Asking the question properly
“Rare-earth-free” is usually the wrong framing, because it bundles together two very different exposures with very different solutions.
For most buyers, the heavy rare earth question is the one that matters
It is where the cost premium sits, where the licensing exposure sits, and where the supply is genuinely scarce. It is also the one with a mature, commercially available answer — and the least discussed, because it is unglamorous compared with new materials. If you read nothing else here, read section 02.
How the options sort
| Tier | Options | Status |
|---|---|---|
| Available now | Grain boundary diffusion, ferrite redesign, recycled NdFeB, cerium substitution | Commercially purchasable today, at scale, with real supply chains |
| Real but narrow | SmFeN bonded magnets, advanced La-Co ferrite, alnico | Commercial, but suited to specific application windows |
| Emerging | Iron nitride, MnAlC, MnBi | Early commercial or pre-commercial; limited grades, limited geometry, limited supply |
| Research | Fe₂NiZn, AlFe₂Ni, Co₂MnSb and similar computational candidates | Papers, not parts. Years from qualification even if they succeed |
Grain boundary diffusion — the available answer
The most useful and least publicised option on this page. Grain boundary diffusion achieves high-temperature coercivity using substantially less dysprosium or terbium than conventional bulk alloying, and it is a mature production process available from established mills today.
How it works
Coercivity in sintered NdFeB is governed at the grain boundaries, where reverse domains nucleate. Conventional high-temperature grades add dysprosium or terbium throughout the alloy, so the heavy rare earth is distributed everywhere — including through the grain interiors, where it raises cost and slightly reduces remanence without contributing to coercivity.
Grain boundary diffusion instead applies the heavy rare earth to the surface of an already-sintered magnet as a hydride, fluoride or metallic compound, then heat-treats so it diffuses along the grain boundary network. The heavy element concentrates where it does the work.
Do this first, and it may be all you need
Before evaluating exotic materials, run the two conventional checks: is the temperature class over-specified, and can the required coercivity be met by grain boundary diffusion instead of bulk alloying? Between them these two questions resolve the majority of heavy rare earth exposure in real designs, using material you can buy today with no redesign. The load-line method for the first is in the temperature guide.
Ferrite redesign — the largest saving
The only genuinely rare-earth-free option that is available at unlimited scale, at low cost, with a mature supply chain, today. It is also the one most often dismissed without analysis.
| Attribute | Ferrite versus sintered NdFeB |
|---|---|
| Rare earth content | None. Iron oxide with strontium or barium carbonate |
| Export licensing | Not applicable |
| Cost per unit energy | Roughly an order of magnitude lower |
| Energy product | Far lower — the whole trade-off. Requires substantially more volume for the same field |
| Corrosion | Already an oxide; needs no coating |
| Tariff classification | Falls under a different subheading from metallic magnets — see tariffs and origin |
| Low temperature | Requires care — ferrite coercivity behaviour on cooling is the opposite of NdFeB, and cold-environment demagnetization is a real failure mode |
Where the substitution genuinely works
- Holding, latching and door applications where the envelope has room and mass is not penalised.
- Magnetic separation — already predominantly ferrite in most industrial equipment.
- Loudspeakers outside weight-critical applications.
- Low-duty motors and actuators where power density is not the design driver.
- Sensor targets at short working gaps, where the field requirement is modest.
Where it does not
- Traction and high power-density motors, where the mass and volume penalty defeats the purpose of the machine.
- Compact assemblies with no room to grow the magnet.
- Precision sensing at longer gaps, where the field simply is not available.
- Anywhere the design has already been optimised around NdFeB’s energy density.
Advanced ferrite is worth a second look
Lanthanum-cobalt substituted ferrite grades and improved microstructure control have pushed ferrite performance meaningfully above the classic Y-series figures many engineers carry in their heads from a decade ago. It remains far below NdFeB, but the gap at the margin has narrowed, and some applications that were judged infeasible on older ferrite data are feasible now. If your last ferrite evaluation is more than a few years old, the arithmetic may have changed. Properties are in the material comparison.
Recycled NdFeB
Not a different material — the same sintered NdFeB, made from recovered feedstock rather than newly mined ore. It addresses supply concentration and sustainability reporting without changing the engineering.
Practical considerations
- Qualify it as a source change, because it is one. Composition and magnetic properties from recycled feedstock require the same cross-comparison as any second source.
- Grade range is narrower than primary material, particularly at the top of the energy product range and in high-coercivity classes.
- Capacity is still small relative to demand. Treat it as one strand of a supply strategy rather than a replacement for primary sourcing.
- Ask for the recycled content percentage and its basis, not a general claim — the same discipline you would apply to any compliance declaration.
The emerging materials, assessed honestly
These attract most of the coverage and almost none of the purchase orders. Each is real; none is a drop-in replacement for a production NdFeB part today.
| Material | Genuine strength | Real constraint | Practical status |
|---|---|---|---|
| SmFeN samarium iron nitride |
Excellent coercivity; well-proven material system; good temperature behaviour | Still a rare earth — and samarium is itself export-controlled. Decomposes at sintering temperatures, so it is produced as bonded or injection-moulded parts, limiting energy product | Commercial from Japanese producers; a genuine option for bonded applications, not a sintered NdFeB replacement |
| Iron nitride Fe₁₆N₂ |
Iron and nitrogen only — abundant, cheap, no supply concentration. High Curie temperature | The widely quoted energy product figures are theoretical maxima for the pure phase. Realised commercial products are far below them. Phase stability and scale-up remain the hard problems | Early commercial, narrow product range. Positioned between ferrite and neodymium rather than replacing neodymium |
| MnBi manganese bismuth |
Coercivity increases with temperature — the opposite of NdFeB, and genuinely valuable for hot applications | Low energy product; difficult phase purity control; bismuth is not abundant | Research and niche. Interesting where the thermal behaviour is the whole point |
| MnAlC | Abundant elements; low density; previously commercialised so the process base exists | Modest performance; was displaced by NdFeB for good reasons | Being revisited with modern processing; not a general option |
| Cerium-substituted NdFeB | Cerium is abundant and co-produced, so it is cheap; partially replaces neodymium | Lower remanence and coercivity; still contains rare earths, just less valuable ones | Commercially used in cost-driven, low-duty applications |
| Computational candidates Fe₂NiZn and similar |
High-throughput screening is producing genuinely new candidates | Synthesis, scale-up and commercial validation all unproven. The gap between a promising calculation and a qualified part is measured in years | Research. Not a procurement input |
How to read the claims
Two specific things to watch in coverage of new magnet materials. First, theoretical energy products quoted as though they were product specifications — the maximum attainable for an ideal single phase is not what a manufactured, coated, machined part delivers. Second, rare-earth-free being conflated with supply-secure: a material can be free of rare earths and still depend on a scarce or concentrated input.
The question that cuts through it: can I buy a hundred thousand parts to my drawing, to a stated specification, with a lead time and a certificate? For most of the table above, in mid-2026, the answer is not yet.
What to actually do
A sequence that produces results, ordered by return per unit of engineering effort.
| # | Action | Reduces | Effort |
|---|---|---|---|
| 1 | Right-size the temperature class against a load-line analysis at the real maximum | Heavy rare earth content, cost, lead time, licence exposure | Low — a calculation and a requalification |
| 2 | Move to grain boundary diffused material where geometry permits | Heavy rare earth content, at equal performance | Low to medium |
| 3 | Audit the portfolio for ferrite candidates — holding, latching, separation, low-duty | All rare earth content, and most of the cost | Medium — redesign per part |
| 4 | Qualify recycled-content NdFeB on suitable part numbers | Primary supply concentration; supports sustainability reporting | Medium — treat as a source change |
| 5 | Consider SmFeN for bonded applications already using bonded material | Nothing in licensing terms — but a genuine alternative supply route | Medium |
| 6 | Monitor iron nitride and MnBi without designing around them yet | Nothing today | Low — watch, do not commit |
The summary
There is no drop-in rare-earth-free replacement for sintered NdFeB, and anyone offering one is describing a research result. What does exist is a set of practical moves that materially reduce exposure: use less heavy rare earth by specifying the class correctly and by using grain boundary diffusion; use no rare earth at all where ferrite can carry the application; and diversify the feedstock where recycled material qualifies. Together those cover most of what is actually achievable in 2026 — and most organisations have not done the first two.
The parallel supply-side moves — buffering, second-sourcing and domestic stock — are covered in inventory programs and the second-source playbook. Material substitution and supply strategy work together; neither is sufficient alone.