electronic structure of light rare earth permanent magnets...2019/02/01  · electronic structure of...

Electronic structure of light rare earth permanent

magnets

H. Akai

Institute for Solid State Physics, University of Tokyo

Kashiwa-no-ha, Kashiwa, Chiba 277-8581

Despite intensive efforts to develop perma-nent magnets whose performances exceed thatof the Nd2Fe14B magnet, no essentially novelmagnet has been developed so far. In this situ-ation, one might question whether it is possibleat all to obtain a permanent magnet materialthat is superior to the currently available max-imum performance.

To answer this, one should have a perspec-tive on the possible maximum performance ofpermanent magnet materials. One of way todo this is to estimate the upper limits of mag-netization JS, Curie temperature TC, and low-est order uniaxial magnetic anisotropy con-stant K1, which make a prospect about theperformance of magnets. We discuss each ofthese quantities on the basis of the results ob-tained through first-principles calculation.

The discussions are based on all-electronelectronic structure calculations performedwithin the local density approximation (LDA/GGA) of density functional theory (DFT).Machikaneyama (AkaiKKR) KKR-CPA pack-age [1] was used, and for the calculation of TC,Liechtenstein’s method [2] was employed.

Figure 1 shows the calculated magnetizationJS of 3d elements as a function of the latticeconstant a and atomic number Z. The frac-tional atomic number of a fictitious atom isused. The number of total electrons per atomis equal to Z. The crystal structure is assumedbcc. A prominent feature is that it has a dome-like structure appearing around a = 2.65 Aand z=26.4, where JS takes the maximum

value of 2.66 T. It is pointed out that thisis related to the fact that in the bcc struc-ture, the interatomic distance between nearestneighbor pairs becomes small, forming a con-siderable bonding–antibonding splitting with apseudo gap in between. Unfortunately, the lat-tice constant a = 2.65 A is 7 % too small com-pared with the equilibrium lattice constant ofbcc Fe. Contrary to the general behavior of themagnetic moment that increases as the volumeincreases, the magnetic polarization increaseswith decreasing a up to some point where themagnetic state collapses.

Figure 1: Saturation magnetic polarization JS

of the system plotted against the lattice con-stant and the fictitious atomic number[3].

Magnetic polarization takes on a large valueat one of the corner points in the Z-a plane,

Z = 25 and a = 3.2 A, but this is not real. Inthis region, the antiferromagnetic state is morestable than the ferromagnetic state. Combin-ing this fact with the information given byFig. 1, we may conclude that a large JS isexpected only in the vicinity of the dome-likestructure, and the upper limit of JS would notexceed ∼ 2.7 T.

Figure 2 shows the behavior of magnetictransition temperature TC as a function of Z

and a. Here, we again see a dome-like struc-ture near Z = 26.5 and a = 2.9 A. This posi-tion approximately coincides with the positionof the similar dome-like structure in JS. Thisindicates that if Z = 26.5 and a = 2.9 A isforced by crystal structure, chemical composi-tion, pressure, temperature, etc., JS ∼ 2.7 Tis achieved. TC drops rapidly toward the cor-ner in the Z-a plane, Z = 25 and a = 3.2 A,where TC becomes negative, meaning that theantiferromagnetic state should be the groundstate. Now, we may say that the upper limit ofTC is ∼ 2000 K (if fcc structure were assumed,the upper limit would be ∼ 1500 K).

Figure 2: Magnetic transition temperature TC

of the system plotted against the lattice con-stant and the fictitious atomic number[3].

The main origin of magnetocrystallineanisotropy is spin orbit coupling. For Sm

(Sm3+ in Sm-type Sm element), assuming thatthe orbitals are firmly bound to the lattice, theupper limit of the magnetic anisotropy con-stant K1 estimated from the strength of thespin-orbit coupling , together with the valuesof 〈L〉, is as high as ∼ 1000 MJm−3. The up-per limit of K1 for other lanthanides, if scaledby the value of L, also would be similar tothat of Sm. However, K1 of rare earth mag-netic materials is one to three orders of magni-tude smaller than this values. This is becausethe anisotropy in lattice geometry is not largeenough to firmly bind the orbital to the lat-tice: the 4f electron density rotates in linewith magnetization to some extent. There-fore, the upper limit of K1 is bound by thelattice geometry. Also, the magnetic momentcarried by 3d orbitals of transition metal ions isonly weakly coupled to the 4f orbitals of rareearth ions (through 3d–5d indirect and 5d–4f

direct exchange coupling), the latter producinga large magnetic anisotropy. Accordingly, themagnetization is rather loosely bound to thelattice. The effect is in particular prominent athigh temperature T & (2/3)TC, where the cou-pling between 3d and 4f becomes progressivelyweaker because of the thermal fluctuation.

In conclusion, calculations based on densityfunctional theory conclude that the plausibleupper limits of saturation magnetic polariza-tion, magnetic transition temperature, and themagnetocrystalline anisotropy constant of per-manent magnet materials could be ∼2.7 T,∼2000 K, and ∼1000 MJm−3.

References

[1] H. Akai, AkaiKKR, http://kkr.issp.

u-tokyo.ac.jp/ (2002).

[2] A. I. Liechtenstein, M. Katsnelson,V. Antropov, V. Gubanov, J. Magn.Magn. Mater. 67 (1987) 65.

[3] H. Akai: Scripta Materialia (2018), inpress.