This is the first deliverable of the Cu₂Sb Screening + ML Property Prediction plan — a structural survey of four Mn-bearing Cu₂Sb-type compounds as candidates for rare-earth-free permanent magnets.
The Cu₂Sb structure (space group P4/nmm, #129) is a tetragonal arrangement with Z=2 formula units (6 atoms per cell) sitting on three distinct Wyckoff positions:
2a (0,0,0) — typically occupied by the transition metal or larger cation
2b (0,0,½) — typically the pnictogen/chalcogen (Sb, Ge, P)
2c (0,½,z) — the variable-z site, occupied by the second metal
What makes this structure interesting for magnet design is the two-sublattice geometry. When both 2a and 2c are occupied by the same magnetic species (as in Mn₂Sb), the antiparallel alignment of moments on inequivalent sublattices creates ferrimagnetic behavior. When only one sublattice is magnetic (as in MnAlGe with Mn on 2a), you get straightforward ferromagnetism with the full moment available.
Cu2Sb-type (anti-type) Mn2Sb. P4/nmm #129, Z=2, 6 atoms. a=4.074 Å, c=6.570 Å, c/a=1.613. ICSD 103470. Ferrimagnetic, Tc≈550K. Mn(I) at 2a, Mn(II) at 2c (z≈0.280), Sb at 2b.
Mn₂Sb is the textbook Cu₂Sb ferrimagnet. Two Mn sublattices (2a and 2c) couple antiparallel, each carrying substantial moment (~2.1 μB and ~1.7 μB respectively), but the net polarization is only ~0.4 μB/f.u. The Curie temperature is high at 550 K, which is a genuine advantage, but the small net moment limits the energy product ceiling. ICSD reference: entry 103470. The c/a ratio of 1.613 is close to the ideal close-packing value.
Cu2Sb-type MnAlGe. P4/nmm #129, Z=2, 6 atoms. a=3.915 Å, c=5.840 Å, c/a=1.492. Ferromagnetic, Tc≈505K, ~1.5 μB/Mn. Strong uniaxial anisotropy. Mn at 2a, Al at 2c (z≈0.25), Ge at 2b.
MnAlGe is the standout candidate for hard magnet screening. It's a true ferromagnet (Tc ≈ 505 K) with ~1.5 μB/Mn, and critically, it exhibits strong uniaxial magnetocrystalline anisotropy — the property that actually matters for permanent magnets. Mn occupies the 2a site with Al and Ge on 2c and 2b respectively, so there's no competing sublattice diluting the net moment. The c/a ratio of 1.492 is notably lower than the other candidates, reflecting a compressed tetragonal distortion that likely contributes to the anisotropy.
Cu2Sb-type MgMnGe. P4/nmm #129, Z=2, 6 atoms. a=4.120 Å, c=6.880 Å, c/a=1.670. Antiferromagnetic, TN≈480K. Large local Mn moments. Mg at 2a, Mn at 2c (z≈0.28), Ge at 2b.
MgMnGe presents the AFM challenge. The Mn moments on the 2c sublattice are large, but they couple antiparallel (TN ≈ 480 K). The Mg on 2a is non-magnetic, so this isn't a simple ferrimagnetic compensation — it's genuine antiferromagnetism within the Mn sublattice. That said, the large local moments and the tunability of the 2a site make this worth screening computationally. If aliovalent substitution (e.g., replacing Mg with a magnetic atom) or epitaxial strain can destabilize the AFM coupling, the large underlying moments could become useful.
Cu2Sb-type KMnP. P4/nmm #129, Z=2, 6 atoms. a=4.140 Å, c=6.990 Å, c/a=1.688. Magnetic order TBD. Novel pnictide candidate. K at 2a, Mn at 2c (z≈0.30), P at 2b.
KMnP is the wildcard. K is a large electropositive cation that expands the lattice significantly (a=4.140, c=6.990 Å, c/a=1.688 — the largest of the four). The magnetic order isn't well characterized experimentally. The Mn sits on the 2c site with z≈0.30, and the P on 2b. The valence electron count differs from the other three (K contributes 1 electron vs. Mg's 2), which changes the band filling and potentially the magnetic ground state. Worth a computational look, but the experimental uncertainty makes this lower priority for near-term screening.
Compound | a (Å) | c (Å) | c/a | Magnetic Order | Tc/TN (K) | Net Moment | Priority |
|---|---|---|---|---|---|---|---|
MnAlGe | 3.915 | 5.840 | 1.492 | Ferromagnetic | 505 | ~1.5 μB/f.u. | 1 |
Mn₂Sb | 4.074 | 6.570 | 1.613 | Ferrimagnetic | 550 | ~0.4 μB/f.u. | 2 |
MgMnGe | 4.120 | 6.880 | 1.670 | Antiferromagnetic | 480 | 0 (AFM) | 3 |
KMnP | 4.140 | 6.990 | 1.688 | Unknown | — | TBD | 4 |
With ICSD-anchored CIFs for all four compositions in hand, the pipeline proceeds to:
Orb v3 Gate 1 stability check — relax all four structures and verify they survive without triclinic P1 collapse (the GPSK-05 failure mode we documented extensively during Laves screening)
Materials Project hull energy — cross-validate thermodynamic stability using the MP 'Calculate energy above hull' route (our standard secondary check after learning that JARVIS ALIGNN has systematic +1.6 eV/atom bias)
ALIGNN magnetic property prediction — saturation magnetization, MAE, and Curie temperature for the stable structures (pending resolution of the 9-anchor residual analysis block from Apollo's ICSD 53768 patch)
MAE ranking — the end goal is a ranked list of Cu₂Sb-type compositions by magnetocrystalline anisotropy energy, which is the property that determines whether a material can function as a permanent magnet
MnAlGe goes first. Its ferromagnetic ground state, confirmed experimental Tc, and reported uniaxial anisotropy make it the highest-confidence candidate. If it passes Orb v3 and MP hull, it moves immediately to full magnetic property screening.
Part of the Cu₂Sb Screening + ML Property Prediction plan. ICSD-anchored CIFs generated from experimental lattice parameters and Wyckoff positions.
On this page
Structural survey of four Mn-bearing Cu₂Sb-type compounds for rare-earth-free permanent magnet screening. ICSD-anchored CIFs, lattice parameters, Wyckoff positions, and magnetic properties.