RESEARCH NEWS

FePt Hard Magnets

By Oliver Gutfleisch,* Julia Lyubina, Karl-Hartmut Mller, Ludwig Schultz
The FePt alloys have recently attracted considerable attention due to their excellent intrinsic magnetic, chemical and mechanical properties. Their possible usage ranges from permanent magnets for special applications (e.g. in micro-electro-mechanical systems, magnetic MEMS, and in aggressive environments) to ultra-high density magnetic storage media. The article describes general aspects concerning the phase formation and magnetic properties of materials based on the L10 FePt phase. Both thin film and bulk approaches are considered. The production of bulk nanocrystalline Fe100-xPtx powders by mechanical alloying and subsequent annealing is described. Various combinations of phases, away from thermodynamic equilibrium, have been obtained using this technique.

1. Introduction
The compounds FePt and CoPt (with the tetragonal L10-type structure; space group P4/mmm) are of fundamental interest because their magnetic anisotropy based on 3d and 5d electrons is as large as that of the well-known rare earth transition metal high performance magnets (RPMs) based on the 3d-4f-systems (see Tab. 1 for a comparison with other magnet materials[13]). The large magnetic anisotropy of the 3d-5d ferromagnets is due to the large spin-orbit coupling in the 5d-element. Magnets made of the L10 phase have attracted much attention recently because of their potential usage in ultra high-density data storage[4] and permanent magnets for special applications. Distinct advantages of these compounds are, as opposed to the RPMs, that they are very ductile and chemically inert. Selforganised magnetic arrays (SOMA) of dispersed L10 particles (diameter ~ 5 nm) are expected to be superparamagnetic only at elevated temperature. Therefore they are considered as candidates for future thermally stable magnetic recording media with densities as high as 1 Tbit/in2.[5] In bulk form, their high cost limits L10 based permanent magnets to very
Table 1. Some intrinsic magnetic properties for selected materials. Tc: Curie temperature, Ms: spontaneous magnetisation, HA: anisotropy field, d: domain wall width, Dc: singledomain particle diameter (data are from [1-3]). phase Co CoPt FePt FePd Sm2Co17 SmCo5 symmetry hexagonal tetragonal tetragonal tetragonal rhombohedral hexagonal Tc [K] 1390 720 750 760 1190 1020 585 l0 Ms [T] 1.81 1.0 1.43 1.39 1.22 1.07 1.61 l0 HA [T] d [nm] Dc [nm ] 0.76 12.3 11.5 3.5 6.5 39 7.6 14 7.4 6.3 11.5 8.6 3.6 3.9 68 1000 560 330 490 1530 214

specialised applications such as in magnetic micro-electromechanical systems (MEMS) or in dentistry, where they are used as magnetic attachments to retain prostheses in the oral cavity. According to the Fe-Pt or Co-Pt phase diagrams, (close to) equiatomic FePt and CoPt alloys exhibit a typical disorderorder transformation from the disordered face-centred cubic A1-type structure (with low magnetic anisotropy) to the L10type superstructure (with high magnetic anisotropy) when the temperature is lowered below 1300 C and 825 C, respectively. In the L10 phase the cubic symmetry is broken due to the stacking of alternate planes of the 3d element (Fe) and the 5d element (Pt) along [001] direction. According to the equilibrium Fe-Pt phase diagram (Fig. 1[6]), in alloys deviating from the 35 x 55 stoichiometry, formation of the stable cubic superstructures FePt3 (L12) and Fe3Pt (L12) is expected (compare Fig. 2). X-ray diffraction is one of the widely used methods for characterising chemical order in binary alloys. As a result of ordering, additional peaks called superstructure peaks appear in x-ray diffraction patterns. The integrated intensity of these superstructure reflections gives a measure for the longrange order. To quantify the degree of chemical order, the following expression for the order parameter S can be defined: S=rPt+rFe1=(rPtxPt)/yFe=(rFexFe)/yPt, where rPt(Fe) is the fraction of Pt(Fe) sites occupied by the correct atom, xPt(Fe) is

Nd2Fe14B tetragonal

[*] Dr. O. Gutfleisch, Prof. J. Lyubina, Dr. K.-H. Mller, Prof. L. Schultz IFW Dresden, Institute for Metallic Materials P.O. Box 270016, D-01171 Dresden, Germany E-mail: o.gutfleisch@ifw-dresden.de

208

2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

DOI: 10.1002/adem.200400183

ADVANCED ENGINEERING MATERIALS 2005, 7, No. 4

Gutfleisch et al./FePt Hard Magnets

RESEARCH NEWS
Fig. 1. Equilibrium phase diagram of the Fe-Pt system (adapted from [6]). Shown schematically are crystallographic structures of the phases in the corresponding phase equilibrium regions.

Fig. 2. The magnetic behaviour and magnetic properties at room temperature of the main phases in the Fe-Pt system (para paramagnetic; ferro ferromagnetic; af antiferromagnetic).

the atom fractions of Pt(Fe), and yPt(Fe) is the fraction of Pt(Fe) sites.[7] When the order is perfect, the order parameter S reaches unity, while for a completely random atom arrangement S is equal to zero. Thus, the order parameter can reach its maximum (S = 1) only for stoichiometric composition. For nonstiochiometric compositions, S is always less than unity. It has to be noted, that the above definition for S is derived for homogeneous materials. Frequently, however, materials consist of regions having high degree of chemical order and regions being nearly or completely disordered. In this case, as an additional parameter, the volume fraction of the chemically ordered phase, should be used to quantify the order in such inhomogeneous materials. Despite the crystallographic simplicity of the L10 phase and its well-defined magnetic properties, its formation as well as the accomplishment of complete long-range order require sophisticated preparation procedures and are not yet fully understood. Many approaches have been used to obtain the L10 phase.[3,8-14] Typically, the synthesis starts with the A1 precursor. The subsequent annealing to produce the desired L10 phase has to be conciled with problems of grain growth, coalescence or sintering. The result of the disorder-order transformation of bulk samples with stoichiometries close to equiatomic is a nontextured material characterised by a polytwinned microstruc-

ture with antiphase boundaries and different types of twin boundaries. However, the details of microstructure are sensitive to the specific kinetic regimes characterised by annealing temperature and/or cooling rates.[811] The polytwinned microstructure in relatively coarse grained materials is beneficial for achieving a hard magnetic material as the mentioned planar defects act as pinning centres for magnetic domain walls. The coercivities achieved for these conventionally processed Fe-Pt alloys usually do not exceed 0.45 T.[8,12] Moreover, the macroscopic isotropy of these materials results in a relatively low remanence Mr. Improved coercivity values (up to ~ 1 T) were obtained for alloys subjected to mechanical deformation followed by annealing at temperatures below the order-disorder transformation.[3,5,10,11,15] This processing results in so-called combined solid state reactions that take place in solids in which thermodynamical equilibrium is approached by more than one elementary reaction.[16] Frequently, the combined reactions occur if prior to the heat treatment a high density of crystal defects is introduced into the material. In the case of heavily deformed Fe-Pt alloys annealing leads simultaneously to the decrease of defect density, nucleation and growth of the new phases concurrent with the ordering process. As a result of the combined reactions, the microstructure is drastically changed and material mostly consisting of fine L10 FePt grains is obtained. In thin film-form, the L10 FePt phase can be directly prepared by deposition onto heated substrates.[17] Ordered FePt thin films have been prepared as well by subsequent annealing of multilayers produced by sequential deposition of separate Fe and Pt layers of thickness in the order of several nanometers.[18] Nanoparticulate systems can be produced by chemical synthesis[19] leading to self-organised magnetic arrays (SOMA) or by a gas-phase based process.[20] Cold deformation[15] is used to make FePt foils and electrodeposition[21] can be employed to prepare layers of several microns, both suitable

ADVANCED ENGINEERING MATERIALS 2005, 7, No. 4

http://www.aem-journal.de

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

209

Gutfleisch et al./FePt Hard Magnets for MEMS applications. Because of the ductility of the L10 phases, the conventional powder-metallurgical route, often used for producing permanent magnet materials, is not easily adaptable here. However, mechanical alloying of elemental Fe and Pt powders, performed at liquid-nitrogen temperature, has been successfully used to prepare fine-grained Fe100-xPtx (x = 4060) powders.[10] The mechanisms of phase formation and ordering, and thus the final magnetic properties, differ depending on the approach used to obtain the L10 phase.[811,13,14,20,2224] Our research focuses on the improved understanding of thermodynamics and phase formation, changes in stochiometry and the enhancement of ordering kinetics by e.g. alloy additions and, at the same time, restraining grain growth. As the performance of conventional isotropic bulk permanent magnets is rather limited, we also investigate two-phase materials with exchange coupled grains of the magnetically highly anisotropic L10 FePt phase and the less anisotropic L12 Fe3Pt phase with larger spontaneous magnetisation (see Fig. 2). This coupling results in enhanced remanences and higher energy products (BH)max due to so-called exchange-spring behaviour.[25] As a model system, we have chosen nanocrystalline FePt powders produced by mechanical ball milling and subsequent annealing. However, the obtained data should be equally valuable for other nanoparticulate FePt systems. level. As the milling time increases, the amount of the A1 FePt phase increases, while the fractions of a-Fe and Pt are reduced. The lamellae thickness and the grain size of the phases are reduced with milling time and after 7 h of milling the powders are predominantly of A1 FePt structure. Heat treatment of the as-milled powders induces the aforementioned combined solid-state reactions. In this case, the ordering transformation involves nucleation and growth of ordered phases assisted by healing out deformation-induced defects. Furthermore, the ordering kinetics was found to be enhanced for the fine lamellar structure of Fe/FePt(A1)/Pt.[24] Annealing of the 2 h milled Fe100-xPtx (x = 39.5; 45; 50; 55; 60) powders with the lamellar microstructure induces immediate formation and ordering of L10 FePt, the maximum amount of which is about 85 vol.% for Fe50Pt50 after annealing at 450 C/48 h. As an example, Figure 4 shows the x-ray diffraction pattern of this powder. Rietveld refinement of the x-ray data reveals the presence and allows the quantification of additional phases such as A1 FePt, Fe3Pt (L12) and FePt3 (L12), additionally to the L10 phase. The amount of the disordered A1 FePt phase remains almost unchanged (610 vol.%), whereas the fraction of the Fe3Pt and FePt3 phases varies with the Pt concentration x. In Fe-rich samples, the Fe3Pt phase in the amount of 38 vol.% (Fe60.5Pt39.5) is present, whereas in the Ptrich powders FePt3 phase is observed (40 vol.% for Fe40Pt60). The fractions of the Pt(Fe) sites occupied by the right atom rPt(Fe) in the L10 FePt phase, extracted from the Rietveld refinement, were used for the determination of the long-range order parameter S in L10 FePt. In the heat-treated Fe100-xPtx (x = 39.5; 45; 50) powders S was found to be approximately 0.9, whereas for the Pt-rich powders (x = 55; 60) it is close to unity. Thus, heat treatment of the as-milled Fe100-xPtx powders with the lamellar microstructure results in the formation of the chemically highly ordered equiatomic L10 FePt phase together with additional phases. This multiphase state is different from the state of (global) thermodynamic equilibrium, which is a homogeneous nonequiatomic L10 phase with an order parameter of, consequently, less than unity. Probably the observed metastable state can be considered as a state of local equilibrium provoked by the heterogeneous as-milled precursor material.

RESEARCH NEWS

2. Results
Bulk nanocrystalline hard magnetic Fe100-xPtx powders have been prepared by mechanical ball milling of elemental Fe and Pt powders at 77 K and subsequent annealing at moderate temperatures as described in.[10] This method gives the possibility to obtain various combinations of phases for concentrations, where according to the equilibrium phase diagram, only one phase region should exist. As an example, Figure 3(a) presents the microstructure of an as-milled powder particle (typical particle size ~ 100 lm) of nominal composition Fe60.5Pt39.5, showing a finely divided layered structure of the constituent elements. After milling for only 2 h, the grain sizes of a-Fe and Pt are substantially reduced (down to ~ 20 nm) compared to the starting powders with typical grain sizes of 80 lm. Rietveld analysis[26] of x-ray diffraction data (not shown here) reveals that the A1 FePt phase is present in the as-milled powders, which is due to alloying of the elements on an atomic

Fig. 3. SEM image (backscattered electron mode) of a particle of Fe-rich FePt powder milled at liquid nitrogen temperature for 2 h (left) and milled for 4 h and annealed at 450 C/48 h (right).

Fig. 4. X-ray diffraction profiles of the Fe50Pt50 powder milled at 77 K for 2 h and annealed at 450 C/48 h. The vertical bars represent the Bragg reflection positions of the observed phases (from top to bottom: L10, A1, Fe3Pt and FePt3). Marked by arrows are the superstructure reflections of the L10 FePt phase.

210

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

http://www.aem-journal.de

ADVANCED ENGINEERING MATERIALS 2005, 7, No. 4

Gutfleisch et al./FePt Hard Magnets The combination of the highly ordered L10 FePt and L12 Fe3Pt phases is attractive from the point of view of improvement of hard magnetic properties. Adding the less anisotropic Fe3Pt phase with its high magnetisation enhances the remanence, while the high magnetic anisotropy of the L10 FePt phase ensures a sufficiently large coercivity. However, in order to really profit from the combination of the beneficial properties of the constituent phases the demagnetisation behaviour of the composite material must be magnetically single phase. Therefore, the soft magnetic phase should be exchange-coupled to the hard magnetic one and the microstructure has to be on the nanoscale.[25] Remanence enhancement, single-phase demagnetisation behaviour and rather steep recoil loops are general criteria for the presence of exchange-spring behaviour. In fact, the large reversible portion of the demagnetisation curve corresponds to an energy recovery, as in the case of a mechanical spring, from which the term exchange-spring magnet is derived. A very fine lamellar structure of L10 FePt and Fe3Pt was achieved by heat treatment of Fe100-xPtx (x=40; 45; 50) powders milled for 4 h.[27] As can be seen in Figure 3(b), after heat treatment at 450 C for 48 h the intermetallic compounds FePt (L10) and Fe3Pt (L12) have been formed in Fe-rich Fe100-xPtx powders. The distribution of the chemically highly ordered L10 FePt and Fe3Pt phases is sufficiently fine to yield effective exchange interactions between hard and soft magnetic grains. The Fe50Pt50 powder is essentially single-phase (about 98 vol.% of L10 FePt was detected, the rest being A1 FePt). Figure 5 shows the demagnetisation curves of magnets produced by embedding the heat-treated Fe100-xPtx powders in epoxy resin. The coercivity and remanence of such magnets reach from l0JHc=0.44 T, l0Mr=1.05 T (for x = 40) to l0JHc=0.87 T, l0Mr=0.77 T (for x = 50), where the values of Mr are normalised to the volume of the powder in the bonded magnet. For the Ferich samples, a spring-magnet behaviour has been clearly detected (Fig. 5). As a consequence, a high energy product (BH)max of 121 kJ/m3 was obtained for the Fe55Pt45 powder.

3. Summary
The formation of the well-ordered L10 FePt phase requires sophisticated preparation conditions including heat treatments at elevated temperatures (above ~ 400 C). In particulate thin film samples this leads to unavoidable grain growth and sintering, which result in microstructures not useful for magnetic recording. Therefore, the production of monodisperse nanoparticles (diameter ~ 5 nm), which carry a spatially and temporally stable magnetisation, as well as the generation of patterned and textured media still remain a demanding task. In bulk L10 based magnets, combined solid state reactions, such as decrease of defect density and grain growth, are beneficial for the enhancement of the ordering kinetics. It was shown that mechanical alloying at 77 K followed by annealing at moderate temperatures results in large amounts of the highly ordered L10 FePt phase and, at the same time, slow grain growth. This preparation technique gives a possibility to obtain various combinations of phases for concentrations, where according to the equilibrium phase diagram, only one phase region exists. Nanocomposite FePt/Fe3Pt bulk magnets with high energy products (BH)max have been produced. In order to further improve the permanent magnet performance, densification of the powders, as well as crystallographic alignment of the L10 grains, yielding textured nanocomposites, should be achieved. Received: November 19, 2004 Final version: December 06, 2004

RESEARCH NEWS

Fig. 5. Demagnetisation curves of the Fe100-xPtx (x= 40; 45; 50) powders annealed at 450 C for 48 h in dependence on Pt concentration x. Recoil loops are shown for Fe60Pt40. Remanence enhancement, single-phase demagnetisation behaviour and steep recoil loops characterised by fair reversibility indicate exchange-spring behaviour.

[1] R. Skomski, J. M. D. Coey, Permanent Magnetism, IoP Publishing, Bristol 1999. [2] N. I. Vlasova, G. S. Kandaurova, N. N. Shchegoleva, J. Magn. Magn. Mater. 2000, 222, 138. [3] O. A. Ivanov, L. V. Solina, V. A. Demshina, L. M. Magat, Phys. Metall. Metalloved. 1973, 35, 92. [4] S. Sun, C. B. Murray, D. Weller, L. Folks, A. Moser, Science 2000, 287, 1989. [5] T. J. Klemmer, C. Liu, N. Shukla, X. W. Wu, D. Weller, M. Tanase, D. E. Laughlin, W. A. Soffa, J. Magn. Magn. Mater. 2003, 266, 79. [6] T. B. Massalski (editor), Binary Alloy Phase Diagrams, ASM International, Ohio 1990, p. 1752. [7] B. E. Warren, X-ray Diffraction, NY, Dover Publications, Inc 1990. [8] B. Zhang , W. A. Soffa, Phys. Stat. Sol. (a) 1992, 131, 707. [9] Q. F. Xiao, E. Brck, Z. D. Zhang, F. R. de Boer, K. H. J. Buschow, J. Alloys Comp. 2002, 336, 41. [10] J. Lyubina, O. Gutfleisch, K.-H. Mller, L. Schultz, N. M. Dempsey, J. Appl. Phys. 2004, 95, 7474. [11] J. Lyubina, O. Gutfleisch, K.-H. Mller, L. Schultz, J. Magn. Magn. Mater., in press. [12] K. Watanabe, H. Masumoto, Trans. Jpn. Inst. Met. 1983, 24, 627. [13] A. Cebollada, R. F. C. Farrow, M. F. Toney, in Magnetic Nanostructures (Editor: H. S. Nalwa), Stevenson Ranch, CA: American Scientific 2002, p. 93.

ADVANCED ENGINEERING MATERIALS 2005, 7, No. 4

http://www.aem-journal.de

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

211

Gutfleisch et al./FePt Hard Magnets [14] R. A. Ristau, K. Barmak, L. H. Lewis, K. R. Coffey, J. K. Howard, J. Appl. Phys. 1999, 86, 4527. [15] N. H. Hai, N. M. Dempsey, M. Veron, M. Verdier, D. Givord, J. Magn. Magn. Mater. 2003, 257, 139. [16] E. Hornbogen, Metall. Trans. A 1979, 10A, 947. [17] M. Weisheit, L. Schultz, S. Fhler, J. Appl. Phys. 2004, 95, 7489. [18] J. P. Liu, C. P. Luo, Y. Liu, D. J. Sellmyer, Appl. Phys. Lett. 1998, 72, 483. [19] H. Zeng, J. Li, J. P. Liu, Z. L. Wang, S. Sun, Nature 2002, 420, 395. [20] B. Rellinghaus, S. Stappert, M. Acet, E. F. Wassermann, J. Magn. Magn. Mater. 2003, 266, 142. [21] K. Leistner, E. Backen, B. Schupp, M. Weisheit, L. Schultz, H. Schlorb, S. Fhler, J. Appl. Phys. 2004, 95, 7267. [22] F. E. Spada, F. T. Parker, C. L. Platt, J. K. Howard, J. Appl. Phys. 2003, 94, 5123. [23] K. Barmak, J. Kim, S. Shell, E. B. Svedberg, J. K. Howard, Appl. Phys. Lett. 2002, 80, 4268. [24] J. Lyubina, O. Gutfleisch, R. Skomski, K.-H. Mller, L. Schultz, Scripta Mater., in press. [25] E. F. Kneller, R. Hawig, IEEE Trans. Mag. 1991, 27, 3588. [26] R. A. Young (editor), The Rietveld Method, Oxford University Press, Oxford, UK 1993. [27] J. Lyubina, O. Gutfleisch, K.-H. Mller, L. Schultz, Phys. Rev. B, submitted.

RESEARCH NEWS

______________________

212

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

http://www.aem-journal.de

ADVANCED ENGINEERING MATERIALS 2005, 7, No. 4