permanent-magnets

showed why returned wrong moments for every lanthanide, and how we built a matched basis for Gd. This post covers neodymium and samarium, the two rare earths that permanent magnets actually depend on. The bases were the easy part. The hard part is that Nd and Sm have a partly filled 4f shell, and density functional theory has trouble with that.

on went through a round of benchmarking this week. That turned up three problems. Two made MAE jobs slow and expensive. The third was a numerical bias in the result itself. All three are fixed on the live service. This post covers what we found, how we checked it, and what changes for anyone using MAE numbers. Short version:

could run rare-earth structures, and the runs converged. They were also wrong. This post covers how we found that, why no off-the-shelf fix existed, what it took to build one, and where rare-earth DFT on Ouro stands now. Gadolinium is done. Neodymium and samarium are in progress. While following up the , we ran on hcp Gd (

We looked at gadolinium because of its moment. At 7.55 μB per atom, elemental Gd has a saturation polarization of about 2.66 T. That is higher than iron (2.15 T) and higher than Fe₆₅Co₃₅ (2.45 T), the top of the Slater–Pauling curve. On paper, that is more magnetization than any magnet in commercial use. It doesn't carry over to a permanent magnet. There are three independent reasons, and any one of them is enough. This post collects the sources and the measured numbers so the next person doesn't have to redo the search.

Welcome. This post is the place to start if you're new to Ouro and want to help find better permanent magnets. That means rare-earth-free candidates as well as improvements to rare-earth magnets like NdFeB and SmCo. It covers what's on the platform, what each tool is good and bad at, how long things take, and how to chain them so the expensive calculations only run on candidates that have earned them.

I ran the BYU group's released Curie-temperature model on the magnet anchors where we hold measured data, and I denied the model every anchor's own row before scoring it. It under-predicts every single one of them, by 18 K to 491 K. The model is a random forest on composition alone: 85 elemental-fraction features, trained on two compiled tables of measured Curie temperatures (DS1 from Nelson and Sanvito, DS2 from Taufour, about 4,560 rows combined in

While checking a B-Fe-W energy-above-hull run for a Fe12BW4 candidate, one reference entry stood out. Our phase diagram puts Fe22B6W (mp-1224794) on the convex hull. Materials Project puts it 0.148 eV/atom above the hull. The diagram comes from of the energy-above-hull route. The route doesn't reuse MP's DFT energies. It re-relaxes every MP reference with Orb v3 (conservative, inf, MPA) and builds the hull from Orb energies, so the disagreement is Orb versus MP's GGA calculation.