Wade

Investors seem irritated with the mine-to-magnet companies’ stock prices, which reflect that they still can’t produce at scale. Nobody who has ever made a magnet is surprised by that. January 1, 2027 is when DFARS stops taking excuses. The physics doesn’t care about deadlines or share prices. Making a sintered Nd-Fe-B magnet from separated metals takes roughly a dozen tightly controlled steps. A defect early in the process can limit everything that follows. Some highlights: • Melting and strip casting. The metals are melted under vacuum and rapidly solidified on a spinning water-cooled copper wheel into thin flakes. The quench has to be fast enough to avoid leaving free iron, or the material won’t powder or align properly in later steps. • Hydrogen decrepitation. The flakes are exposed to warm, pressurized hydrogen. A rare earth-rich phase absorbs the gas, swells, and cracks the flakes into a coarse powder. The hydrogen is then baked out under vacuum. Leaving it in hurts coercivity, the resistance to demagnetization. • Jet milling. The powder is reduced to particles a few microns across, ideally single crystals. From here on it stays under inert gas, because fine Nd-Fe-B burns in air (it was always a good day in my lab when dirty tools caught fire upon leaving the glovebox). Particle size largely determines grain size after sintering, which strongly influences coercivity. • Alignment and pressing. The powder is blended with a lubricant, poured into a die, aligned in a magnetic field, and compacted. Alignment drives remanence, a measure of how strong the magnet is, and the pressing method alone can move it by 5–8%. • Vacuum sintering. Above 1000°C the compact densifies as particles bond and grains grow. Excessive temperature or time lets a few grains grow enormous, creating weak spots where demagnetization can begin. • Annealing. Lower-temperature heat treatments modify the chemistry and distribution of the thin rare earth-rich phase at the grain boundaries. One study reported a 37% gain in coercivity from annealing alone. • Machining, coating, magnetizing. Sintered blocks are hard and brittle; shaping them takes specialized tooling and is the biggest material loss in the chain (roughly 10–40%). Parts are coated because corrosion attacks along the grain boundaries the anneal just built, then magnetized with pulse fields more than twice the magnet’s coercivity. These steps require tight control, and they can introduce impurities. Oxygen from milling, carbon from the pressing lubricant, and nitrogen from the mill gas all consume the small excess of Nd that forms the grain boundary phase, resulting in crappy coercivity. Upstream processing defects cannot be fixed downstream. Their cumulative effects ultimately show up when the finished magnet is tested. Magnet grades are specified by the performance, not the process to get there, and each plant has to tune its own recipe. I’m only describing one stage. Mining, separation, metallization, alloying, and magnet making are each their own discipline, with their own steps and their own learning curve. Process knowledge is built by running real material through real equipment. This knowledge cannot be found in a book. A company promising the whole chain has signed up to climb every one of those learning curves at once, on one schedule and one balance sheet. A slip at any stage propagates through everything downstream of it. When the national rebuild rides on one or two of these companies, their slips become America’s slips. The DFARS deadline is really counting qualification. One magnet that meets spec proves nothing. The customer starts qualifying when the supplier can reproduce the performance batch after batch. The integrated bets aren’t necessarily a mistake, just a more complex route. Some of these bets will pay off, and US supply chains will be stronger when they do. But they are not where speed comes from, and the stock prices are reflecting that. Speed comes from companies mastering one link, with one learning curve to climb and one balance sheet to climb it with.