The production and properties of sintered Nd permanent magnets

The production and properties of sintered Nd permanent magnets

The production and properties of sintered Nd permanent magnets 7.0 7 Introduction As discussed at several points in the preceding chapters, there a...

4MB Sizes 0 Downloads 19 Views

The production and properties of sintered Nd permanent magnets 7.0

7

Introduction

As discussed at several points in the preceding chapters, there are two major families of NdFeB permanent magnets. These include magnets produced from rapidly solidified powder, specifically isotropic-bonded Nd magnets and anisotropic hot-deformed NdFeB magnets, and sintered Nd magnets. Although sintered Nd permanent magnets are not rapidly solidified, a review of the production and properties of this important family of permanent magnets is provided here to provide the reader with the means of comparing the two different families of magnets. Sintered Nd magnets were discovered at the Sumitomo Special Metals Corporation in 1982 (Sagawa et al., 1984a,b,c) and are produced by the so-called orient-press-sinter (OPS) process, which is the same process used to produce sintered Sm-Co magnets. A number of patents were filed, including Sagawa et al., US Patent 4,792,368 (issued 1988) and Yamamoto et al., US Patent 4,601,875 (issued 1986). Today, sintered Nd magnets are by far the largest market of rare earth-transition metal permanent magnets and are used in a wide variety of applications including voice coil motors for hard disk drives, magnetic resonance imaging devices, windmill generators, and many different motors. It is reported that the drive motors in a modern hybrid vehicle uses 28 kg of sintered Nd magnets.

7.1

Sintered Nd production process

Although the properties of HP-3 (see Chapter 6: Hot-deformed NdFeB permanent magnets) magnets are fairly similar to those obtained for sintered Nd magnets, the method by which the anisotropy is obtained and the resulting microstructure are significantly different. To provide the reader a comparison, the process for producing sintered Nd magnets is covered here in some detail. The several steps in the process are shown in Fig. 7.1 and consist fundamentally of converting an isotropic alloy into a powder and then reassembling the powder into an anisotropic, fully dense intermetallic compact by compacting in an alignment field followed by sintering. The OPS process for producing sintered Nd magnets has been extensively investigated and review articles have been written by, among others, Ormerod (1985, 1989), Sagawa et al. (1987), Buschow (1988), and Burzo and Kirchmeyer (1989).

7.1.1 Alloy preparation The first step in the process of producing a sintered Nd magnet is the preparation of the starting alloy. Unlike the melt-spinning process, where the microstructure of Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets. DOI: http://dx.doi.org/10.1016/B978-0-08-102225-2.00007-7 Copyright © 2018 Elsevier Ltd. All rights reserved.

298

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

Alloy production Crush alloy Jet mill powder Align and press powder Sinter green compact Finish grind Apply coating

Alloy is strip cast to produce a single phase microstructure Strip cast alloy is hydrogen decrepitated Crushed alloy is jet milled to a fine 1–5 μm powder Powder is aligned and while being compacted Green compact is sintered and heat treated under Ar atmosphere All sintered magnets must be ground to final dimension Ni-Cu-Ni coating is applied to protect magnet from corrosion

Figure 7.1 Steps in the production of sintered Nd magnets.

starting alloy is unimportant, the alloy used for producing sintered magnets is critically important. Fig. 7.2 shows an isothermal section of the Nd-Fe-B phase diagram at 1000 C, which shows the Nd2Fe14B intermetallic phase (Φ) at the intersection of the isotherms in the lower left-hand corner (Schneidner et al., 1987). This phase diagram shows the equilibrium phases, which form at this temperature for compositions surrounding the Nd2Fe14B composition. These include α-Fe, Fe2B, Nd2Fe17, and the Nd11hFe4B4 phase, which is indicated as η. The 1 1 h indicates that this Brich phase has a large homogeneity region. There are two important requirements when producing alloy for sintered Nd magnets. These include that the alloy be as single phase as possible and that the Nd content be as low as possible, in order to achieve the highest magnetic remanence. The major problem with producing NdFeB alloy is that the Nd2Fe14B intermetallic compound forms by a peretectic reaction, which is a reaction where a liquid and solid phase reacts to form a second phase. In this instance, Nd2Fe14B forms by peretectic reaction from a liquid 1 γ-Fe. At normal cooling rates, however, this reaction does not go to completion and the cast alloy is found to contain a mixture of the Nd2Fe14B, Nd11hFe4B4 and α-Fe: the (f.c.c.) γ-Fe converts to (b.b.c.) α-Fe at 910 C. An example of the microstructure of an as-cast NdFeB ingot is shown in Fig. 7.3 and shows the complicated mixed phase microstructure that forms under normal cooling conditions (McGuiness et al., 1989). Here the lighter phase is the Nd2Fe14B phase while the darker microstructure is a mixture of the α-Fe and Nd11hFe4B4 phases. The presence of the α-Fe, in particular, is a problem for sintered magnet producers because it is a comparatively ductile phase compared with the brittle Nd2Fe14B alloy and makes crushing and grinding of the ingot into a powder much more difficult. Another serious problem for sintered Nd magnets is that the α-Fe is magnetically soft and results in a reduction in magnetic performance in the finished magnets. Elimination of secondary phases can be accomplished by a lengthy hightemperature anneal during which the secondary phases react together to form the desired Nd2Fe14B intermetallic phase. However, this annealing or homogenization

The production and properties of sintered Nd permanent magnets

299

FeB η

40

Fe2B

B[ at%

]

30

Φ+η+ Fe2B

L + Φ+η

20

10

Φ + Fe+ Fe2B

L L+Φ

7 Nd

Fe 1

Fe

2

Φ + Fe + Fe17Nd2

L + Φ + Fe17Nd2 20

30

40

Nd [at%]

Figure 7.2 Isothermal section of the Nd-Fe-B phase diagram at 1000 C showing the equilibrium phases that form in this composition region. Here the Nd2Fe14B phase is shown as Φ (Schneidner et al., 1987).

Figure 7.3 The microstructure of as-cast NdFeB ingot showing the mixed phase microstructure that forms when the alloys is cooled under normal conditions (McGuiness et al., 1989).

300

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

process is slow and costly. The best way to circumvent the formation of the α-Fe phase is to rapidly cool the alloy through the peretectic temperature, so that formation of the α-Fe is suppressed. In addition, the rapid cooling allows alloy to be produced with lower Nd content, resulting in higher remanence in the finished magnets. In the early stages of the development of sintered Nd magnets, the ingot was cast into book molds, which are molds in which the casting cavity is slot with a cross section on the order of 1 cm or less in width, resulting in more rapid cooling of the ingot. This helped but did not solve the problem of α-Fe precipitates. The problem was finally solved by the development of strip casting, which is a process in which the molten alloy is poured into a trough-shaped tundish, which contains a long narrow slot-shaped nozzle (Bernardi et al., 1989). A sheet of molten alloy pours from the nozzle and is quenched on a rotating water cooled drum to form a continuous sheet of cast alloy. This alloy sheet is typically ,1 mm thick and, as with the melt-spun ribbon, the brittle NdFeB sheet cracks into small flacks measuring from 0.5 to 1.0 cm in diameter when cooled. The cooling rate of the molten alloy is typically 500 1000 K/s. The microstructure of a (Nd,Dy)14.1(Fe,Al)80B5.9 strip cast alloy is shown in Fig. 7.4, where (A) is an image normal to the surface of the quench rim and shows the long columnar grains that grow from nucleation sites (C) toward the free surface of the cast strip. The grains are conical shaped and have a diameter of 5 25 μm near the quench surface and 25 60 μm near the free surface of the rim and are almost completely free of α-Fe. Fig. 7.4B shows the microstructure on the top surface of the strip cast sheet and shows the complicated mosaic that delineates the tops of the conical-shaped grains that form. Each of the grains is surrounded by the Nd-rich intergranular phase, which is a necessary requirement for all NdFeB permanent magnets. The application of a higher cooling rate cannot be employed because this would produce smaller, polycrystalline grains that are unsuitable for the sintered Nd process. The development of strip casting was a major development in the production of sintered Nd magnets and resulted in substantial improvements in magnetic properties. However, it is of no practical use for the production of melt-spun magnetic powder for bonded Nd magnets. This is because the starting ingot or alloy is remelted during the process and its precursor microstructure is irrelevant to the final melt-spun product. As discussed in Chapter 3, The properties of melt-spun NdFeB alloys, and Chapter 4, Production of rapidly solidified NdFeB magnetic powder, the cooling rate of the melt-spinning process can be as high as 105 K/s, a cooling rate which completely suppresses the formation of any α-Fe in the final ribbon microstructure.

7.1.2 Powder preparation The next step in the process is to grind the strip cast alloy into a fine powder with a particle size ranging from 1 to 10 μm, but preferably 1 to 5 μm, and a narrow particle size distribution. Sintered magnets are produced by aligning the powder in a magnetic field while it is being compacted and, it goes without saying, that any particles that contain more than one grain will not completely align during this pressing operation. Therefore, an essential requirement is that virtually all of the grains

The production and properties of sintered Nd permanent magnets

301

(A)

(B)

Figure 7.4 (A) Optical micrograph of a cross section of strip cast alloy showing the columnar grains of the NdFeB magnetic phase which form at the nucleation centers (C) with their major axis normal to the quench surface. (B) Optical micrograph of the top surface of the strip cast alloy (Bernardi et al., 1989).

be single crystal particles. This is why the strip cast alloy must be quenched at an intermediate rate, which produces Nd2Fe14B grains that are somewhat larger than the desired average grain size of the powder. Crushing the powder is carried out in two steps. In the first step, the strip cast alloy is hydrogen decrepitated by simply subjecting the alloy to a hydrogen atmosphere at room temperature for, typically ,1 hour (Harris et al., 1985, 1987; McGuiness et al., 1989). During this process step, hydrogen is absorbed into the Nd-rich intergranular phase, which expands and causes the alloy to fracture and decrepitate into a powder having an particle size of ranging from 200 to 1000 μm. At this temperature, the hydrogen does not absorb into the Nd2Fe14B intergranular phase. As a consequence of fracturing the intergranular phase, an amount of Nd-rich fines are also produced. Because of their high Nd content, these fine particles are much more easily oxidized than the majority Nd2Fe14B grains and serve to getter or react with the oxygen during this process step. A scanning electron micrograph (SEM) image showing these fines on the

302

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

Figure 7.5 SEM image of the surface of a hydrogen decrepitated particle showing the fine Nd- and oxygen-rich particles that form when the intergranular phase is fractured (McGuiness et al., 1989).

surface of a hydrogen-decrepitated NdFeB ingot is shown in Fig. 7.5. As discussed later, an important part of the processing of sintered magnets is the removal of at least some of these fine particles. The next step in the process is to further grind this coarse powder into a much finer powder using a jet mill. Jet milling is one of the most critical steps, possibly the most critical step, in the manufacture of sintered Nd magnets for the reasons stated. A schematic drawing showing the operation of a fluidized bed jet mill is shown in Fig. 7.6. This grinder operates by feeding particles into the crushing chamber using a screw feeder. Inside the crusher the particles are entrained in two or more high-velocity gas streams (usually three), which crush the powder by impacting the grains violently against each other. The fluidized air flow then carries the crushed particles upward into a turbo selector, which is an integral part of the grinder. The turbo selector has a squirrel cage rotor, which rotates at high speed and allows particles smaller than a certain size to exit from the grinding chamber and rejects particles larger than a certain size back into the grinding chamber. As depicted in the drawing, these larger grains are again entrained in the highvelocity gas streams and further ground. The gas stream containing the fine particles exits the grinder and flows into a cyclone, which separates and collects the fine particles from the air stream, while at the same time removing the extremely fine particles. In a cyclone separator, the particles and gas enter the chamber tangentially where the larger particles, due to their higher inertia, move outward to the walls of the chamber and fall down into a collection chamber at the bottom. At the same time, the finest particles, having lower inertia, remain entrained in the gas stream and exit at the top of the cyclone. This stream passes into a second cyclone, where the bulk of these ultrafine particles fall to the bottom into a second collection bin. Particles that are still entrained in the gas stream are filtered from the gas

Gas containing ultrafine particles (dust) Not collected with fine powder

Classfying retor

Clean gas returned to compresser

Classified powder and gas output

Coarse powder input

High volocity gas jets

High volocity gas jets

Removed Ultrafine particles

Filter Milling chamber

Collected Fine Powder

Cyclone Figure 7.6 Operation of a fluidized bed jet mill similar to that used to grind alloy for sintered Nd magnets. Source: Courtesy John Ormerod Consulting.

304

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

stream before exiting to the atmosphere. This includes removal of these very fine (,1.0 μm) Nd- and oxygen-rich particles that are produced by the expansion and shattering of the Nd-rich intergranular phase. This adjusts the overall oxygen content of the magnet which, if too high, results in lower density and magnetic properties for the final sintered magnets. Removing the Nd-rich particles also adjusts the overall rare earth content of the magnet. The jet mill shown in Fig. 7.5 is well suited for grinding NdFeB powder since the natural operation of the grinding mill requires that a certain amount of gas must be continuously removed from the grinder and this requires that the smallest, ultrafine particles must also be removed or filtered from the exiting gas stream. Because of the large amount of gas consumed, most jet milling is carried out using N2 gas since the NdFeB powder is nonreactive with the N2 except at higher temperatures. One problem with the resulting fine-grained powder is that it is still highly susceptible to oxidation when removed from the grinding chamber. To reduce the possibility of uncontrolled oxidation, a small amount of oxygen, typically on the order of 4000 ppm, is added to the chamber during the grinding operation. This results in a very thin oxide layer that passivates the surface of the grains, prevents uncontrolled oxidation, and fires and allows the powder to be more easily handled in a plant environment. Powder that has not been given this passivation step can actually combust from the heat generated during the subsequent pressing of the powder. A lubricant is also added to the jet mill during the grinding operation to prevent or minimize scoring of the die by the highly abrasive NdFeB powder during the subsequent pressing operation. This lubricant can vary, but is typically a fatty ester diluted with a petroleum solvent. Much of the solvent volatilizes and is removed with the exiting gas stream and the remaining lubricant is distributed uniformly on the surface of crushed powder.

7.1.3 Powder alignment and compaction In sintered Nd magnets, anisotropy is achieved by aligning the individual grains so that their c axes are all aligned, while at the same time compacting the powder to a green density high enough to enable handling of the compacted part. Even with the oxygen-passivated powder, care must be taken to prevent additional oxidation of the powder during this step. In the early stages of sintered magnet development and production, attempts were made to completely enclose the pressing and handling of the green compact in an inert atmosphere of N2 or Ar. However, this proved to be too unwieldy for high-volume manufacturing. Most manufacturing today is carried out by pressing in air but care is taken to keep the relative humidity (,65%) and temperature (,27 C) in the factory as low as is practicable. This minimizes chemisorption of moisture on the powder, which can disassociate into oxygen and hydrogen and is the primary source of oxidation in all NdFeB magnets. Pressing of the magnet is very similar to that described for bonded Nd magnets in Section 5.1.3, except the tooling must include a magnetizing coil for aligning the powder or the tooling must be placed inside the field of a traditional electromagnet. The operation of the press is also similar to the production of a bonded Nd magnet

The production and properties of sintered Nd permanent magnets

305

in that the powder is automatically loaded by a shoe and pushed off the press by the next loading step, as is shown in Fig. 5.6 in Chapter 5. However, the press cycle time is much slower to allow the powder to align and to remain aligned while the powder is slowly compacted. This slower cycle time results in a higher overall capital cost for pressing sintered Nd magnet. Although pulsed magnetic fields have been used to align the powder, most manufacturing is carried out today using a DC field of between 10 and 12 kOe, which is about the maximum field that can be generated by a DC field given the space limitations. In some manufacturing plants a combination of DC and pulsed magnetic field is used. Irregardless of how the powder is aligned, once the green compact has been produced, it is necessary to demagnetize the part to enable it to be handled until the compact is sintered. This demagnetization occurs prior to removing the green compact from the die by using a decaying ac field or by demagnetizing the part to a point slightly into the third quadrant and then allowed to recoil back to a largely demagnetized state. There are two different compaction techniques that can be used to produce the green compact, transverse compaction and axial or parallel compaction. The difference between these two techniques is shown in Fig. 7.7. The direction of the powder alignment is shown by the red arrows and the motion of the punches is shown by the black arrows. In axial compaction the alignment field is parallel to the press direction, whereas in transverse compaction the alignment field is normal to the press direction. One can sense intuitively that the transverse compaction technique would align the c-axis of the individual grains more effectively than axial pressing, which would tend to mechanically misalign the grains at higher compaction levels. This is observed in practice, with transverse pressing generating as much as 5% higher magnetic remanence than axial compaction when compacted to the same green density. However, for both techniques the powder will become mechanically

H Upper punch

Aligned green compact

Upper punch

H

Aligned green compact

Electomagnet Die Fe-Co core Electomagnet

Lower punch

Die Lower punch

Parallel or axial compaction

Transverse compaction

Figure 7.7 Rendering of the two compaction techniques that are used to align and press the jet milled powder. In transverse pressing, the applied field is normal to the press direction, while in axial pressing the applied field in parallel to the press direction. The red arrows show the alignment direction of the c-axis of the Nd2Fe14B grain pressure oil.

306

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

misaligned if the compaction pressure is too high. For this reason, compaction pressures are generally in the neighborhood of only 1.5 tons/cm2 (10 tons/in.2), which results in a density of B50%. This produces a green compact that is still fairly friable but has sufficient strength to allow handling. Final densification of the magnets occurs during the sintering process. This compaction pressure is significantly lower than that used to produce compression-molded Nd magnets, which can be as high as 9 tons/cm2. In some sintered Nd factories, the powder is aligned and compacted just enough to allow handing of the green compact and addition compaction is carried out by isostatic pressing, which applies pressure on all sides of the green compact. This involves sealing the compact in a rubber or plastic shell and then compacting under high oil pressure. One advantage of isostatic pressing is that pressing occurs from all sides of the part and compacting to a higher green density does not tend to misalign the grains that have been aligned by the magnetic field. Also, the precursor green compact that is used for isostatic pressing is often aligned using a pulsed magnetic field. Some of the highest magnetic properties reported for sintered Nd magnets have been achieved using a pulsed magnetic field followed by isostatic pressing (Rodewald et al., 2000). The disadvantage of isostatic pressing is that it is somewhat unwieldy and less adaptable to high-volume production. However, isostatic pressing is often used today to manufacture big blocks of sintered Nd magnets, which are targeted for slicing and cutting into small magnets. There is a large market world-wide for very small magnets, which are impractical to produce by pressing and sintering small individual magnets.

7.1.4 Sintering and postsintering heat treatment In modern sintered Nd factory, sintering and heat treatment of the green compact are carried out in long, interconnected vacuum furnaces in which pallets of magnets are pushed from one stage to the next through vacuum doors. The pallets of green compacts are first heated in the first stage at B150 C to remove or burn off the lubricant that is added to the jet milling operation. Following this, the magnets are pushed into the actually sintering chamber, where they are typically heated at between 1050 and 1150 C for between 1 and 2 hours or for longer sintering times at lower temperatures, for example, B1050 C for 4 hours. Following this the magnets are then rapidly cooled to below 300 C using cooled Ag gas and then given a heat treatment, which can consist of a single stage treatment, for example, 625 C for 1 hour or, in some reports, a higher temperature stage, for example, 800 C for 1 hour, followed by a lower temperature heat treatment at between 500 and 600 C for 1 hour. This heat treatment that can result in a dramatic improvement is the coercively of the magnet and was recognized in some of the earliest studies of sintered Nd magnets (Sagawa et al., 1984a,b; Tokunaga et al., 1985; Hirago et al., 1985). This improvement is believed to be due to a change in the structure or morphology of the Nd-rich intergranular phase. The changes in microstructure, which accompany the anneal, are discussed below in Section 7.3.

The production and properties of sintered Nd permanent magnets

307

7.1.5 Finish grinding and coating During the sintering process, there is a significant shrinkage in the volume of the magnet. Because of this all sintered magnet must be given a finish-grinding step to adjust the shape to the specification dimensions. Needless to say, the compacted green part must be sized significantly larger than the final dimension of the magnet to allow for this shrinkage. The shrinkage that occurs is the primary reason that sintered magnets are limited to more simple shapes and the process cannot be used to produce complicated shapes like rings. As with almost all NdFeB magnets, a coating must be applied to sintered Nd magnets to product the magnet from corrosion that can occur from condensed water and other even more corrosive agents. Coating can be a spray coat like a phenolic epoxy but is usually an electroplate, which is applied in layers of Ni-Cu-Ni.

7.2

Magnetic properties of commercial sintered Nd magnets

Fig. 7.8 displays demagnetization curves of sintered Nd magnets that can be produced using the manufacturing technique described earlier. These magnets have among the highest reported for commercially available materials and include a Dyfree magnet with an energy 50 MGOe combined with an Hci of B14 kOe. Also shown is a 48 MGOe magnet with an Hci of over 22 kOe. The composition of these magnets would be in the range of those reported in Table 7.1. These magnets would have been produced using the manufacturing procedures discussed earlier, including the use of the best grade of strip cast alloy and careful handling of the powder 16

50 MGOe Dy free 8 48 MGOe contains Dy

M (KG)

12

4

0 24

20

16

12 –H (kOe)

8

4

0

Figure 7.8 Demagnetization characteristics of high end sintered Nd magnets that are commercially available.

308

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

Table 7.1 Components and typical composition range of commercial sintered Nd magnets

during the alignment and compaction steps, including keeping the factory as dry and cool as possible. Compaction would have been carried out using transverse alignment technique as shown in Fig. 7.7. The magnet with the highest coercivity would contain at least 4 wt% of Dy to increase the coercivity. There are magnet grades having Hci . 30 kOe, but these grades would contain a correspondingly higher amount of Dy. However, as was discussed briefly in Chapter 1, The development of rare earth permanent magnets, there is now a critical supply problem for Dy. For this reason, there has been extensive research aimed at reducing or eliminating the Dy in sintered Nd magnets. This research has focused on two areas including diffusion of a rare earth with higher magnetocrystalline anisotropy into the grain boundary phase surrounding the Nd2Fe14B grains. For example, Hirota et al. (2006), Watanabe et al. (2007), Li et al. (2008), Suzuki et al. (2009), and Sepehin-Amin et al. (2010) reported significant increases in the coercivity of NdFe-B sintered magnets by diffusing Dy or Tb into the critical grain boundary region using varying techniques. As discussed in Section 7.3, which deals with the microstructure and magnetization process in sintered Nd magnets, the region near the grain boundaries is the most important for the development of coercivity. All of these later studies were carried out by treating the surface with a Dy fluoride coating or by vapor deposition or sputtering of the Dy or Tb onto the surface of thin layers of Dy-free magnets, followed by heat treating. While these techniques were not practical for production, they did indicate that it might be possible to increase Hci by producing a thin shell of high anisotropy Dy or Tb around periphery of the Nd2Fe14B grains. It is not known if this Dy diffusion technique has been applied to production magnets at this time. The second technique that continues to be evaluated is to increase Hci in Dy-free sintered magnets by changing the process. For example, Kobayashi et al. (2013) have reported achieving a significant increase in Hci by using a milled powder with a finer (B1 μm) average particle size. The processing details discussed earlier for the manufacture of sintered Nd magnets are for high-end magnets. However, there is a large variation in the manufacturing process used by various manufacturers, including the alloy composition, the type of alloy used and the care with which the powder is milled into a fine

The production and properties of sintered Nd permanent magnets

309

Table 7.2 Typical properties of common grades of commercially available sintered Nd magnets

powder and handled. As a consequence, there are a large number of different grades of sintered Nd magnets that are commercially available. The properties of the most common grades that are produced worldwide are listed in Table 7.2. These magnets have a wide range of properties that are targeted for a wide range of applications. For many of these applications, cost is of overriding importance and the highest magnetic properties are of lesser important. The composition of sintered Nd magnets varies depending on the producer and target application. Much of this production takes place in China, which is now the world’s largest producer of sintered Nd magnets. However, it is believed that most of the highest grades of sintered Nd magnets are still produced by various companies in Japan.

7.3

The microstructure of sintered Nd magnets

There have been a large number of high-resolution SEM, magnetic force microscopy (MFM), transmission electron microscopy (TEM), and Lorentz TEM studies of sintered Nd magnets including those by Sagawa et al. (1984b), Fidler (1985), Hadjipanayis et al. (1985), Hiraga et al. (1985), Mishra et al. (1986), Hirosawa and Tsubokawa (1990), Makita and Yamashita (1999), Vial et al. (2002), Shinba et al. (2005), Li et al. (2009), Goto et al. (2012), and Yazid et al. (2016). Fig. 7.9 shows an SEM image of the polished surface of an N42 Grade sintered Nd magnet from the study by Yazid et al. (2016). The microstructure shown here is fairly typical of contemporary sintered Nd magnets. This magnet would have been produced by the process discussed earlier, including hydrogen decrepitation of a strip cast alloy followed by jet milling to produce the fine powder. During the final jet milling, the finest particles would have been removed to adjust the Nd and O2 content of the final sintered magnet. As can be seen, the microstructure consists almost entirely of grains of the Nd2Fe14B intermetallic phase with smaller amounts of the lighter Ndrich phase in the triple junctions between grains and in the grain boundaries. The microstructure was also reported to contain Nd-oxide precipitates and pores, which are largely located at the junctions of the grains. Although the grain boundaries of this SEM image are not well defined, all of the grain boundaries would contain a

310

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

Figure 7.9 SEM image of the surface of a sintered Nd magnet (Yazid et al., 2016).

thin B1.5 nm layer of the Nd-rich intergranular phase. As has been discussed in the previous chapters, dealing with melt-spun NdFeB materials, this thin Nd-rich boundary layer is an extremely important microstructural feature in all types of NdFeB permanent magnets and closely related to the development of coercivity, since, if it is absent, little or no coercivity develops. As mentioned in Section 7.2, a postsinter heat treatment can result in a dramatic increase in the coercively of sintered magnets. The earliest studies of sintered Nd magnets recognized the relationship between the intergranular phase and the coercivity of sintered Nd magnets, and proposed that the increase in coercivity resulting from the heat treatment was probably due to structural changes in the intergranular phase (Sagawa et al., 1984a,b, 1987; Hirago et al., 1985; Mishra et al., 1986). An example of the microstructural change that occurs in the grain boundary phase during the heat treatment of sintered Nd magnets is shown in Fig. 7.10 from the study by Vial et al. (2002). Image (A) shows an SEM of a portion of a sintered Nd magnet in the as-sintered state and after heat treating (B) for 1 hour at 520 C. The twobottom panels display TEM images of the boundary phase from the same magnets in the as-sintered state (C), and after heat treating (D): the as-sintered magnet has an Hci of 5.5 kOe while the magnet heat treated at 520 C for 1 hour had an Hci of 11.1 kOe. Whereas the grain boundary phase in the as-sintered magnet is ill-defined and of varying thickness, the grain boundary phase in the heat-treated sample appears to be smooth and well defined with a uniform thickness. From this, it was concluded that the increase in coercivity was due to structural improvement of the grain boundary phase, resulting in a reduction in certain defects, which act as nucleation sites for reverse domains. Shinba et al. (2005), carried out a comprehensive TEM study of the Nd-rich intergranular phase in sintered Nd magnets using Lorentz microscopy. The samples

The production and properties of sintered Nd permanent magnets

(A)

As-sintered

(C)

As-sintered

311

(B)

Heat treated 560°C

(D)

Heat treated 560°C

Figure 7.10 (A) SEM images of the grain boundaries in the as-sintered magnet and (B) after heat treating at 560 C for 1 hour. (C) TEM image of a grain boundary in the as sintered magnet and (D) TEM image of a grain boundary after the 560 C heat treatment. The as sintered magnet had an Hci of 439 kA/m (5.5 kOe) while the magnet heat treated for 1 hour at 560 C had an Hci of 1094 kA/m (11.1 kOe). Source: Adapted from Vial et al., 2002. J. Magn. Magn. Mater. 242, 1329.

investigated had a composition of Nd15Fe79B4 and were prepared from a 3 5 μm jet milled powder that was compacted and sintered at 1054 C for 1 hour followed by a two-stage heat treatment. The samples that were examined included the as sintered magnet (A) and samples that were given a postsinter heat treatment at 800 C for 1 hour (B) followed by a second heat treatment at 500 C for 1 hour (C). The overall microstructure of all three magnets was very similar in appearance to the SEM image in Fig. 7.9 and consisted almost entirely of grains of the Nd2Fe14B phase, with an diameter of B5 μm, and precipitates of the Nd-rich intergranular phase. The Nd-rich phase was found as large precipitates at the triple junction between grains as well as a thin layer between the grain boundaries. The Nd-rich phase was also found as precipitated within some of the Nd2Fe14B grains. Fig. 7.11 displays the demagnetization curves of the these three samples: the coercivities of Samples A, B, and C were reported as 314 kA/m (3.9 kOe), 419 kA/m (5.2 kOe), and 759 kA/m (9.5 kOe), respectively. A bright field TEM images of the grain boundary phase at triple junctions in all three samples are shown in Fig. 7.12. The lines radiating out from the precipitates, most evident in Samples A and B, are from strain fields surrounding the triple junctions. While there appears to be no effect on the morphology or shape of these triple junction precipitates as a result of

312

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

B

A

1.0

B (T)

C

0.5

1.0

0.5 –H (MA/m)

0

Figure 7.11 Demagnetization curves of sintered Nd magnets in the as-sintered state (A) and after a heat treatments at 800 C for 1 hour (B) followed by a second heat treatment at 500 C for 1 hour (Shinba et al., 2005).

the heat treatment, it is clear that the postsinter anneal reduces the intense strain field that exists around these precipitates. The circular features highlighted by the arrows in Sample A are also strain fields surrounding Nd-rich precipitates that were found to exist with the interior of grains in all of the samples. Fig. 7.13A shows an HIREM image of a triple junction from Sample C while (B) shows a higher magnification image of the rectangular area region highlighted in image (A). X-ray diffraction of the intergranular phase in the triple junction found it to be a crystalline phase with an f.c.c. structure and unit cell diameter of 0.058 nm, which is in agreement with earlier studies of sintered Nd magnets by Ramesh et al. (1986), Mishra et al. (1986), and Schrey (1986). This Nd-rich phase was found to contain a considerable amount of oxygen, and the f.c.c. phase was reported to be a variant of Nd2O3, which is stabilized by the addition of O2 and Fe. Although the Nd-rich phase in the triple junctions is clearly a crystalline compound, this phase was found to become increasingly disordered as the width of the grain boundary decreases and was found to be completely amorphous for the thin boundary separating two adjacent grains. The width of the grain boundary phase was found to be B1.5 nm, which is close to the 1 2 nm reported by Mishra (1986) for fine-grained melt-spun NdFeB and by Mishra and Lee (1986), Mishra (1987a), and Kirchner et al. (2004) for hot-deformed NdFeB magnets produced from melt-spun ribbon. This is quite remarkable given the dramatically different processing routes for these various materials and suggests that there is some energetic reason for the formation of this particular intergranular wall thickness.

The production and properties of sintered Nd permanent magnets

313

(A)

(B)

(C)

Figure 7.12 Bright field TEM images of the triple junction of sintered Nd magnets. (A) As sintered, (B) 800 C heat treatment for 1 hour, and (C) after heat treatment for 1 hour at 500 C. Source: Adapted from Shinba et al., 2005. J. Appl. Phys. 97, 053504.

Fig. 7.14 shows a TEM image of one of the Nd-rich precipitates that were found to form within the Nd2Fe14B grains in all three of the samples investigated. These precipitates varied in diameter but some were as large as 100 nm. These precipitates were also found to have an f.c.c. structure but the amount of oxygen was less than that found in the Nd-rich triple junction and grain boundary phase. The 110 planes

314

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

(A)

(B)

Figure 7.13 (A) Bright field TEM of the intergranular phase at a triple junction in Sample C (500 C heat treatment) and (B) higher magnification image of the rectangular region from image (A) showing the intergranular phase between two adjacent grains (Shinba et al., 2005).

Figure 7.14 A TEM image of one of the Nd-rich precipitates found in the interior of the Nd2Fe14B grains (Shinba et al., 2005).

The production and properties of sintered Nd permanent magnets

315

of these Nd-rich predicates was found to parallel with the 011 planes of the Nd2Fe14B grains. X-ray analysis of the precipitates found them to be coherent with the Nd2Fe14B phase but with a severe lattice mismatch, leading to high-strain fields around the grains. TEM images comparing of the grain boundaries phase in the as-sintered magnet (Sample A) and after heat treating at 500 C (Sample C) are displayed in Fig. 7.15. Whereas the boundary phase in the as-sintered material is undulating and of varying thickness, the grain boundary in the annealed sample is smooth and more uniform. This study proposed that, since the low temperature 500 C anneal is below the melting point (665 C) of the Nd-Fe eutectic phase, that the heat treatment serves to structurally change the amorphous grain boundary into a smooth uniform layer as a result of a reduction in interfacial energy. The heat treatment results in a smooth grain boundary phase that is coherent with the Nd2Fe14B grains as well as a reduction in the strain fields around the boundary phase and the Nd-rich precipitates residing within the grains. The heat treatment does not, however, completely remove the strain from the grain boundary phase since there is still considerable lattice mismatch between the intergranular phase, the precipitates and the Nd2Fe14B grains. Fig. 7.16 shows Lorentz images of the domain structure in Sample C from this study. Image (A) shows that the domain structure proceeds continuously across a thin Nd-rich grain boundary, as indicated by the observation that the polarity of the domains in the same on either side of the boundary. In contrast, image (B) shows that the magnetization is discontinuous across a large Nd-rich precipitate of nearly 100 nm in width. Image (C) shows a triple junction where the polarization is clearly disrupted. The polarization is also disrupted across the thicker intergranular phase boundary leading off from the triple junction on the left. Again, this is clear from the change in polarity that occurs on opposite sides of the boundary. The study concluded that the domain structure and magnetization process in sintered Nd magnets are governed by longer range magentostatic or dipole interactions rather than exchange interactions and that this interaction weakens as the width of the grain boundary increases. One puzzling observation is the presence of a domain structure within the larger Nd-rich precipitates as seen in images (B) and (C). This phase has long been reported

(A)

(C)

Figure 7.15 Bright field TEMs of the grain boundary phase in the as sintered sample (Sample A) and after heat treating at 500 C (Sample C) for 1 hour. Source: Adapted from Shinba, Konno, Ishikawa, et al., 2005. J. Appl. Phys. 97, 053504.

316

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

(A)

(B)

(C)

Figure 7.16 Lorentz TEM images of the domain structure in Sample C (A) across a grain boundary with a thin layer of intergranular phase, (B) in the vicinity of a large Nd-rich precipitate residing within a Nd2Fe14B grain, and (C) near a triple junction and a grain boundary with a thicker layer of the Nd-rich intergranular phase (Shinba et al., 2005).

as paramagnetic in nature and classical magnetism teaches that a paramagnetic material cannot support a domain wall. It is therefore noteworthy that recent studies by Murakami et al. (2014), Kohashi et al. (2014), and Nakamura et al. (2014) have reported that the Nd-rich phase in sintered Nd magnets is actually ferromagnetic. In these studies, sintered magnets were fractured under high vacuum and the surfaces of the grain boundary phase examined using various analytical techniques. If this is the case, then it would explain the presence of this domain structure. If the f.c.c. grain boundary phase is ferromagnetic in sintered Nd magnets, then is also likely the same for the Nd-rich intergranular phase in hot-deformed NdFeB magnets and would go far to explain why the domain walls in these magnets are seen to move so readily through the Nd-rich grain boundary phase. Examples of this were shown in the Lorentz TEM images of hot-deformed magnets shown in Figs. 6.20 and 6.30.

7.4

The magnetization process in sintered Nd magnets

As with all rare earth-transition metal permanent magnets, the origin of the coercivity in sintered Nd magnets is the large magnetocrystalline anisotropy of the Nd2Fe14B intermetallic phase. Because of the high anisotropy, magnetization

The production and properties of sintered Nd permanent magnets

317

reversal must occur by the motion of domain walls, since coherent rotation is not energetically feasible. The domains in sintered magnets have been examined extensively by various techniques including the Kerr effect (Livingston, 1985), scanning transmission X-ray microscopy (Ono et al., 2011); magnetic force microscopy (MFM) (Folks et al., 1996; Szmaja et al., 2004; Yazid et al., 2016), small angle neutron scattering (Perigo et al., 2015), and Lorentz TEM microscopy (Hadjipanayis et al., 1985; Sagawa et al., 1985; Fidler, 1987; Fidler and Knock, 1989; Shinba et al., 2005; Li et al., 2009; Ono et al., 2011; Woodcock et al., 2012) and all have found that the grains consist of multiple domain in the thermally demagnetized state and that these domains move easily upon application of a magnetic field. An example of the observation of the domain pattern using Lorentz microscopy by Shinba et al. (2005) was shown in Fig. 7.16. One of the best techniques for observing the domain structure in permanent magnet materials is MFM and an example of a MFM image from an N42 sintered Nd magnet is shown in Fig. 7.17 from the study by Yazid et al. (2016). Here the domains can be clearly distinguished as stripes running parallel with the applied field and roughly parallel with the c-axis of the Nd2Fe14B grains. The observation that the domains in the separate grains are not exactly parallel shows that complete alignment of the grains was not achieved during the alignment, compaction, and sintering process. The grains in this magnet appear to be magnetically isolated from each other by the Nd-rich grain boundary layer, as indicated by the opposite polarity of the domains on either side of the grain boundaries. This is at odds with the TEM image shown in Fig. 7.16A, which appears to show magnetic continuity

Figure 7.17 MFM image of an N42 Grade sintered Nd magnet. Source: Adapted from Yazid et al., 2016. IEEE Trans. Magn. 52, 2100610.

318

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

across a thin boundary layer but not across the thicker boundary layers. As found in all of the previous studies of the domain structure in sintered Nd magnets, the domain walls in thermally demagnetized samples were found to move easily upon application of an applied field. This high initial susceptibility is believed to reflect a dominant nucleation controlled coercivity mechanism (Sagawa et al., 1984b; Hadjipanayis and Tao, 1985; Heinecke et al., 1985; Li and Strnat, 1985; Durst and Kronmuller, 1987; Kronmuller et al., 1988; Hirosawa and Sagawa, 1987; Eckert et al., 1990). To cite one example, Pinkerton and Van Wingerdon (1986) compared the initial magnetization of various types of NdFeB permanent magnets including fine-grained melt-spun ribbons, hot-deformed magnets produced from melt-spun materials, and sintered Nd magnets. Fig. 7.18 shows a plot of the normalized Br [Br/Br(N) where Br(N) is the saturation magnetization of the magnets] versus normalized Hci [Hm/Hci(N), where Hci(N) is the coercivity value at full saturation]. Normalized data are used so that better comparison of the different materials can be shown. As seen here, the Br of both the sintered and hot-deformed magnets rises rapidly and achieve nearly full saturation at an applied field level well below the coercivity level of the samples. For sintered Nd magnets, this is believed to result because the grains contain multiple domain walls, which can move easily, resulting in relatively easy magnetization of the individual Nd2Fe14B grains and complete polarization at relatively low applied field (H{Hci). The fact that it does not completely saturate is believed due to the slight misalignment of some of the grains, as is indicated in the MFM image in Fig. 7.17. The same is at least partially true for anisotropic hot-deformed materials, which are fine-grained materials but consist of large interaction domains in the thermally demagnetized state. The domain walls of these interaction domains also move easily during the first stage of 1.2

1.0

Br/Br (∞)

0.8

0.6

0.4

Ribbons Hot pressed Die upset Sintered

0.2

0 0

0.5

Hm/Hci (∞)

1.0

1.5

Figure 7.18 Remanence Br (normalized by the remanence Br(N) of a fully saturated sample) versus the magnetizing field Hm (normalized to Hci(N)) for various NdFeB permanent magnet materials (Pinkerton and Van Wingerdon, 1986).

The production and properties of sintered Nd permanent magnets

319

the magnetization process, resulting in a comparatively high susceptibility but then appear to become pinned at Nd-rich particles or in the Nd-rich intergranular phase at the edges of the platelet shaped grains. This is consistent with the Lorentz TEM studies of these materials (see Sections 6.3 and 6.5). In contrast, the individual Nd2Fe14B grains in both the fine-grained melt-spun ribbon and hot-pressed magnets made from this hot-pressed ribbon are initially free of domain walls and require the formation of a domain wall for magnetization of the samples to occur. This results in a resistance to an increase in the magnetization and a slow rise in Br and Hci for these materials. The fact that the susceptibility of the hot-pressed material lies somewhere between the melt-spun ribbon and the anisotropic magnets suggests that many of the grains in this material may also contain multiple domains, which initially move easily and allow partial magnetization of the sample. However, the susceptibility decreases at higher applied field levels because the smaller, single-domain particles require the nucleation of a domain wall for magnetization reversal to proceed. The magnetization process in these isotropic fine-grained materials is discussed in Section 3.3 and for hot-deformed materials in Section 6.5. Various studies have found that the coercivity of sintered Nd magnets saturates at a field level significantly below that of the coercivity (Pinkerton and Van Wingerden, 1986; Durst and Kronmuller, 1987) and Durst and Kronmuller (1985, 1987) and Kronmuller et al. (1988) and have argued that if Hsat , Hci, the coercivity must be controlled by the nucleation of domain walls. The magnetization process that is believed to be applicable for a thermally demagnetized sintered Nd magnet is depicted in Fig. 7.19. Here the domain structure of the Nd2Fe14B grains is shown as the magnet is magnetically saturated in the first quadrant and then demagnetized into the second quadrant. Starting from a thermally demagnetized state with strips of domains (A), the application of a magnetizing field results in easy domain wall motion and the growth in the domains with polarity parallel to the applied field (B). This is accompanied by a rapid rise in the magnetization of the magnet. Magnetization continues until all of the domain walls are swept from the magnet and nearly complete saturation is achieved at an applied field level well below the Hci value of the magnet (C). Demagnetization from this saturated state requires the nucleation of new reverse domains walls as shown in rendering (D), which depicts new domains nucleating from a triple junction and from a Nd-rich precipitate within one of the grains. There is a resistance to the formation of these new walls, resulting in a significant coercive force. As the applied field is increased the new domain walls grow and develop a new domain structure until, at the Hci value, it would form a domain structure once again similar to (A) where M 5 0 and exactly half of the domain have a polarity in one direction and half in the other. However, as with fine-grained melt-spun materials and hotdeformed magnets, there is still a difference of opinion on the exact nature of the coercivity mechanism in sintered Nd magnets. Although some early studies believed that the domain walls were pinned in the intergranular phase and become unpinned during the demagnetization process, this is not the general accepted opinion. Most of the investigations have concluded that reverse domain walls nucleate from defects in or near the grain boundary phase, where the local anisotropy is

320

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

(A)

(C)

(B)

N N

S S

N

N S

N

S N S N

S

S N

N

S N S N

S S

N

S

N S

(D)

S

N

16 N

14 D

C

12 S M (KG)

10 8 B 6

A 10

6

2

N N S

4

18

N

S

S

2 0 –2

–6

–10

–14

New reverse domains

–H (kOe)

Figure 7.19 Magnetization process believed applicable for sintered Nd magnets. (A) thermally demagnetized state. (B) domain structure after magnetizing to a field level approximately have the Hci value. (C). Domain structure after full magnetization and (D) the formation of new domains at the knee in the demagnetization curve.

lower and that the postsintering heat treatment reduces the number or character of these nucleation sites by structural changes to the Nd-rich precipitates and grain boundary phase. Most conclude that this is accomplished either by removing or reducing defects on the surface of the Nd2Fe14B grains or by reliving stress between the grain boundary phase and the Nd2Fe14B grains (Fidler, 1987; Fidler and Knock, 1989; Vial et al., 2002; Ono et al., 2011; Woodcock et al., 2012; Goto et al., 2012; Bance et al., 2014). Several TEM studies (Fukagawa and Hirosawa, 2008; Matsuura et al., 2009), which included micromagnetic modeling simulations, have reported that the coercivity of sintered Nd magnets is likely controlled by a thin, coherent boundary layer of Nd2O32x (x . 0) between the Nd-rich intergranular phase and the Nd2Fe14B phase. This thin boundary layer produces distortions in the Nd2Fe14B intermetallic phase, which produces magnetoelastic strain and a drop in local anisotropy. Heat treating the sample can reduce the strain by reducing the distortions in the Nd2Fe14B grain adjacent to the boundary.

The production and properties of sintered Nd permanent magnets

321

There are also studies that report that the coercivity mechanism in sintered Nd magnets is controlled by both nucleation and pinning of domain walls at defects in the Nd-rich grain boundary phase and Nd-rich precipitates. For example, Yazid et al. (2016) study the nature and size of defects in sintered Nd magnets using MFM. While they did observe new reverse domain walls forming at the Nd-rich boundary phase and precipitates, they also observed that the walls appear to became pinned at other defects. They also reported that the defects were oxide in nature, rather than metallic, and further concluded that defects are an important microstructural feature in sintered Nd magnets and that higher coercivity could likely be obtained by controlling the size and distribution of defects between the Nd2Fe14B grains. Another commonly held theory regarding the coercivity mechanism in sintered Nd magnets is that the coercivity is controlled by the degree of magnetic isolation between the grains and that the heat treatment increases the coercivity by structural changes to the grain boundary phase that increase this magnetic isolation (Eckert et al., 1987; Schneider et al., 1990; Zhou et al., 1990; Makita and Yamashita, 1999; Vial et al., 2002). Hirosawa and Tsubokawa (1990) were the first to propose that the coercivity was dependent on the level of magnetic isolation between the Nd2Fe14B grains. This view is supported by various micromagnetic modeling studies (Fidler and Schrefl, 2000; Schrefl et al., 1993) that have concluded that magentostatic interactions dominate in permanent magnets with large grains, including sintered Nd magnets. These modeling studies further conclude that coercivity is controlled by changes in the demagnetization field in the individual Nd2Fe14B grains and, moreover, that the magnetization reversal of any given grain occurs when the total internal field in the grain, which is equal to the externally applied field and the demagnetization field, equals the field needed to nucleate a new domain wall. Since the demagnetizing field is related to the level of magnetostatic interaction, these studies conclude that the presence of the Ndrich grain boundary phase increases the coercivity by increasing the magnetic isolation between grains and, thereby, significantly changing the level of magnetostatic interaction. Various micromagnetic modeling studies have also shown that the demagnetizing field is also dependent on grain size, grain shape, and grain orientation, which may explain why the coercivity of rare earth-transition metal permanent magnets increase with decreasing grain size and the degree of grain orientation (Thielsch et al., 2013; Bance et al., 2014; Sepehin-Amin et al., 2014; Yi et al., 2016; Jujisaki et al., 2016). A somewhat similar explanation was first proposed by Hirosawa and Sagawa (1987) to explain the dependence of Hci and Br on magnetizing field. They concluded that domain walls experience an energy barrier near grain boundary regions that originate from magnetostatic effects around defects, such as sharp protrusions on the Nd2Fe14B grains. The increase in coercivity as a result of a postsinter that treatment would increase the coercivity because the action of the Nd-rich phase removes these defects during the heat treatment. To summarize, there is still no consensus on the coercivity mechanism in sintered Nd magnets. Almost all of the recent studies of these materials conclude that

322

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

reverse domains are nucleated at the Nd-rich grain boundary phase and the Nd-rich precipitates. However, whether this nucleation occurs at defects or at regions of high stray or demagnetization field is still being debated. Also, whether the nucleated domains become pinned during part of the demagnetization process as reported in the MFM study by Yazid et al. (2016) also remains an open question.

References Bance, S., Seebacher, B., Schrefl, T., Exl, L., Winkhofer, M., Hrkac, G., et al., 2014. J. Appl. Phys. 116, 233903. Bernardi, J., Fidler, J., Sagawa, M., Hirose, Y., 1989. J. Appl. Phys. 83, 6396. Burzo, E., Kirchmeyer, H., 1989. In: Gschneidner Jr., K., Eyring, L. (Eds.), Handbook of the Physics and Chemistry of Rare Earths, vol. 12. North Holland Press, Amsterdam. Durst, K.D., Kronmuller, H., 1985. In: Strnat, K. (Ed.), Proceedings of the Forth International Symposium on Magnetic Anisotropy and Coercivity in Rare EarthTransition Metal Alloys, University of Dayton, Dayton, OH, 725. Durst, K.D., Kronmuller, H., 1987. J. Magn. Magn. Mater. 68, 63. Eckert, D., Hinz, D., Handstein, A., Schneider, J., 1987. Phys. Status Solidi. A. 101, 563. Eckert, D., Mueller, K.H., Handstein, A., Schneider, J., Grossinger, R., Kreenka, R., 1990. IEEE Trans. Magn. 26, 1834. Fidler, J., 1985. IEEE Trans. Magn. MAG. 21, 1955. Fidler, J., 1987. IEEE Trans. Magn. MAG. 23, 2106. Fidler, J., Knock, K.G., 1989. J. Magn. Magn. Mater. 80, 48. Fidler, J., Schrefl, T., 2000. J. Phys. D Appl. Phys. 33, R-135. Folks, L., Street, R., Woodward, R.C., Babcock, K., 1996. J. Magn. Magn. Mater. 159, 109. Fukagawa, T., Hirosawa, S., 2008. J. Appl. Phys. 104, 013911. Goto, R., Matsuura, M., Sugimoto, S., Tezuki, N., Une, Y., Sagawa, M., 2012. J. Appl. Phys. 111, 07A739-3. Hadjipanayis, G.C., Lawless, K.R., Dickenson, R.C., 1985. J. Appl. Phys. 57, 4097. Hadjipanayis, G.C., Tao, Y.F., 1985. J. Phys. (Paris). 46, C6 237. Hadjipanayis, G.C., Tao, Y.F., Lawless, K.R., 1985. In: Strnat, K.J. (Ed.), Proceeding of the Forth International Symposium on Magnetic Anisotropy and Coercivity in Rare Earth Transition Metal Alloys, University of Dayton, Dayton, OH, 657. Harris, I.R., McGuiness, P.J., Jones, D.G.R., Abel, J.S., 1987. Script. Phys. T19, 435. Harris, I.R., Noble, C., Bailey, T., 1985. J. Less Common Metals. 106, L1. Heinecke, U., Handstein, A., Schneider, J., 1985. J. Magn. Magn. Mater. 53, 236. Hirago, K., Hirabayashi, M., Sagawa, M., Matsuura, Y., 1985. Jpn J. Appl. Phys. 24, L30. Hirosawa, S., Sagawa, M., 1987. J. Magn. Magn. Mater. 71, L1. Hirosawa, S., Tsubokawa, Y., 1990. J. Magn. Magn. Mater. 84, 309. Hirota, K.H., Nakamura, T., Minowa, Honshima, M., 2006. IEEE Trans. Magn. 42, 2909. Jujisaki, J., Furuya, A., Uehara, Y., Shimizu, K., Ataka, T., Tanaka, T., et al., 2016. AIP Adv. 6, 056028. Kirchner, A., Thomas, J., Gutfleisch, O., Hinz, D., Mu¨ller, K.-H., Schultz, L., 2004. J. Alloys Compounds 365 (1 2), 286. Kobayashi, K., Urushibata, K., Une, Y., Sagawa, M., 2013. J. Appl. Phys. 113, 163910. Kohashi, T., Motai, K., Nishiuchi, T., Hirosawa, S., 2014. Appl. Phys. Lett. 104, 232408.

The production and properties of sintered Nd permanent magnets

323

Kronmuller, H., Durst, K.D., Sagawa, M., 1988. J. Magn. Magn. Mater. 74, 291. Li, D., Strnat, K.J., 1985. J. Appl. Phys. 57, 4143. Li, D., Suzuki, S., Kawasaki, T., Mashida, M., 2008. Jpn J. Appl. Phys. 47, 7876. Li, W.F., Ohkubo, T., Hono, K., 2009. Acta Materialia 57, 1337. Livingston, J.D., 1985. J. Appl. Phys. 57, 4137. Makita, K., Yamashita, O., 1999. Appl. Phys. Lett. 74, 2056. Matsuura, M., Sugimoto, S., Goto, R., Tezuka, N., 2009. J. Appl. Phys. 105, 07A741. McGuiness, P.J., Devlin, E., Harris, I.R., Rozendaal, E., Ormerod, J., 1989. J. Mater. Sci. 24, 2541. Mishra, R.K., 1986. J. Magn. Magn. Mater. 54, 450. Mishra, R.K., 1987a. In: Sankar, S.G., Herbst, J.F., Koon, N.C. (Eds.), High Performance Permanent Magnet Materials, Materials Research Society Symposium Proceedings, vol. 96. Materials Research Society, Pittsburg, CA. Mishra, R.K., Chen, J.K., Thomas, G., 1986. J. Appl. Phys. 59, 2244. Mishra, R.K., Lee, R.W., 1986. Appl. Phys. Lett. 48, 773. Murakami, Y., Tanigaki, T., Sasaki, T.T., Takino, Y., Park, H.S., et al., 2014. Acta Mater. 71, 370. Nakamura, T., Yasai, T., Kotani, Y., Fukagawa, T., Nishiuchi, T., et al., 2014. Appl. Phys. Lett. 105, 202404. Ono, K., Araki, T., Miyamoto, N., Shoji, T., Kato, A., et al., 2011. IEEE Trans. Magn. 47, 2672. Ormerod, J., 1989. Powder Metallur. 32, 244. Pinkerton, F.E., Wingerdon, D. J. Van, 1986. J. Appl. Phys. 60, 3685. Perigo, E.A., Gilbert, E.P., Michaels, A., 2015. Acta Mater. 87, 142. Ramesh, R., Krishna, K.M., Goo, E., Thomas, G., Okada, M., Homma, M., 1986. J. Magn. Magn. Mater. 54, 563. Rodewald, W., Blank, R., Repel, G.W., Zilig, H.D., 2000. In: 16th International Workshop on RE Magnets and Their Applications, Sendai, Japan, 119. Sagawa, M., Fujimori, S., Togawa, M., Matsuura, Y., 1984a. J. Appl. Phys. 55, 2083. Sagawa, M., Fujimura, S., Togawa, H.N., Yamamoto, H., Matsuura, Y., 1984b. J. Appl. Phys. 55, 2083. Sagawa, M., Fujimura, S., Yamamoto, H., Matsuura, Y., Hiraga, K., 1984c. IEEE Trans. MAG. 20, 1584. Sagawa, M., Fujimura, S. Yamamoto, H. Matsuura, Y. Hirosawa, S., Hiraga, K. 1985. In: Strnat, K.J. (Ed.) Proceedings of the 4th International Symposium on Magnetic Anisotropy and Coercivity in Rare Earth-Transition Metal Alloys, University of Dayton, Dayton, OH, 587 (1985). Sagawa, M., Fujimura, S. Matsuura, Y. Magnetic Materials and Permanent Magnets, US Patent 4,792,368, issued 1988. Schneider, G.T., Henig, E.-T., Missell, F.P., Petzow, G., 1990. Z. Metallkds 81, 322. Schneidner, G.T., E.-T., Henig, H.H. Stadelmeir and G. Petzow, 1987. In: Proceedings of the 9th Rare Earth Magnet Conference, 347. Schrefl, T., Schmidts, H.F., Fidler, J., Kronmuller, H., 1993. IEEE Trans. MAG. 29, 2878. Schrey, P., 1986. IEEE Trans. Magn. MAG. 22, 913. Sepehin-Amin, H., Ohkubo, T., Gruber, M., Schrefl, T., Hono, T., 2014. Script. Mater. 89, 29. Sepehin-Amin, H., Ohkubo, T., Hono, K., 2010. J. Appl. Phys. 107, 09A745. Susuki, H., Satsu, Y., Komuro, M., 2009. J. Appl. Phys. 105, 07A734. Szmaja, W., Grobel, J., Cichomski, M., Makita, K., 2004. Vacuum 74, 297. Thielsch, J., Suess, D., Schultz, L., Gutfleisch, O., 2013. J. Appl. Phys. 114, 223909.

324

Rapidly Solidified Neodymium-Iron-Boron Permanent Magnets

Tokunaga, M., Meguro, N., Endoh, E., Tanigawa, S., Hirada, H., 1985. IEEE Trans. Magn. MAG. 21, 1964. Vial, F., Joly, F., Nevalainen, E., Sagawa, M., Hirago, K., Park, K.T., 2002. J. Magn. Magn. Mater. 242, 1329. Watanabe, N., Itakura, M., Kuwano, N., Li, D., Suzuki, S., Machida, K., 2007. Mater. Trans. 48, 915. Woodcock, T.G., Zhang, Y., Hrkac, G., Ciuta, G., Dempsy, N.M., Schrefl, T., 2012. Script. Mate. 67, 536. Yamamoto, H., M. Sagawa, S. Fujimura and Y. Matsuura, Process for Producing Magnetic Materials, US Patent 4,601,875, issued 1986. Yazid, M.M., Olsen, S.H., Atkinson, G., 2016. IEEE Trans. Magn. 52, 2100610. Yi, M., Gutfleisch, O., Xu, B.-X., 2016. J. Appl. Phys. 120, 033903. Zhou, G.F., Fu, S.Y., Sun, X.K., Chang, Y.C., 1990. Phys. Status Solidi A 121, 257.

Selected Readings Buschow, K.H.J., 1988. In: Wohlfarth, E.F., Buschow, K.H.J. (Eds.), Ferromagnetic Materials, vol. 4. North Holland Press, Amsterdam, p. 1. Ormerod, J., 1985. The physical metallurgy and processing of sintered rare earth permanent magnets. J. Less Comm. Met. 111, 49. Sagawa, M., Hirosawa, S., Yamamoto, H., Fujimura, S., Matsuura, Y., 1987. Nd-Fe-B permanent magnet materials, Japan. J. Appl. Phys. 26, 785. Shinba, Y., Konno, T.J., Ishikawa, K., Hiraga, K., Sagawa, M., 2005. Transmission electron microscopy study on Nd-rich phase and grain boundary structure of Nd Fe B sintered magnets. J. Appl. Phys. 97, 053504.