyclone separation. Sample Nd.sub.2Fe.sub.14B Nd.sub.2O.sub.3 Nd.sub.2Fe.sub.14BH.sub.x Nd(OH).sub.3 Starting material 99.3 0.7 n/d n/d Non-milled HD Air 0.5 1.2 88.9 9.1 UF-1 n/d n/d 91.3 8.4 UF-2 n/d n/d 96.2 3.8 UF-3 n/d n/d 97.1 2.8 n/dNot detected

(18) The main aim of subjecting the HD powder to hydrocyclone separation was to extract all the GBP in order to have a clean fraction based on the Nd.sub.2Fe.sub.14B hydride matrix phase alone. It is clear from the ICP-OES and XRF results that, although there are some subtle differences in the Nd ratios between the different analytical techniques, there is a clear trend showing that the Nd ratio is falling in the UF with increasing number of cycles on the hydrocyclone. This is clear evidence that the hydrocyclone is capable of stripping out the small Nd-rich GBP particles.

(19) The stoichiometric Nd.sub.2Fe.sub.14B composition should have an Fe/ZREE ratio of around 2.71 by wt %. The Fe/ZREE ratio, using ICP-OES data in Table 2, increased from 1.68 in the input HD material to 1.81, 2.23 and 2.25 after the 1.sup.st, 2.sup.nd and 3.sup.rd cycle of hydrocyclone separation.

(20) It is interesting to note from the XRD results in Table 3 that the Nd.sub.2Fe.sub.14B.sub.x matrix particles are still intact, even after exposure to water in the cyclone and after heating in air at 80 C. This powder was also exposed to air for over 4 months prior to measurement on the XRD. No evidence of -Fe could be observed in the XRD traces which would be a sign of the matrix grains breaking down. The XRD results showed that the Nd-rich phase had transformed into Nd(OH).sub.3 on exposure to water in the hydrocyclone.

(21) The XRD data showed an increase in Nd.sub.2Fe.sub.14B matrix hydride and a subsequent decrease in neodymium hydroxide, with increasing number of cycles of hydrocyclone separation, yielding a maximum of 68.89 wt % after the 3.sup.rd cycle, close to the 72.3 wt % present in the stoichiometric Nd.sub.2Fe.sub.14B. It can therefore be stated that both phases have shown significant separation during hydrocyclone processing.

(22) The hydrocyclone separated HD powder (3 cycles) and a powder blend containing the same powder with an additional 5 at % NdH.sub.2.7 was re-sintered into new NdFeB magnets. The magnetic properties of the resultant magnets are shown in Table 4.

(23) TABLE-US-00004 TABLE 4 Properties of starting material and recycle sintered NdFeB-based magnets blended with neodymium hydride. .sub.rel H.sub.c B.sub.r (BH).sub.max NdH.sub.2.7 [at %] [g cm.sup.3] [%] [kA m.sup.1] [mT] [kJ m.sup.3] Starting material 7.59 99.98 1,191 1,140 242 0 at % 5.03 66.48 0 0 0 5 at % 6.98 93.06 545 934 146 is the density of the material; .sub.rel is the density relative to the starting material; H.sub.c is the coercivity; B.sub.r is the remanence; and (BH).sub.max is the maximum energy product.

(24) It was evident from the BSE SEM images (not shown) that only a few light particles of the Nd-rich phase could be observed in the 0 at % NdH.sub.2.7 sample, which was in agreement with the ICP-OES in Table 2. This confirms that separation was successfully performed but that there is still room for improvement in order to fully extract all of the remaining GBP particles bound to the Nd.sub.2Fe.sub.14B.sub.x surfaces. Table 4 shows that recycled re-sintered NdFeB magnets without any NdH.sub.2.7 additions, exhibited a much lower density than the starting material (5.03 g cm.sup.3). This equates to a relative density of 66.48%, when compared with the results from the starting EOL NdFeB sintered magnet (Table 2). This is not surprising given the very small amounts GBP left in the magnet and the fact that this has been converted to Nd(OH).sub.3. Therefore there was insufficient liquid phase during sintering. It was not possible to measure the magnetic properties as the magnet broke apart after pulse magnetising.

(25) 5 at % NdH.sub.2.7 was blended, into the hydrocyclone HD NdFeB powder (after 3 cycles), in an attempt to enhance liquid phase sintering and hence to obtain a fully dense recycled magnet. In this case, the density and relative density of the resultant re-sintered magnet increased to 6.98 g cm.sup.3 and 93.06%, respectively. The recycled magnet still showed significant porosity and it is clear that further NdH.sub.2.7 is required to recover the full density for this sample.

(26) The recycled re-sintered magnet made from hydrocyclone separated powder blended with 5 at % NdH.sub.2.7 demonstrated an intrinsic coercivity, remanence and maximum energy product of 545 kA m.sup.1, 934 mT and 146 kJ m.sup.3 respectively. The decrease in coercivity is likely to be as a result of the porosity observed in the magnet, which will act as sites for reverse domains to nucleate.

(27) It is particularly interesting to note that the remanence obtained in this magnet was only 18% lower than the starting material despite a density of only 6.98 g cm.sup.3. This implies that the matrix phase is intact after hydrocyclone separation which is confirmed by the XRD results shown in Table 3.

(28) When further cyclone separations are applied to the HD powder and when >5 at % NdH.sub.2.7 is added to the hydrocyclone separated powder then it should be possible to recover all of the remanence and coercivity.