Sharing some notes as I go through this paper:
: A Computational High-Throughput Study Lorenzo A. Mariano, Vu Ha Anh Nguyen, Valerio Briganti, and Alessandro Lunghi Journal of the American Chemical Society 2024 146 (49), 34158-34166 DOI: 10.1021/jacs.4c14076
This paper came up as I was researching whether to go after MAE or coercivity as a next prediction/screening model.
MAE was the answer as it's intrinsic to the material and upstream of coercivity anyways. This paper shows that it's fundamental to predicting coercivity, so it'll be good to have.
This paper makes their computed data available here (published September 6, 2024):
https://zenodo.org/records/13712318
I'm still looking into it but ideally we can clean it to get material structure (CIF) to MAE constants.
Goal: Find cobalt molecules that keep their magnetic “north” and “south” better than today’s best single-molecule magnets.
What they did
Built a big digital library of 15,000-plus virtual cobalt complexes.
Optimized each one with quantum chemistry to measure its magnetic “lock-in” strength (the axial anisotropy D).
Big discovery: You don’t need fragile two-coordinate cobalt.
A simple rule works:
Put two strong, straight-line (“axial”) ligands very close to cobalt and keep all other bonds weaker or farther away.
This “pseudo-linear” setup lets the cobalt keep a big orbital moment → huge magnetic anisotropy.
Numbers that matter
Nearly 200 new designs beat |D| > 100 cm⁻¹, many in 4–5 coordination.
One record hit reached |D| ≈ 250 cm⁻¹—better than any cobalt magnet reported so far.
Why it matters
Chemists now have a clear design recipe that is easier to synthesize and more stable than ultra-low-coordination compounds.
All data are released as the COMPASS database, so anyone can train ML models or pick promising targets for the lab.
Bottom line: give cobalt a straight, two-bond “spine,” relax the side bonds, and you unlock record-size magnetic stickiness—opening new paths to strong molecular and bulk magnets.
2. Why iron is trickier than cobalt
Iron’s built-in “magnet locking” power (spin-orbit coupling) is only about one-third of cobalt’s.
So we need to help iron by:
Keeping the same two-bond line, and
Adding things that boost the effect—heavier neighbor atoms, stretched lattices, or extra strain.
3. Simple design rules for iron magnets
Short north-south bonds: Put atoms like N, B, Sb, or Ni directly above and below each iron atom.
Loose east-west bonds: Make the side bonds longer or weaker so they don’t cancel the effect.
Heavy helpers: Surround iron with a few heavier atoms (Ni, Sb, W, Pt) to amplify the locking power.
Built-in stretch: Use crystal structures that are naturally longer in one direction (tetragonal) or apply strain during processing.
4. Real-world crystal families that already follow these rules
Example material | Why it works | Status |
|---|---|---|
Fe₁₆N₂ (“α″-phase”) | N sits on the two short axial sites; lattice is tall and skinny | Made in thin films; high magnet strength reported |
L1₀-FeNi (tetrataenite) | Alternating Fe and Ni layers give iron two Ni “caps” | Now producible in bulk with Cu-assisted casting |
Fe₃Sn₁-ySbᵧ | Swapping some Sn for heavier Sb stretches the lattice along one axis | Lab tests show strong anisotropy |
Fe₅SiB₂ / Fe₅PB₂ | B or P occupy the axial spots in Fe square pyramids | Cheap elements, high Curie temperature |
Strained Fe-Co films | Growing Fe-Co on a mismatched substrate stretches c-axis | Good for nano-layer magnets |
Bottom line:
Treat each iron atom like a tiny bar magnet: clamp it firmly from the top and bottom, leave its sides free, and give it a heavier buddy or a bit of stretch. Do that, and cheap, earth-abundant iron can inch closer to the performance of today’s rare-earth super-magnets.