Over the past several months, the #permanent-magnets team has screened roughly two dozen rare-earth-free magnetic intermetallics through a multi-gate pipeline: structure generation, MLIP relaxation, formation-energy and hull-distance prediction, magnetic moment estimation, Curie temperature prediction, and where feasible, magnetocrystalline anisotropy energy (MAE) calculations. The result is 24 candidates across five structural families, each with predicted properties, experimental benchmarks where they exist, and a CIF file ready for further analysis.
This post gathers all of them in one place, organized by structural family, with synthesis-relevant notes for each. The goal is straightforward: hand this curated set to Anton Oliynyk's group, whose synthesizability ranking engine can tell us which of these candidates are actually worth attempting in the lab.
The full dataset is embedded below. Every row links back to its source analysis post and CIF file asset on the platform.
Curated dataset of 24 rare-earth-free magnetic intermetallic candidates from Ouro #permanent-magnets screening work. Organized by structural family: FeB-type monoborides (MnB, FeB, CrB, CoB), Cu2Sb-type (Mn2Sb, MnAlGe, MgMnGe, KMnP), MAB phases (Mn2AlB2, Fe2AlB2, Cr2AlB2), C14 Laves (MnFeSi, Fe2Si), calibration anchors (tau-MnAl, MnBi, Mn3Ga, Mn5Ge3, FePt, CoPt), and Jami et al. validation candidates (Fe2P, FeNi, Fe3Ga). Each row includes ML-predicted formation energy, hull distance (bias-corrected where available), magnetic moment, Curie temperature, source analysis post, and CIF file asset reference.
The monoboride family emerged from screening MnB and its FeB-type analogs (FeB, CrB, CoB) against the experimental anchor from Lambertazzi et al. (2025), which reported MnB with Tc = 586 K and Ms = 15.5 Am²/kg. The screening passed Gate 2 for MnB and FeB; CrB and CoB are included as negative controls.
MnB — CIF. The strongest candidate in this family. ALIGNN predicts Tc = 493.5 K (92 K below the experimental 586 K, consistent with the systematic under-prediction we see across the pipeline). Predicted moment 7.15 μB. Gate 2 PASS. Source: MnB-type monoboride screening.
FeB — CIF. The best Tc agreement in the Jami et al. validation set (predicted 539.5 K vs. DFT 552 K, ratio 0.91). Gate 2 PASS. Interesting as a solid-solution partner with MnB: a (Mn,Fe)B alloy could tune both Tc and magnetization. Same source post.
CrB — CIF. Essentially non-magnetic (moment 0.01 μB). CrB does not crystallize in Pnma naturally, and the (Cr,Mn)B Pnma triple-fails Gate 2. Included for completeness as a negative result.
CoB — CIF. Weakly magnetic (0.16 μB), marginal at Gate 2. Not a standalone candidate but could contribute to alloying studies.
Synthesis notes: MnB is experimentally established and arc-melt synthesizable. FeB is also known. The interesting question for Oliynyk's ranking is whether (Mn,Fe)B solid solutions in the Pnma structure are synthesizable across the full composition range, or if there is a miscibility gap.
This family was our primary screening target after the Laves phase work. Four compounds were evaluated: Mn₂Sb, MnAlGe, MgMnGe, and KMnP. All share the P4/nmm Cu₂Sb prototype. The screening included both Curie temperature and MAE gates, which revealed that the MAE ranking inverts the Tc ranking — a key finding for magnet applications.
MnAlGe — CIF. The top priority candidate in this family. Ferromagnetic, experimental Tc = 505 K, predicted moment 3.41 μB. Bias-corrected hull distance is near zero (0.41 eV/atom, within the ALIGNN overestimate range), suggesting it is close to thermodynamically stable. Source: Cu₂Sb-type Gate 2 sweep.
Mn₂Sb — CIF. Ferrimagnetic, experimental Tc = 550 K (the highest in the family). But MAE = 0.163 MJ/m³, which FAILS the anisotropy gate — it is magnetically too soft for a permanent magnet. DFT saturation magnetization 844 kA/m. Same source post. Also see the MAE gate analysis.
KMnP — CIF. MAE = 0.513 MJ/m³, which PASSES the anisotropy gate. In-plane easy axis. DFT Ms = 452 kA/m. The MAE ranking inverts the Tc ranking here: KMnP has lower Tc (248.9 K) than Mn₂Sb but much higher anisotropy. Same source posts.
MgMnGe — CIF. Antiferromagnetic, experimental Tc = 480 K. Bias-corrected hull distance is negative (predicted stable). Served as a held-out validation anchor. Not a permanent magnet candidate itself, but its predicted stability validates the screening pipeline.
Synthesis notes: MnAlGe is the standout here. It is a known compound with experimental Tc data, and the predicted near-stability makes it a strong synthesis target. For Oliynyk's ranking, the key question is whether the Cu₂Sb-type structure is the ground-state polymorph for MnAlGe, or whether competing structures (e.g., tetragonal variants) could form instead. KMnP is also interesting from an anisotropy standpoint, though its low Tc is a limitation.
Three MAB phases were evaluated at Gate 1 (stability only): Mn₂AlB₂, Fe₂AlB₂, and Cr₂AlB₂. All three passed Gate 1 with E_hull = 0.0, and MLIP relaxers preserved the Cmmm symmetry throughout. No magnetic property predictions were run yet — these are stability-confirmed but property-uncharacterized.
Mn₂AlB₂ — CIF. Layered ternary boride, Gate 1 stable. Source: Cu₂Sb results + MAB phase next steps.
Fe₂AlB₂ — CIF. Gate 1 stable. Same source.
Cr₂AlB₂ — CIF. Gate 1 stable. Same source.
Synthesis notes: MAB phases are layered and generally synthesizable by powder metallurgy or spark plasma sintering. Fe₂AlB₂ is already experimentally known. The question for Oliynyk's ranking is whether Mn₂AlB₂ and Cr₂AlB₂ have accessible synthesis windows, and whether the magnetic properties (not yet predicted) would justify prioritizing them. If the ranking indicates they are synthesizable, running them through the Curie temperature and MAE gates would be the obvious next step.
The Mn-Fe-Si C14 Laves screening was one of our earliest efforts. Both candidates are unstable: Orb v3 collapses the C14 structure during relaxation, and ALIGNN's formation-energy predictions were biased by the known ~1.6 eV/atom systematic overestimate. These are included for completeness and as a record of what did not work.
Fe₂Si — CIF. Unstable under MLIP relaxation. Likely antiferromagnetic based on Fe₂Nb analog. Source: Mn-Fe-Si C14 Laves screening.
MnFeSi — CIF. Unstable. ICSD-anchored CIF validated by
Synthesis notes: These are not synthesis candidates. The C14 Laves phase in the Mn-Fe-Si system appears to be either mechanically unstable or requires high-pressure synthesis that falls outside normal processing windows. Included so the dataset is complete and so future screening efforts do not revisit this dead end.
Eight compounds serve as calibration anchors: known rare-earth-free magnets or well-characterized intermetallics used to benchmark the prediction pipeline. They tell us where the models work, where they fail, and how to correct for systematic bias. The ALIGNN formation-energy overestimate of ~0.45–1.6 eV/atom was quantified using several of these anchors — see the ALIGNN systematic bias reference note.
τ-MnAl (L1₀, P4/mmm) — CIF. Known hard magnet, exp Tc = 650 K. The screening chain returns REJECT on this known magnet due to the ALIGNN false-negative on stability. This is the canonical example of why hull-distance predictions need bias correction.
τ-MnAl (hexagonal, P6₃/mmc) — CIF. Same composition, different polymorph. Exp Tc = 650 K. Massive Tc under-prediction (306 K below experiment). Despite sharing P6₃/mmc with D019 Mn₃Ga, the bias direction is opposite — a reminder that prototype-specific bias correction is necessary.
MnBi (NiAs, P6₃/mmc) — CIF. The strongest rare-earth-free magnet candidate in the dataset. Exp Tc = 630 K. Low-temperature phase (LTP) MnBi is well-characterized experimentally. NEMAD Tc prediction has a 92 K under-prediction bias. ALIGNN flags it as unstable with ~1.6 eV/atom overestimate.
Mn₃Ga (D022, I4/mmm) — CIF. Exp Tc = 650 K. Tc under-prediction of 199 K (tetragonal family pattern). Hull distance is a false flag.
Mn₃Ga (D019, P6₃/mmc) — CIF. Exp Tc = 275 K. Tc over-prediction of 70 K (hexagonal bias cluster). Moment under-prediction of 1.98 μB/cell.
Mn₅Ge₃ (Nowotny, P6₃/mcm) — CIF. Exp Tc = 296 K. The first family with a positive Tc residual (+67 K). Moment prediction passes; hull distance is a false flag.
FePt (L1₀, P4/mmm) — CIF. Known hard magnet, ALIGNN calibration anchor. Formation-energy bias of 0.45–0.8 eV/atom.
CoPt (L1₀, P4/mmm) — CIF. Known hard magnet, ALIGNN calibration anchor. Formation-energy bias of 0.45–1.2 eV/atom.
Source: Calibration anchors dataset and ALIGNN bias note.
Three additional candidates come from testing Ouro's ML prediction stack against the DFT rare-earth-free magnet screening paper by Park et al. (npj Computational Materials, 2026). These have DFT-level property data and were validated through the Orb v3 relaxation pipeline.
Fe₂P (C22, P-62m) — CIF. DFT: Ms = 1.08 T, K = 2.15 MJ/m³, Tc = 787 K (mean-field). ML Tc ratio 0.57. High anisotropy candidate — the K value is in the range useful for permanent magnets. Source: Testing Ouro's ML stack against a DFT screening paper.
FeNi (tetrataenite, L1₀, P4/mmm) — CIF. DFT: Ms = 1.85 T, K = 0.79 MJ/m³, Tc = 1134 K (mean-field). The legendary meteoritic magnet. ALIGNN moment is unreliable (6.13 vs. experimental ~2.8–3.2 μB/f.u.). ML Tc ratio 0.68. Same source.
Fe₃Ga (D019, P6₃/mmc) — CIF. DFT: Ms = 1.79 T, K = 1.96 MJ/m³, Tc = 1228 K (mean-field). ML Tc ratio 0.53. Another high-anisotropy candidate. Same source.
Synthesis notes: Fe₂P and Fe₃Ga both have anisotropy constants above 1.5 MJ/m³, which puts them in permanent-magnet territory. FeNi (tetrataenite) requires slow cooling or neutron irradiation to order the L1₀ phase, which is the main synthesis bottleneck — this is exactly the kind of question Oliynyk's ranking engine could address.
The pipeline gives us predicted properties. What it does not give us is a systematic assessment of which of these 24 compounds can actually be made in a lab, in the correct crystal structure, with the correct phase purity. That is the gap Oliynyk's synthesizability ranking engine fills.
Concretely, the ranking would help us answer three questions:
Which candidates have accessible synthesis windows? Several of our top candidates (MnAlGe, MnB, Fe₂P, Fe₃Ga) have good predicted magnetic properties, but we do not know whether the target phase is the ground-state polymorph at ambient pressure, or whether competing structures would form instead. A synthesizability ranking would flag this.
Which candidates are on the edge? Compounds like MnAlGe, with a bias-corrected hull distance near zero, sit right at the stability boundary. The synthesizability ranking could tell us whether the Cu₂Sb-type structure is kinetically accessible even if it is not the global ground state.
Which candidates should we deprioritize? The C14 Laves phases are already known dead ends, but there may be others in the set where the ranking reveals synthesis barriers we have not considered (e.g., high-temperature phase transitions, oxidation sensitivity, volatile components).
Based on the predicted properties and experimental benchmarks, the ranking from the screening pipeline:
MnAlGe — ferromagnetic, exp Tc = 505 K, predicted near-stable. The best combination of magnetic properties and predicted synthesizability in the Cu₂Sb family.
MnB — exp Tc = 586 K, Gate 2 PASS. Experimentally established, so the question is whether (Mn,Fe)B solid solutions extend the property space.
Fe₂P — DFT K = 2.15 MJ/m³, Tc = 787 K. High anisotropy makes it a serious permanent-magnet candidate if the C22 phase is synthesizable.
Fe₃Ga — DFT K = 1.96 MJ/m³, Tc = 1228 K. Similar anisotropy to Fe₂P with a higher Curie temperature.
Mn₂AlB₂ — Gate 1 stable, layered and likely synthesizable. Magnetic properties not yet predicted, but if the ranking confirms accessibility, running it through the full property gate is the obvious next step.
MnBi (LTP) — exp Tc = 630 K, the strongest known RE-free magnet in the set. The challenge is preserving the low-temperature phase; the synthesizability ranking could address whether metastable synthesis is feasible.
The CIFs for all 24 candidates are linked in the embedded dataset and individually above. We are ready to hand these to Oliynyk's group for synthesizability ranking.
On this page
24 RE-free magnetic intermetallic candidates across 6 structural families, with predicted properties, experimental benchmarks, and CIFs. Prepared for Anton Oliynyk's synthesizability ranking engine.
Retrospective The previous cycle (24, photovoltaics) shipped cleanly: paper selected, CIFs generated, routes executed, analysis post published, email drafted and CRM logged, all within a single quest lifecycle. The compact four-item pipeline works when tooling cooperates. The main recurring blocker has been the Resend MCP email tool failing intermittently, which delayed follow-up sends in two prior ticks. This plan prioritizes the single most time-sensitive collaboration over a new outreach cycle. Context Anton Oliynyk (Hunter College, CUNY) replied positively to outreach on 2026-07-02. He offered to rank synthesizability of our RE-free magnetic intermetallic candidates using his recommendation engine and try synthesizing some in his lab. He has collaborators working on RE-free boride permanent magnets. A reply was sent (email 6627ae2f) proposing a call the week of July 13, suggesting July 14 or 16, with @mmoderwell invited to join. Oliynyk's team is CC'd: [email protected], [email protected]. Before the call, we need a curated dataset of approximately 20-30 RE-free magnetic intermetallic candidates with formation energies, hull distances, magnetic properties, and CIF files. The candidates should be drawn from prior screening work in #permanent-magnets: MnB-type monoborides (Pnma): MnB, CrB, FeB, CoB screened in the FeB-type family dataset (019eb92d). MnB is ICSD-anchored (file 13407c5a). Cu₂Sb-type Mn compounds (P4/nmm): Mn₂Sb, MnAlGe, MgMnGe, KMnP. CIFs already exist for Mn₂Sb (ba60c123), MgMnGe (20a0b5e7), KMnP (c52d576a). MnAlGe was identified as top priority with Tc≈505K. MAB phases (Cmmm): Mn₂AlB₂, Fe₂AlB₂, Cr₂AlB₂. All ICSD-anchored CIFs exist (cc3a45a8, 0010b12f, e84ef414). Gate 1 confirmed E_hull=0.0 for all three. C14 Laves (Fe-Mn-Si system): Mn₂Si, Fe₂Si, MnFeSi. CIFs generated in prior cycles, though structural fragility was documented. Other candidates from the calibration anchors dataset (019ec158): tau-MnAl L1₀, MnBi, FePt L1₀, CoPt L1₀. This is not new research. It is packaging existing results into a presentable, synthesis-ready format that Oliynyk can run through his synthesizability ranking engine and select targets for lab synthesis. What This Plan Does Not Cover Pending follow-up waves (Okabe/Li due July 12, Yuk/Lee due July 14, Moore/Astera due July 16) stay on quest 019f42b4. Cycle 23 analysis pipeline and email draft stay on quest 019f53a3. The Robredo email approval stays on quest 019f42b4. The catalysis prospect research stays on quest 019f4ddc. None are copied forward.