This loop asked whether tungsten can do for tetrataenite, the ordered iron–nickel magnet L1₀ FeNi, what platinum does for FePt: put a heavy atom with strong spin–orbit coupling in the iron layer and raise the anisotropy.
New candidate: Fe₅WNi₆. It is on the measured rung: every criterion of the objective has a measurement inside its target, while stability in one PBE setup, a converged basis, and whether it can be made are still open. It reads 1.27–1.63 MJ/m³ along c, depending on how spin–orbit coupling is treated, at 1.26 T, with a Curie point of at least about 770 K, for 14.11 $/kg. That is behind our cobalt-free lead Mn₂FeB₂ on anisotropy and magnetization (1.74–1.84 MJ/m³ at 1.49 T), and ahead of it on the hull and the Curie point.
Tungsten goes onto the iron sites of tetrataenite, never the nickel sites. There it raises the anisotropy faster than it lowers the magnetization. Plain tetrataenite reads 0.60 MJ/m³ at 1.72 T on our route; with one tungsten in twelve atoms it reads 1.27 at 1.26 T, and both of our targets hold at once. The catch is the one tetrataenite already has. Its iron–nickel layering forms only below about 320 °C, where atoms barely move, and tungsten makes that ordering drive weaker, not stronger. So this is a better tetrataenite if tetrataenite can be made, not a way around making it. Nobody has computed or made tungsten in tetrataenite as far as we found. Nickel–tungsten compounds themselves are all non-magnetic, and tungsten in our layered Fe–Ni–Al–Si lead was far off the hull on every site.
PBE, DZP basis, kspacing 0.2 Å⁻¹, each cell relaxed in DFT and measured on its own relaxed cell; K is along c against the plane, and the in-plane directions agreed in every cell. Hull distances are on the Orb surface; the last column puts the same cells on published-PBE references. Each tungsten cell is the lowest-energy arrangement of the ones we screened at its composition.
Cell | W on Fe sites (at.%) | Geometry | Novelty | K along c (MJ/m³) | Js (T) | Curie route (K) | Orb hull (meV/atom) | Published-PBE references (meV/atom) |
|---|---|---|---|---|---|---|---|---|
L1₀ FeNi (parent) | 0 | |||||||
6.25 | ||||||||
8.3 | ||||||||
12.5 |
The targets are K ≥ 1 MJ/m³ along one axis, Js ≥ 1 T, a Curie point of 450 K or more, and at most 75 meV/atom above the hull. Fe₅WNi₆'s +1.27 is from our default perturbative treatment of spin–orbit coupling, which is known to fail for heavy atoms in L1₀ alloys. The two methods that do not expand it read higher: +1.42 by the force theorem and +1.63 by self-consistent total energy. The parent's Curie reading puts the route at most 1.19× high, so Fe₅WNi₆'s Curie point is at least about 770 K. Fe₅WNi₆ has no imaginary phonon mode (run
Fe₅WNi₆ as relaxed: alternating iron and nickel layers along the easy axis c, with tungsten taking half the sites of one iron layer in three, in a checkerboard:
Tungsten on the iron site raises tetrataenite's anisotropy at 1.4 T or more: partly. At 6 % tungsten K rose 1.32× (predicted at least 1.2×), but Js came out at 1.39 T, 0.01 T under the line.
Tungsten behaves like molybdenum in tetrataenite: held. It takes the iron site by about 0.4 eV, and it costs the iron–nickel ordering preference 109 meV per tungsten (predicted 60–200).
Somewhere between 6 and 12.5 % tungsten both targets hold together: held. At 8.3 %, 1.27–1.63 MJ/m³ at 1.26 T (predicted 0.95–1.35 at 1.18–1.32).
In the layered Fe₄Ni₂AlSi lead, tungsten takes an iron site near the hull: refuted. Every site is 40–76 meV/atom above the parent on Orb.
Nickel–tungsten has no magnet of its own, but tungsten is a working anisotropy lever in tetrataenite. Each tungsten removes about 4.3 μB from the cell, so the trade sets the composition: 6 % is short on anisotropy and 12.5 % is short on magnetization.
How the tungsten is arranged matters as well as how much there is: at 8.3 %, a checkerboard in one iron layer of three is 20 meV/atom below a full tungsten layer.
Disordered fcc nickel–iron dissolves a few to about 10 at.% tungsten (the binder of tungsten heavy alloys), so the composition is not exotic. The ordering is.
Fe₅WNi₆ is not a validated magnet. Its stability rests on the Orb surface and on published-PBE references scored by Orb, with no same-setup PBE check yet. Its anisotropy is in the DZP basis, on a route that reads plain tetrataenite at about half its measured value. It assumes the iron–nickel layering of tetrataenite plus an ordered tungsten arrangement. Tetrataenite forms naturally only in meteorites cooled over millions of years, and tungsten weakens the drive to order. A sample would most likely be disordered fcc (Fe,Ni,W), which is a soft magnet.
Earlier: loop 62
known (meteoritic) |
1.717 |
988 (≥ 830 measured) |
not checked |
1.390 |
new composition |
1.258 |
not checked |
0.950 |
Cell + ionic relaxation with ABACUS DFT; 0.04 eV/Å threshold; final energy = -46499.8954 eV; energy change = -0.0283 eV
W on the Fe sites of L1₀ FeNi raises K from 0.60 to 1.27–1.63 MJ/m³ at 1.26 T (8.3 % W); measured rung, with stability and makeability open.