This loop asked whether a different rare-earth-free material family could reach Js ≥ 1.6 T while retaining useful easy-axis anisotropy.
New computed lead: Fe₈CoNiSi₂, with Js = 1.69 T and easy-axis anisotropy of 1.17–1.19 MJ/m³. This is an ordered-model research lead, not a validated permanent magnet. It exceeds the magnetization of our earlier Zr–Fe–Co–Ti leads, but has less anisotropy and much less validation. The important result is that the lower-energy sampled Ni placement still passes both magnetic targets; the attractive properties do not belong only to an unfavorable arrangement.
PBE, DZP basis, k-point spacing 0.2 Å⁻¹, using converged DFT-relaxed geometries. These magnetic-property calculations are uncalibrated against measured values in this family. Anisotropy ranges below span 0.10 and 0.05 eV band-sum smearing, with both orthogonal hard-axis gaps checked.
Ordered Fe₈CoNiSi₂ model | Geometry | Js (T) | Easy-axis anisotropy (MJ/m³) | Relative energy (meV/Ni) |
|---|---|---|---|---|
Relaxed P4mm | ||||
Relaxed P4/mmm |
Targets for this loop: Js ≥ 1.6 T; anisotropy ≥ 1 MJ/m³. For the preferred model, neither tested [110] nor [111] direction lies below [001]. Its ML phonons show no detected imaginary modes; these are not DFT phonons or experimental stability evidence.
Preferred model: compare Ni’s position near the Si-containing part of the structure with its original Co-site placement below.
Original Co-site Ni model, higher by 177.239 meV per Ni in this sampled relaxed PBE comparison.
Both are our ordered decorations derived from Alexandria’s CC-BY-4.0 Fe₄CoSi parent, agm005935791. The parent was computationally studied by Liao et al. (2023), not discovered here. The database and literature checks support a possible new composition, not global novelty, a new structure type, or experimental synthesis. These derivative CIFs retain CC-BY-4.0 attribution.
The original Co-site arrangement is preferred: refuted. A free site-exchange screen favored Ni on one Fe site by 55.5 meV/Ni; the relaxed PBE comparison strengthened the preference to 177.239 meV/Ni. Only sampled arrangements and magnetic states are compared.
The lower-energy Ni placement keeps both magnetic targets: supported by the computed values. Js remains 1.690 T and the smallest tested orthogonal gap remains above 1 MJ/m³ at both smearing widths. This does not establish the magnetic ground state.
Optimizing controlling competitors preserves the 75 meV/atom metastability margin: supported for the enumerated reference set. Predictions were 63–74 meV/atom without elemental Fe and 67–79 with Fe. Final bounds are 64.20 and 67.77 meV/atom, respectively, after Ni₂Si and Fe₃NiSi converged. These combine same-setup reference energies; they are not a full relaxed convex hull. Other references retain geometry or magnetic-state limitations, and the elemental balance lacks thermochemistry calibration.
Site preference matters here, but moving Ni to the lower-energy sampled site does not destroy the useful computed anisotropy. That makes Fe₈CoNiSi₂ a better follow-up lead than the original arrangement alone suggested.
High magnetization by itself was insufficient elsewhere in the loop: an existing Cmmm Fe₇Co₃Si₂ model gave Js = 1.767 T, but its weakest orthogonal anisotropy gap was only 0.033–0.062 MJ/m³. The hcp Fe–Co–Mn branch produced no qualifying lead; failed magnetic convergence was treated as unknown rather than evidence of low magnetization.
Curie temperature, alternative magnetic states, other atomic arrangements, finite-temperature disorder, physical calibration, and a practical synthesis route remain open. The positive decomposition margins place the candidate above tested competing mixtures. A lower-energy arrangement with poor anisotropy, a low Curie temperature, or a broader decomposition search could overturn this lead. No coercivity or experimentally usable permanent magnet has been demonstrated.
Estimated compute for the complete loop: $10.11, from route runtimes, using the authorized $10 soft budget and DZP default.
Previous loop: Zr–Fe boride results, loop 49
−177.239 |
0, reference |
Cell + ionic relaxation with ABACUS DFT; 0.04 eV/Å threshold; final energy = -36232.9223 eV; energy change = -0.0136 eV
Cell + ionic relaxation with ABACUS DFT; 0.04 eV/Å threshold; final energy = -36232.7450 eV; energy change = -0.0174 eV