This loop left the structure family of the last four loops and asked whether swapping iron and other cheap elements into four-element compounds from the crystal databases gives a new iron-rich compound near stability.
No new candidate.
Of 836 cells built this way, none is stable and only nine are within 25 meV/atom of the stable phases. All nine are the same thing: the known compound ZrFe₂ with titanium and niobium mixed onto the zirconium sites, an alloy and not a new compound. Every other structure type is at least 38 meV/atom away, and the types that are iron-rich in their rare-earth originals do not survive zirconium in the rare earth's place.
With the three-element search of an earlier campaign, this closes one whole method: substituting cheap elements into structures the databases already hold does not produce a new iron-rich magnet candidate. The loop cost nothing and nothing in it was taken to DFT. ZrFe₉Co₂Ti from loop 37 remains the iron-based candidate.
The pool: four-element compounds in Materials Project, Alexandria and GNoME with a uniaxial space group (75–194), at most 30 atoms, within 25 meV/atom of the published-PBE hull and one late transition metal making up 53 at.% or more. One representative was taken per structure type and pattern of element classes (38 in all). The majority metal was replaced by Fe and each of the other three by cheap elements of its class: Zr, Ti or Nb for a large atom; Si, Al, P or Ge for a p-block atom; B or C; Co, Mn or Ni for a second late metal. Each cell was relaxed with the Orb potential and placed on the Orb hull by Energy above the convex hull
Cells | Share of 828 | |
|---|---|---|
On the hull | 0 | 0 % |
Within 25 meV/atom | 9 | 1.1 % |
Within 50 meV/atom | 13 |
The nine cells within 25 meV/atom, all with two thirds iron:
Cell | What it is | Hull, Orb (meV/atom) | Hull, PBE references (meV/atom) |
|---|---|---|---|
ZrTi₂NbFe₈ | ordering of C15 ZrFe₂ (12 atoms) |
They decompose into ZrFe₂, TiFe₂, NbFe₂ and TiNbFe₄. With titanium and niobium replaced by zirconium, the 12-atom cells are the made cubic Laves phase ZrFe₂ and the 30-atom cells are another stacking of the same building blocks, so these are Laves-phase alloys. Our magnetic order screen reads the two lowest at 1.11 T
The best cell of the next ten structure types, named by the database compound each was built from:
Space group | Built from | Cells | Best cell | Best, Orb hull (meV/atom) | Median (meV/atom) |
|---|---|---|---|---|---|
139 | TmV₂Ga₂Rh₇ | 191 | Ti₂NbAl₂Fe₇ |
The first two of these are iron-rich lattices with the atoms merely relabelled (lattice decorations), not compounds with their own structure. Some linked runs show as failed: the calculation finished and the platform failed while saving it, and the number is taken from that run's own record.
Some cell within 25 meV/atom is neither a relabelled lattice nor an ordering of a known type, and is a strong ferromagnet: not supported in substance. Predicted 2–7 % of cells within 25 meV/atom and up to three real survivors; found 1.1 % and none. Our classifier did label six of the nine "unmatched", because it tests two elements sharing a site and these have three; checked by hand they are orderings of a known type.
Substitution into database structures is closed as a search method for iron-rich uniaxial compounds of cheap elements, for three and now four elements.
Four-element database compounds with a late-metal majority are mostly platinum-group and rare-earth compounds; with cheap elements in their place the typical cell is 150–450 meV/atom from stability.
Zirconium does not stand in for a rare earth in the iron-rich carbide and boride types (66–102 meV/atom), as it did not in the simpler rare-earth structures of earlier loops.
Earlier in this campaign: loop 36
1.6 % |
Within 100 meV/atom | 74 | 8.9 % |
Zr₂Ti₂NbFe₁₀ | ordering of another Laves stacking (30 atoms) | not returned |
ZrTi₂(NbFe₅)₂ | ordering of another Laves stacking (30 atoms) |
ZrTi(NbFe₄)₂ | ordering of C15 ZrFe₂ (12 atoms) |
Zr₂Ti(NbFe₅)₂ | ordering of another Laves stacking (30 atoms) |
ZrTi₂(NbFe₅)₂ | ordering of another Laves stacking (30 atoms) |
Zr₂Ti₂NbFe₁₀ | ordering of another Laves stacking (30 atoms) |
Zr₂Ti(NbFe₅)₂ | ordering of another Laves stacking (30 atoms) |
Zr₂TiNbFe₈ | ordering of C15 ZrFe₂ (12 atoms) |
273.5 |
123 | Cd₄Ni₁₂CN₃ | 8 | Al₄Fe₁₂BC₃ | 231.8 |
160 | Mg₃Ti₃Co₁₆Si₇ | 47 | Zr₃Nb₃Fe₁₆Ge₇ | 149.9 |
123 | Li₂YSi₂Pd₇ | 24 | Ti₂NbAl₂Fe₇ | 178.3 |
123 | CeFe₅GeP₂ | 95 | ZrFe₅Si₂Ge | 205.5 |
123 | LiErPPd₅ | 59 | TiNbAlFe₅ | 189.9 |
187 | Tm₄Fe₁₆B₃C | 6 | Zr₄Fe₁₆B₃C | 187.8 |
166 | Tm₂Mn₁₅Fe₂C₃ | 42 | Zr₂Fe₁₅Ni₂C₃ | 201.5 |
191 | NdY₃(Co₄B)₄ | 12 | Zr₃Nb(Fe₄B)₄ | 157.9 |
160 | NdSmFe₄Co₁₃ | 18 | ZrMn₄NbFe₁₃ | 117.5 |
Substituting Fe and cheap elements into 38 near-hull four-element uniaxial prototypes: 0 of 828 on the hull, 9 within 25 meV/atom, all (Zr,Ti,Nb)Fe₂ Laves alloys. Free tier, no DFT.