Assessment of off-stoichiometric Zr_33-xFe_52+xSi₁₅ C14 Laves phase compounds as permanent magnet materials

Until the 2010s, the Laves phase compounds not-containing the critical rare earth elements had rarely attracted attention as potential hard magnetic materials.¹ The current incentive to decrease reliance of permanent magnets on critical elements have brought into consideration the hexagonal MgZn₂-type (C14) compounds found in the (Ti,Zr)Fe₂,² Zr(Fe,Si)₂³ and Hf(Fe,Sb)₂⁴ alloys. Although the binary ZrFe₂ compound is a ferromagnet with a high Curie temperature (584 K to 733 K depending on the composition⁵), its cubic MgCu₂-type (C15) crystal structure is unfavorable for permanent magnet applications. A hexagonal MgNi₂-type (C36) Laves phase was also observed in the binary Zr-Fe alloys;⁵ this phase is reportedly stable between 1513 K and 1618 K in the off-stoichiometric composition Zr_26.6Fe_73.4. A partial Si substitution for Fe is known to stabilize the C14 structure; in the alloys homogenized at 1273 K, this structure was observed in Zr_33.3Fe₅₀Si_16.7⁶ (it was also found in the as-prepared alloys⁷ and in the alloys homogenized at 1373 K⁸). According to Zamora et al.,⁷ the Si substitution for Fe in the stoichiometric Zr(Fe,Si)₂ alloys is very detrimental for the saturation magnetization M_s. On the other hand, a partial replacement of the Zr atoms with the Fe atoms in the C15 ZrFe_2+δ off-stoichiometric compounds increases not only the fraction of the magnetic atoms, but also – due to enhanced magnetic moments of the Fe atoms occupying the Zr sites – the average magnetic moment per Fe atom.⁹ This work was initially aimed at the assessment of both the C36 and C14 hexagonal Zr–Fe–(Si) compounds as materials for the permanent magnets. Attempt to synthesize the C36 structure was not successful, but the C14 compound was prepared and characterized in stoichiometric and off-stoichiometric alloys.

Two series of alloys, Zr₂₇Fe_73-wSi_w (w = 0, 5, 10, 15) and Zr_33-xFe_52+xSi₁₅ (x = 0, 2, 4, 6, 8, 9, 10, 11), were prepared by arc-melting under argon elemental Zr (99.2%), Fe (99.97%) and Si (99.96%). The alloys were re-melted four times to ensure their homogeneity. The ingots wrapped in molybdenum foil were sealed in argon-filled quartz capsules and placed in a furnace preheated to 1538 K for 5 h; the capsules were then quenched in water. Prior to characterization, surfaces of the ingots were machined off to avoid any contaminants. Densities of the alloys were determined with the Archimedes method. Alloy powders were prepared with a hand mortar. X-ray diffraction (XRD) characterization was performed with a Rigaku Ultima IV instrument using the CuKα radiation both for randomly oriented powders and for powders immobilized with epoxy resin under a magnetic field of 19 kOe. The XRD results were analyzed with Powder Cell software;¹⁰ volume fractions of the crystalline phases were determined through Rietveld refinement. The magnetization data were collected with a Quantum Design VersaLab vibrating sample magnetometer. Thermomagnetic scans were recorded when heating 10 mg alloy pieces in a field of 5 kOe. Oriented powders for isothermal magnetic measurements were immobilized with paraffin wax under a magnetic field of 16 kOe; the magnetization curves were corrected for self-demagnetization using demagnetization factors determined for similarly prepared Fe powders.

The XRD characterization of the binary Zr₂₇Fe₇₃ alloy annealed at 1538 K did not reveal the hexagonal C36 structure (Fig. 1). Instead, the alloy was found to contain two phases, the cubic C15 [the lattice parameter a = 0.7029(1) nm] and approximately 10 vol.% of the cubic Zr₆Fe₂₃ [the Th₆Mn₂₃ type, a = 11.742(1) nm]. Although this phase composition is in agreement with some published equilibrium phase diagrams,¹¹ the more recent studies⁵ suggested that the Zr₆Fe₂₃ is not an equilibrium binary phase, but one that is stabilized by oxygen. Addition of 5 and 10 at.% Si at the expense of Fe decreases the lattice parameters of the two observed phases but does not significantly affect the volume fractions of the phases. At 15 at.% Si, the Laves phase adopts the hexagonal C14 structure; this observation is in a good agreement with earlier studies.^6–8 

The C14 Laves phase compounds were systematically studied for the alloys containing 15 at.% Si and 33 – 21 at.% Zr. Although the C14 structure was the principal phase in all the examined Zr_33-xFe_52+xSi₁₅ compositions, only the near-stoichiometric alloys with x ≤ 2 were free form minority phases; the XRD data for the Zr₃₃Fe₅₂Si₁₅ alloy are shown in Fig. 2. The Zr₆(Fe,Si)₂₃ phase, which is stable in the ternary Zr–Fe–Si system,⁸ was detected in the alloys with x ≥ 4 including the Zr₂₇Fe₅₈Si₁₅ and Zr₂₅Fe₆₀Si₁₅ alloys represented in Figs. 1 and 2, respectively. The amount of the Zr₆(Fe,Si)₂₃ phase was found to increase with x. One more minority phase, a bcc Fe–Si solid solution was detected for x ≥ 10. At x = 12 (see Zr₂₁Fe₆₄Si₁₅ in Fig. 2), the volume fractions of the Zr₆(Fe,Si)₂₃ and Fe–Si phases were estimated as 23.7 % and 7.8 %, respectively. In Fig. 3, the lattice parameters of the C14 structure are plotted as a function of the alloy composition. With the Fe substitution for Zr, both lattice parameters decrease linearly until x = 9.5. This result indicates that the Fe solubility in the C14 Laves phase at 1538 K extends to the Zr_23.5Fe_61.5Si₁₅ composition.

Magnetic characterization of the Zr_33-xFe_52+xSi₁₅ alloys was done for the solution range 0 ≤ x ≤ 9. The XRD spectrum recorded for the magnetically oriented Zr₂₅Fe₆₀Si₁₅ powder and shown in Fig. 2 features only the [00l] reflections of the C14 Laves phase, thus indicating that the magnetocrystalline anisotropy of the compounds is uniaxial with the easy magnetization direction lying parallel to the hexagonal axis. The Curie temperature T_C of the stoichiometric Zr₃₃Fe₅₂Si₁₅ Laves phase was found to be 290 K [Fig. 4(a)]. The Fe substitution for Zr increases the T_C by 240 K, to 530 K; the increase of the T_C with x is nearly linear (Fig. 5). Only one magnetic transition was observed in all the two-phase alloys (x ≥ 4), corroborating the earlier finding¹² that the Zr₆(Fe,Si)₂₃ phase is paramagnetic at 300 K. Room-temperature magnetization curves are plotted in Fig. 5 after correction for the self-demagnetization. In order to find the saturation magnetization of the Laves phase, the M(H) data were additionally corrected both for the paramagnetic contribution (evident from the M slope at 15 kOe ≤ H ≤ 30 kOe) and for the volume occupied by the paramagnetic Zr₆(Fe,Si)₂₃ phase. The resulting 4πM_s values are plotted in Fig. 5; the saturation magnetization increases with x reaching the value of 6.7 kG at the Fe solubility limit.

The anisotropy field H_a was determined as the intersect of the "parallel" M(H) curves [Fig. 4(b)] and extrapolated principal slope of the "perpendicular" M(H); the extrapolation was done under assumption that a certain (small) fraction of misaligned crystals requiring a field exceeding the H_a for saturation was present in the oriented powder samples. The obtained H_a values are plotted in Fig.5; except for the alloys x = 0 and x = 2 having their T_C close to 300 K, the room-temperature H_a is nearly constant as 9.5 – 9.8 kOe. Compared to the known hard magnetic materials, these H_a values are too low for permanent magnet development. A somewhat more optimistic assessment can be done based on the magnetic hardness parameter κ = (H_a/8πM_s)^1/2 which was proposed as a quantitative criterion of whether the material can (κ > 1) or cannot (κ ≤ 1) be developed into a viable permanent magnet.¹³ Thanks to their low magnetizations, the Laves phases in the Zr_33-xFe_52+xSi₁₅ alloys with x = 2, 4 and 6 exhibit κ > 1. Unfortunately, the 4πM_s of 4.6 kG limits the maximum energy product of the Laves phase at x = 6 to the low value of 5.3 MGOe. Such values are already achievable in the very inexpensive ferrite magnets, whereas the Zr₂₇Fe₅₈Si₁₅ alloy contains 40 wt.% of the (moderately) expensive Zr.

Despite their ferromagnetism and uniaxial nature of their magnetocrystalline anisotropy, the off-stoichiometric C14 Laves phase compounds Zr–Fe–Si are not interesting as permanent magnet materials. Their weak anisotropy is the primary obstacle for the development of these compounds into viable magnets.

The work was supported by the U.S. Department of Energy, Grant No. DE-FG02-90ER45413.