Cost effective modification of SmCo₅-type alloys

High demand predictions for PM materials are difficult to be met by the current supply chain, mostly due to the scarcity of some critical raw materials.^1–3 A safe solution to this may arise from development of new compounds via targeted modification of established materials, for example reducing Co content. In the same scheme, we can investigate the possibility of using as substitutes raw materials with abundance greater than their demand, like Ce and La, a path already utilized in the case of Nd-Fe-B based PM.^4,5 Sm-Co, a system with excellent magnetic properties has been investigated towards reduction of both constituent elements, especially Co but also for replacing Sm with other rare-earth (RE) atoms.^6–14 Specific properties are positively affected and regarding the supply cost, overall merit of the material is improved. Another important field towards development of new materials suitable for PM applications is the theoretical understanding of the underlying physics. Ab-initio methods are widely used, and especially DFT-based calculations are usually successful in RE-TM intermetallic compounds, however, there are still issues to be clarified.^7,15–17

In this work we have examined the effect of Sm-Co replacement for Mischmetal-LaCe₃ and Fe, Ni respectively on the fundamental properties of the SmCo₅ intermetallic, towards a material with better overall merit, including the cost-criticality, for PM applications. We combine theoretical ab-initio calculations and experimental results, with the additional goal to improve the understanding of the fundamental physics of the system.

Samples of nominal composition Sm_1-xMM_xCo_5-y-zFe_yNi_z (x = 0 – 0.75; y = 0.5 – 1.5; z = 0.5 – 1) were prepared by Ar arc-melting, followed by thermal homogenization at 900-1200 K. Alloys were manually ground down to less than 50 μm using a sieve of 50 mesh. Structural characterization of the powders was carried out using X-ray diffraction (XRD) patterns recorded with a SIEMENS D500 diffractometer (Cu-Kα radiation). XRD patterns of magnetically aligned powder samples with the alignment direction normal to the sample holder were recorded in order to study the magnetic anisotropy of the compounds. The magnetization measurements were performed on magnetically oriented powders fixed in epoxy resin at room temperature using a PAR 155 vibrating sample magnetometer (VSM) in applied fields up to 2 T. The composition was determined by a scanning electron microscope (SEM), equipped with an electron microprobe analyzer (EDAX).

DFT (ab initio) calculations were performed by using VASP package,¹⁸ with the PAW version of the PBE pseudopotentials.^19–21 All calculations were fully non-colinear, with spin-orbit coupling and aspherical corrections inside the PAW spheres and to Kohn-Sham potential being all explicitly considered. For the treatment of strongly correlated electrons, the rotationally invariant DFT+U approach was employed.²² A value of U_eff = U-J = 4.7 eV was used for the 4f electrons of Sm-Ce, along with a value of U_eff = 2.22 eV for 3d electrons of La, Co, Fe, Ni.^23,24 Kinetic energy cutoff for the wave functions was 500 eV. Sm_1-xMM_xCo_5-y-zFe_yNi_z structures were represented by using 96 atom, 2⨯2⨯4 supercells derived from the primitive cell of the hexagonal P6/mmm, CaCu₅-type structure of SmCo₅. It has been previously computationally demonstrated that the substitution of a single Co atom with a Ni one is energetically preferable to occur on the 3g sites; if more Co atoms are substituted with Ni, then one substitution per unit cell occurs at a 3g site and the rest at 2c.²⁴ Fe atoms favor the occupation of 3g sites in the CaCu₅-type structure.^10,16 Therefore, we considered substitution of Co atoms with Fe ones on the 3g sites and with Ni ones at the 2c sites. In all cases calculations simulated the stoichiometries from first principles, independently of the experimental results.

In Fig. 1 X-ray powder patterns for all stoichiometries are presented; in the case of Sm_0.5MM_0.5Co₄Ni a pattern of epoxy-oriented sample is also included. In almost all cases the desired CaCu₅-type structure is formed, with obvious exception of Sm_0.3MM_0.7Co₃FeNi compound. Phenomenological observations were confirmed by Rietveld analysis, Table I. We have assumed that Fe atoms replace Co in 3g positions and Ni atoms in 2c positions as in the case of ab-initio calculations and we have allowed for transition metal content which in all cases but Sm_0.3MM_0.7Co₃FeNi was below 2 wt.%; in the latter case the 1:5 and 2:7 mass fractions were almost equal (41 and 49 wt.%) while excess transition metal accounts for the rest. The basal plane unit cell parameter is slightly smaller than the reported values for SmCo₅ (a = b = 0.5002 nm; c = 0.3961 nm)²⁵ while the perpendicular is slightly larger. This results in a larger c/a ratio. Materials retain the uniaxial character of the magnetocrystalline anisotropy, as observed by using epoxy-oriented samples. In Fig. 1 the corresponding plot for Sm_0.5MM_0.5Co₄Ni is presented as an example, it is evident that only the (0 0 1) and (0 0 2) reflections are present in the oriented sample. The successful alloying of the materials was confirmed by scanning electron microscopy and EDAX analysis (Fig. 2). The refined composition as obtained is close to the starting composition, for example for spot nr. 10 (Fig.2) was found the stoichiometry Sm_0.7MM_0.3Co_2.95 Fe_1.04Ni_0.99.

The magnetic properties of the materials were investigated by means of RT isotherm loops in fields up to 2 T in powder samples of size 50 μm or less with random orientation. The corresponding curves are presented in Fig. 3 and magnetization is included in Table I. In Fig. 3 we distinguish between x = 0.5 samples and the rest. In the first case, magnetization is mostly reduced by Ni content, the compound with less Ni (y = z = 0.5) presents the highest magnetization. For the other two samples where Ni content is the same the material with no Fe present lower magnetization than itscounterpart with 20% replacement of Co by Fe. These results are consistent with previous reports in the literature.^7,16,24 Doping with Ni seems obscure due to its negative effect on magnetic properties, however, there are reports which connect the stability of the Fe-doped compound with Ni inclusion as well.²⁶ The rest of the samples present unexceptional magnetization values, below 70 Am²/kg; Sm_0.7MM_0.3Co₃FeNi is close to Sm_0.5MM_0.5Co₃FeNi, as expected from the similar stoichiometry. The value for Sm_0.3MM_0.7Co₃FeNi is presented as reference since the it is affected by sample’s structural inhomogeneity. Finally, SmCo_2.5Fe_1.5Ni compound has very low magnetization, although in this case the negative -in general-effect of MM in absent. The results of the calculated structural parameters, total magnetization per formula unit and formation enthalpy are presented in Table II. The multiple substitutions of Sm with the MM alloy and Co with Fe and Ni were found to lead to a 2.5-3.4% increase of the in-plane lattice constant and a 9.4-11.7% increase of the c/a ratio with respect to the calculated values for the SmCo₅ compound. Moreover, the introduction of the substituents was found to lead to the presence of a small in-plane structural distortion (0.27-1.82%) which can be detrimental for the magnetic properties; a is not exactly equal to b within the hexagonal plane.²⁷ The introduction of MM at 25% concentration was found to lead to the smaller value of in-plane distortion but also to a reduction of the total magnetization. The substitution of Sm with MM was found to lead to a reduction of the total magnetization by ∼1 μ_B/f.u. in all cases.

Also, from Table III, it is evident that MM (LaCe₃) tends to slightly destabilize the structure, as Ce atoms present lower values of energy per atom than Sm in their groundstate structures. A low energy per atom value in this context can be interpreted as a measurement of the likelihood of an atom to “prefer” to form its corresponding groundstate structure phase, rather than an alloy or an intermetallic compound. The introduction of Ni significantly increases the stability of the structures, as it leads to a reduction of the formation enthalpies with respect to SmCo₅, due to its higher value of energy per atom in its groundstate structure, compared to Co, Table III.

The average values of magnetic moments per each atom type of the rare earth and transition metal elements are shown in Table IV. Strong antiferromagnetic alignment of the Ce atom is evident and can explain the decrease of the total magnetic moments by ∼1 μ_B/f. u. in the MM containing stoichiometries; Ce has a 4f electron with itinerant character which interacts with Co 3d electrons, modifying their weight at the Fermi level.²⁸ The introduction of both transition metal substitutes leads to a slight decrease of the average atomic magnetic moments of Co atoms by 0.1-0.2 μ_B/atom. However, the addition of Fe is strongly beneficial for the hard magnetic properties as its magnetic moment appears ∼1.75 times larger than the one of Co. Its addition, though, is known to destabilize the structure, fact which agrees with its corresponding calculated energy per atom in its groundstate structure. This destabilization appears to be eliminated with the addition of Ni, at the cost of reducing the total magnetization, as Ni atoms present very low atomic magnetic moments. The presence of Fe at concentrations ≥ 1 per formula unit appears also to be associated to an increase of the magnetic moments of Ni atoms.

Overall, according to the predictions of our ab initio simulations, the SmCo_2.5Fe_1.5Ni stoichiometry presents a large value of total magnetization which is comparable to the respective value of SmCo₅, with the partial substitution of Co with both Fe and Ni leading to a favorable interplay between stability and magnetization and the additional benefit of the reduction of the content of Co, which is regarded as a critical material. Further substitution of the also critical Sm with the MM alloy can lead to a further reduction of cost by reducing raw material and processing expenses, but seems to inevitably lead to a drop of the total magnetization by ∼ 1 μ_B/f. u.

In this work we present structural and magnetic properties of nominal composition Sm_1-xMM_xCo_5-y-zFe_yNi_z (x = 0 – 0.75; y = 0.5 – 1.5; z = 0.5 - 1) from both experimental results and ab-initio DFT calculations. All studied stoichiometries with MM content up to 50% can be synthesized in bulk form with conventional metallurgy techniques.

Ni probably has a positive effect in structure stabilization as also depicted by ab-initio calculations, at the cost of weakening the overall magnetization. In both structural and magnetic properties ab-initio calculations and experimental results are in good agreement, calculated magnetization is larger than experimental results; suggested DFT parameters are suitable for RE-TM intermetallics. The suggested substitutional approach may provide PM suitable for some applications with reduced cost and criticality compared to basic SmCo₅ system.

S. Giaremis acknowledges support for the research work by the Hellenic Foundation for Research and Innovation (HFRI) under the HFRI PhD Fellowship grant (Fellowship Number: 962). Furthermore, this work was supported by computational time granted from the Greek Research & Technology Network (GRNET) in the “ARIS” National HPC infrastructure under the project NOUS (pr010034) and the Aristotle University of Thessaloniki (AUTh) HPC Infrastructure and Resources.

The authors have no conflicts to disclose.

The data that support the findings of this study are available from the corresponding author upon reasonable request.