Use, Reduction and Probable Remedies of Heavy Rare Earth Metals in Neodymium Iron Boron Alloys

Shuk Rashidi
Vice President
Tridus Magnetics and Assemblies
Rancho Dominguez, CA


Abstract—The advent of NdFeB permanent magnets ushered in a slate of new applications that were virtually impossible with the existing families of permanent magnet alloys.The lower Curie Temperatures and the associated thermal magnetization losses of Br and Hci, made the use of NdFeB magnets difficult if not down right impossible- in host of high temperature devices.To enhance the anisotropy field of this family of alloys, a portion of light rare earths was replaced by heavy rare earth metals. The latter enhancement was accompanied by some loss of remanence and the problems of a poor balance between the heavy and light rare earth metals. This paper addresses the benefits and the problems associated with the heavy rare earths and the potential attempts to reduce and or avoid the dependence on heavy rare earths wherever and whenever possible.


Up until 1983, SmCo5 and Sm2Co17 made up the bulk of magnets used in miniature, high performance devices, ranging from computer disk drives to a host of factory automation, office automation and automotive components. The extremely high energy products and the superb Curie Temperatures of the latter alloys ushered in a host of new applications which were virtually impossible with other existing families of permanent magnets. The cobalt crisis of 1977, when the Katangese rebels invaded Zaire and flooded the cobalt mines and causing the price of cobalt to escalate to better than US$ 40.00/lb from a low of US$ 4.00/lb - in the gray market, instilled fear in the minds of many manufacturers, not to mention the magnet users. The magnet laboratories through out the industrialized nations started looking at different families of hard magnetic alloys, in quest of inventing, a system or systems of alloys that were not at the mercy of sources of supply raw materials, some of which were in politically and economically unstable regions of the world.

This was the genesis of Neodymium Iron Boron (NdFeB) magnets. In 1983, John Croat of GM and Sagawa of Sumitomo Specialty Metals Corporation invented the so- called MQ ( Magnequench) and sintered NdFeB permanent magnets, almost concurrently. In Fig(1), Yukata Matsuura of Neomax depicts the increasing trend in the energy product of NdFeB alloys.

Figure 1.  Transition of world record energy production

Figure 1.  Transition of world record energy product


The evolution of the NdFeB seemed to have a few more complications than that of its predecessors SmCo5 and Sm2Co17 systems of alloys. The lower Curie Temperature (around 315 degree centigrade), the magnet being prone to corrosion, coupled with lower thermal stability, were some of the unique undesirable characteristics, that for better or for worse separated this alloy from the Sm-Co families of permanent magnets.

The material scientists responsible for the invention of Nd based rare earth transition metal alloys did what any typical inventor does. They based their technology of the manufacturing of the new system of alloys, on an existing process that had proven effective and had been around for years. In this case, most of the manufacturing events were primarily based on the technology of powder metal process used for making Sm-Co.

Unfortunately, there were some vast differences between the two alloys. In the early days of Nd magnets it was not uncommon to prepare an alloy with a composition slightly richer in Nd as it was done with the SmCo alloys. None the less, an excess amount of Nd increased the level of Nd - rich phase in the grain boundary, which in turn made the magnet more prone to corrosion.

The second challenge for the manufacturer and for the end user was the fact that the magnet exhibited a much lower Curie temperature leading to higher reversible and irreversible temperature coefficient of induction and coercive forces. The end user did not have the luxury of taking the magnet to a high temperature and yet expect the device to perform as though the magnets were Sm based. So effectively we had a new family of permanent magnets that had room temperature characteristics of SmCo and yet the Curie temperature was even lower than that exhibited by hard ferrites.


The industry launched a systematic study to improve the unit magnetic properties of the Nd based magnets, thru the implementation of a host of process changes, and changes in the microstructure of the alloy. The lower Curie Temperature of NdFeB magnets is responsible for its higher reversible coefficient. Cobalt was added to increase the Curie temperature. But Cobalt does tend to reduce the anisotropy therefore the intrinsic coercive force, if the level of Cobalt exceeds a certain level. The latter limit is indicated to be around 10%. The coercivity of NdFeB magnets in practice amounts to no more than 20 to 40% of the theoretical limits. Sagawa and coworkers discovered that replacement of a portion of Nd by Dysprosium would enhance the anisotropy of Nd based alloy and thus increase the intrinsic coercive force. Immediately following Sagawa, Ghandahari and Fidler discovered that using Dy2O3 as a sintering additive also had a similar result. The latter was a low cost approach to the enhancement of the Nd based alloy. The addition of DY2O3, however, did introduce a much higher content of oxygen into the alloy (oxide causes decrease in the remanence and possibly provides easy nucleation sites for nucleation of domain walls) exhibited less homogeneous distribution of Dy into 2-14-1 grains. The practice of adding Dy became a more bona fide method for increasing the anisotropy of the alloy. Fig (2) shows the effect of Dysprosium on Br and Hci. Fig (3) is trend curve for the enhancement of Hci between 1980 and 2005.


Figure 2.  The effect of Dysprosium on Br and Hci

Figure 2.  The effect of Dysprosium on Br and Hci


Figure 3.  Improvement of Nd-Fe-B sintered magnet coercive force

Figure 3.  Improvement of Nd-Fe-B sintered magnet coercive force


Thereafter, Dysprosium and Terbium (Dy and Tb) were added to increase the intrinsic coercive force. The latter did result in higher cost for the alloys and a partial loss of induction, as the heavy rare earths couple ferri- magnetically with transition metal and thus reduce the saturation magnetization. The accepted, typical rule was that for adding one percent of heavy rare earth, the intrinsic coercive force of the alloy will enhance by better than 1.5 Kilo Oersted. The important finding of use of Dy was the fact that up to 10% replacement of total rare earth, by Dy did result in increase in the intrinsic coercive force, beyond which there is a strong deviation from linearity. The doping of non- rare earth metals did show some promise of increasing the intrinsic coercive force. Use of metals such as Aluminum and Niobium may enhance the intrinsic coercive force to some extent as well. J.K.Chen and G.Thomas made two alloys where in one a portion of Iron was replaced by Aluminum while in the second alloy a percentage of Boron was replaced by Aluminum. Both showed a marked increase in the intrinsic coercive force with unfortunately higher than 4% decrease in the value of remanence.

The advent of “strip cast” process versus the “ingot cast” (cupcake or book mold type casting), allowed to make alloys with higher homogeneity and far better microstructure - another reason for greater improvement in the unit magnetic properties.
The above improvement made possible, the use of NdFeB in many new devices specially those that required exposure to higher temperatures and higher demagnetizing forces. The NdFeB magnet users and manufacturers therefore fell heavily for magnets using progressively more and more heavy rare elements, as the end users designed their magnetic circuits, conservatively, around alloys which may had much higher coercive force than they really required. The cost of the heavy additive was reasonable enough to justify such change.


The manufacturers decided to adopt a series of grade designations, where the last two letters of the alloy effectively indicated the nominal content of Nd, Dy and Tb. These were designated for example as N35, N38H, N38SH and N38UH, N35EH and etc. So a N38H alloy had a nominal Br of o 12.4 KGauss, and nominal Hci of 17 K Oersted, and supplied by most of the world class suppliers with essentially the same nominal unit magnetic properties. The content of the heavy rare earths typically varied from manufacture to manufacture by 0.5% or thereabouts.

While magnets of the same grade designation may be manufactured by different manufacturers, exhibiting essentially equivalent room temperature properties, the high temperature performance may vary from manufacture to manufacture as the technology of processing may not be of the same level of sophistication. Less technology savvy manufacturers of NdFeB magnets may have used a higher level of HREs in their alloy in an attempt to level the playing field of the magnetic performance, at higher temperatures.
This move, while may have been a good remedy, did seed some problems, both for the magnet suppliers as well as for the magnet users.


Since the premium for the use of HREs was very small, the conservative designer may have used alloys with much higher intrinsic coercive force than he actually needed. In some cases the designer specified the knee of the magnet (at higher temperatures) clear in the third quarter of the magnetization curve. In other words some designer pushed the use of HREs to levels that ultimately became very expensive if not down right impossible. Some designers decided to go away from free Dy alloy to one with some Dy by content, by redesigning the magnetic circuit with thinner magnet, which allowed him (or her) the use of larger air-gap and less problems associated with mechanical dimensioning.


To usher in a semblance of order in the mining, usage and export of rare earths, the Chinese government changes the rules. The changes had a gradually increasing effect on the prices of rare earths across the board, while the HREs felt the effect more intensely as the fear of “balance” took a new dimension.In a nut shell, in the following five and one half years, the price of Nd metal went to higher than 700,000 RMB/ton, from a low of 45,000 RMB/ton, while Dy increased from 500 RMB/Kg ( or thereabouts) to 4,000 RMB/Kg - at the end of the first quarter of 2011.
Fig (4 a) and Fig (4b) shows the change in the price of Nd during the last decade.

Figure 4A.  Neodymium Price - A Historical Perspective

Figure 4B.  The change in the price of Nd during the last decade

Figure 4.  The change in the price of Nd during the last decade



The designer was presented with three choices that could potentially solve his problem of not to cope with the thermal problems in his design.
(1) Use a thicker magnet – no expensive heavy rare earths.
(2) Design the system with much smaller air gap and move the operating point to much higher load line.
(3) Use a magnet replacing part of Nd with Dy or Tb at the expense of some sacrifice in the value of air gap flux or air gap flux density and risk the probability of not being able to buy heavy rare earths at all at some point in the future.
Fig (5) is Modified Operation load line and change of alloy.


Figure 5.  Modified operation load line and change of alloy

Figure 5.  Modified operation load line and change of alloy


Choice No.1 would mean use of larger device and larger foot print, Choice No.2 would make the cost of non- magnet portion of the circuit more expensive. Therefore choice no.3 was the most logical one to him. Un be-known to him and the magnet supplier the prices of Nd, and Dy would escalate from a low of 45,000 RMB /ton to high of better than 650,000 RMB /ton between year 2004 to 2011, with Dysprosium going thru the same multiple of price increases in China.


The word “balance” was used to point out the lop-sided ratio of light rare earths to heavy rare earths. This was an area of concern as for every 1700 pounds of Nd Oxide, they could produce only one pound of Dysprosium Oxide, in a typical Chinese Bastnasite deposit, while this ratio was around 25 times in the Chinese ionic ore. The ratio of Terbium Oxide to that of Nd oxide was even less favorable. Fig ( 6A ).

Figure 6A.  The ratio of Terbium Oxide to Nd Oxide

Figure 6A. The ratio of Terbium Oxide to Nd Oxide


Chinese rare deposits happen to have a healthy of dose of HREs. The next known largest rare earth deposit in Mountain Pass California, now owned by Moly Corp has virtually no HREs, while there is no public report about the content of HREs the central Asian rare earth deposits.
So there could be better incentive than very higher prices generally for rare earth metals, compounded by limited availability of HREs. We see a concerted effort among the US end users to reduce the use HREs to an absolute minimum or purge such usage all together if possible.
Fig.(6B) shows the estimated supply of rare earth elements from known sources as of March 2010.
Figure 6B.  Estimated supply of rare earth elements from known sources as of March 2010

Figure 6B. The estimated supply of rare-earth elements from known sources as of March 2010



(1) They are designing new devices around Dy or Tb free alloys.

(2) The have grandfathered some of the old designs around alloys with HREs as changing those will be a virtual impossibility.

(3) They are looking at a completely different paradigm of design in that they spend a bit more money on the components (tighter tolerance, higher quality return structure) and use less rare earth of less heavy rare earths.
For example, in some motor applications for instance, the use of NdFeB magnets in hybrid cars, some manufacturers use higher heavy rare content alloy and avoid use of cooling system. In light of the scarcity of heavy rare earths, it may be more logical to incorporate cooling system and avoid use of heavy rare earths.

(4) They specify their operating load line and the operating point of the magnet in the second quadrant curve and let the manufacturer decide what alloy they should use. Generally the designer uses a more conservative view in terms of unit properties, specifying more than required content of HREs. The designer specifies the coordinate fail safe coordinate of the magnet in the third quadrant. The manufacturer may be more prudent and cost conscious suggesting alloys that will meet the designers’ requirement, and not break the bank. See Fig (7) and Fig (7B).

Figure 7B.  Tridus Int'l KJS Association Model HG-700 V1.1.4

Figure 7.  Tridus International KJS associations Model HG-700 V1.1.4


(5) The user may expect more transparency from the manufacturer. They put subtle or not so subtle pressure on the supplier to increase the yield and use more efficient technology to produce higher yield parts. When the cost of raw material quintuples, the user pays a lot more for the inefficiency of the supplier if the supplier is not able to recycle. Fig (8A) and Fig (8B) show the methods suggested, recycling Nd, some seven years ago. Recycling or reclamation is a de-facto method of increasing the yield today. This has become possible in light of the fact that the price of Nd and Dy has increased at least eight folds since 2004.

Figure 8A.  Recover of Magnet Sludge

Figure 8B.  Recovery of Nd from magnet sludge using NdF3

Figure 8.  Recovery of Nd from magnet sludge - using Nd3 hydrolysis


(6) Recycling becomes the mantra of the end user. They persuade the supplier to do the latter and share the return.

(7) Use of Dysprosium Grain Boundary Diffusion (GBD) is a technology that is somewhat promising. Nevertheless the cost of added process to accomplish the latter still needs to come down to make the process economically feasible.
This process does use much lower content of Dy and the magnet can undergo the diffusion process selectively. For example a motor magnet can be diffused only in the area of the arc where it is more prone to the armature reaction.

The effect of (GBD) has much lower effect on reducing the remanence of the magnet. GBD requires a much lower content of heavy rate earth to achieve the same level of coercivity as achieved with Dysprosium used in the melt process. Anecdotally this could be half or less than half of Dysprosium used in conventional cast.

If the purpose of GBD is to reduce the usage of Dysprosium, this process is a bona-fide process. If the goal is to reduce the overall cost of manufacturing, there is some cloud over this issue at this point in time.



[1] Yukata Matsuura: Current Status of NdFeB Magnets in Japan-China Magneitcs Shanghai, September 2006. China Magnetics , September 19-21, 2006 – Sofitel Hyland Hotel – Shanghai, China
[2] Y. Matsuura, S. Hirosawa, H. Yamamoto, S. Fujimura and M. Sagawa, “Magnetic Properties of the Nd sub 2 (Fe sub 1--x Co sub x ) sub 14 B System,” Appl.Phys, Vol. 46, no. 3, pp. 308-310, February. 1985
[3] M. Tokunaga, M. Tobise, N. Meguro, and H. Harada, “Microstructure of R-Fe-B sintered magnet,” IEEE Transactions on Magnetics, vol. 22, no. 5, pp. 904-909, September 1986
[4] S. F .Cheng, V. K. Shina, Y. Xu, J. M. Elbicki, E. B. Boltich, W. F. Wallace, S. G. Sankar, and D. E. Laughlin, “Magnetic and structural properties of SmTiFe11-xCox alloys,” Journal of Magnetism and Magnetic Materials, vol. 75, no. 3, pp. 330-338, December, 1988.
[5] J. K. Chen, G. Thomas- Al Substitution in NdFeB Magnets- High Performance Permanent Mag, Mater. Res. Soc. Symp. Proc., vol. 96, 1987.
[6] M. H. Ghandahari, J. Fidler, “Microstructural evidence for the magnetic surface hardening of Dy2O3-doped Nd15Fe77B8 magnets,” Material Letters, vol. 5, no. 7-8, pp. 285-288, July, 1987.
[7] Karl J. Strnat, H. Mildrum, T. K. Tran, “Sintered Nd FeB Magnets Modified with Holium and Cobalt,” The 10th International Workshop pm RE Magnets, vol. 1, pp.523, May 1989.
[8] J.Fedler and K.G.Knoch - The Influence of Dopants on Microstructure and Coercivity of NdFeB Magnets, The Proceedings of the Tenth International Conference on Rare Earth Magnets, 16-19 May 1989.
[9] Private dialogs with Y.Matsurra – in several conferences re Dysprosium Grain Boundary Diffusion- including during Magnetics 2011, February 28- March 2, 2011, San Antionio Texas , USA
[10] Naoko Oono, Massato Sagawa, R.Kasada, H.Matsui , “Production of High Performance Sintered NdFeB magnets by Grain Boundary Diffusion Treatment With Dysprosium–Nickel–Aluminum Alloy”,Journal of Magnetics and Magnetism ,Vol.323, no.3-4, pp.297-300, February 2011.
[11] Shinetsu Rare Earth Magnets – A Commercial Narrative on Grain Boundary Diffusion of Dysprosium – Shinetsu Rare Magnets- A Commercial Data Sheet – printed June 10, 2009.
[12] Lanthanide Resources and Alternatives- Oakdene Hollins Research & Consulting- A Report for Dept. for Transport and Dept. for Business, Innovation and Skills- March 2010 - UK

Tuesday, July 05, 2011