Two Numbers Standard The Old Way GBD First Use New Frontier The Limit The Cost

Insights · Magnet grades

The NdFeB grade chart, explained

Every neodymium magnet grade — and there are roughly two hundred of them in a full catalog — lives at one point on the same two-axis plot. One process moved where that plot's edge sits.

Short versionBr and Hci trade off within a family, because for decades the only way to raise coercivity was to alloy dysprosium or terbium through the whole grain, and that substitution eats into remanence on its way in. Grain boundary diffusion puts the heavy rare earth only where it's needed — at the grain boundary — and leaves the rest of the grain alone. It started as a way to hit an existing spec with less Dy and Tb. It ended up drawing grades that didn't exist before.

Peculiar Materials LLC · Grade points below are representative of a mid-2020s commercial catalog, drawn schematically to show the trend — not a substitute for a datasheet. Real values vary by supplier and by which edge of the tolerance band a given lot lands on. Grades marked with an asterisk are illustrative compositions, not a specific supplier's part number.

What this page skips: bonded and injection-molded NdFeB, which isn't sintered and can't be grain-boundary diffused; anisotropic vs. isotropic powder; and how a magnet is actually made, which is its own page.

Two numbers

Br and Hci, and why they're the whole chart

Standard grades

N35 through N52

The old way

H, SH, UH, EH, AH — and what it cost

Grain boundary diffusion

A couple of microns, not the whole grain

First use

Same spec, less rare earth

New frontier

Grades that didn't exist before

The limit

Depth, grain size, and a ceiling GBD can't touch

The bill of materials

HREE content and cost per kg, at your prices

Scroll to build the chart

Short versionTwo numbers, plotted against each other, are the entire grade catalog.

1 · Two NumbersOne chart, two axes

Neodymium grades get names like N42, N45SH, N38UH. Confusing, until you notice every one of those names decodes to a single point on this plot: how much field the magnet makes, against how much opposing field it survives before that field starts coming back off.

1 · Two NumbersBr and Hci

Br, remanence, is how much flux the magnet puts out at zero opposing field — more Br, stronger magnet, same size. Hci, intrinsic coercivity, is how much reverse field it takes to start erasing that magnetization for good. If you've read the piece on reading a B-H curve, these are two of its four numbers; the other two, Hcb and (BH)max, mostly ride along with these two here.

Rule of thumb: Br sells the spec sheet. Hci decides whether the part still works in six months.

Short versionThe plain N-series is the highest-Br, lowest-Hci corner of the catalog. Fine at room temperature, not fine anywhere hot.

2 · Standard GradesN35 through N52

No suffix means no heavy rare earth added for coercivity, and it shows: N35 up through N52 sit in a narrow band around 11 to 12 kOe, rated to roughly 80°C. The number after the N tracks with energy product — higher number, more flux — and very little else about the grade changes.

In practiceIf a bill of materials just says "52-grade magnet" with nothing else attached, my first question is what temperature it actually sees in service. A fridge magnet couldn't care less. A magnet three millimeters from a winding that's been running for an hour is a different conversation entirely.

Short versionFor decades the only way to buy more Hci was to alloy dysprosium or terbium through the whole grain — and every bit of it fought against the flux you were trying to protect.

3 · The Old WayH, SH, UH, EH, AH

Push a magnet's rated temperature up and it needs more Hci to match, because coercivity falls as a magnet heats. The historic fix: substitute some neodymium with dysprosium or terbium, alloyed through the entire grain. Climb the suffix ladder — M (100°C), H (120°C), SH (150°C), UH (180°C), EH (200°C), AH (230°C) — and the ceiling moves right.

3 · The Old WayWhy it cost flux

Dysprosium and terbium couple antiferromagnetically to the iron in this crystal — their moment points the wrong way. More of them raises coercivity and quietly cancels out some of the very magnetization they were added to protect. Every step right on this chart bought Hci by giving up Br. Follow the old ceiling and it drops the whole way across.

Short versionCoercivity is decided at the grain boundary, in a layer microns thick. The rest of the grain was never the problem — so let's not put the heavies there.

4 · Grain boundary diffusionWhere demagnetizing actually starts

A sintered NdFeB magnet is millions of grains packed together. Reversal — the part of the demagnetizing process Hci measures — doesn't start in the middle of a grain. It nucleates at defects right at the grain boundary, and sweeps across. If you want to prevent that, you work on the grain boundary. Alloying dysprosium through the whole grain to fix a boundary problem is like putting herbicide on your entire lawn to get rid of a few dandelions.

4 · Grain Boundary DiffusionDope the skin, not the block

Grain boundary diffusion applies a dysprosium or terbium compound — a coating, or a melt-spun alloy powder — to the surface of an already-sintered, already-shaped magnet, then heats it just enough for the heavy rare earth to migrate inward along the grain boundaries. It reinforces the boundary layer and mostly stays out of the grain core, where the flux actually lives.

Origin: proposed by Nakamura and coworkers in the early 2000s and commercialized by Hitachi Metals (now Proterial) through most of that decade.^[1]

Short versionThe first payoff wasn't a new grade. It was the old grade, made with a fraction of the dysprosium.

5 · First UseSame spec, less rare earth

Hit an existing spec — say N42SH, 20 kOe of coercivity at the 42 energy-product class — with grain boundary diffusion instead of bulk alloying, and you land on the same point on this chart. What changed is underneath it: dysprosium confined to the boundary skin instead of spread through the whole grain buys the same Hci from far less heavy rare earth.

In practiceThis mattered because dysprosium and terbium are the genuinely scarce, geographically concentrated part of this supply chain — not neodymium. A process that gets the same spec sheet out of less of them is a supply chain story before it's a materials science one.

Short versionOnce Hci and Br could move almost independently, the obvious next move was to stop matching the old spec and start beating it.

6 · A New FrontierThen somebody asked the next question

If grain boundary diffusion barely touches Br while it raises Hci, the old ceiling was never a law of the material — it was a limit of bulk alloying. Push the same technique further and you can chase a grade that combines a Br you used to only get at a lower temperature class with an Hci you used to only get by giving that Br away.

6 · A New FrontierGrades that didn't exist before

The result sits above and left of the old dashed line — the Hci of an old H, SH, UH or EH grade, at roughly the Br of the next size up. Still not free: Dy and Tb are still in the part, just far less of it, placed far more efficiently.

Short versionDiffusion reaches as far as diffusion reaches. Grain size and part thickness set the limit, not chemistry.

7 · The LimitA different ceiling, not a missing one

Grain boundary diffusion works from the outside in, along boundaries, so how far it penetrates depends on how fine the grains are and how thick the part is. Patent data on the process puts effective penetration around 2 to 3 millimeters from the surface,^[2] which is exactly why it showed up first in thin rings and discs rather than large blocks — and why it pairs so well with the finely jet-milled powder from the sintering process: more grain boundary area per gram, more places for the diffusion to reach.

7 · The LimitThe ceiling nobody diffuses past

There is a real ceiling here, and it isn't a process problem. Nd₂Fe₁₄B's own anisotropy field caps how much coercivity the crystal can hold, full stop — about 75 kOe. Commercial magnets, GBD or not, typically reach only 15 to 20% of that;^[3] the gap has its own name, Brown's paradox, and forty years of literature trying to close it. Grain boundary diffusion moved the practical frontier a long way. It never touched the theoretical one.

Next: what any of this costs — below the chart.

Magnet grades

8 · The bill of materials

What the heavy rare earth actually costs

The chart above is physics. This is the part that shows up on a purchase order. Every number below is calculated live from the prices in the boxes — change them and watch the argument for grain boundary diffusion get stronger or weaker, because that is exactly what it does in the market.

Price basis: · GBD diffusion source:

NdPr alloy $/kg

Dy metal $/kg

Tb metal $/kg

Balance $/kg

GBD step $/kg

Class	Hci	Bulk HREE	GBD HREE	Bulk $/kg	GBD $/kg	Δ cost w/ GBD

Model: 31 wt% total rare earth in the magnet, base coercivity 12 kOe with no heavy rare earth added. Bulk alloying is taken at roughly 2 kOe of Hci per wt% Dy; grain boundary diffusion at about three times that efficiency for Dy and six times for Tb, since the heavy rare earth sits where reversal actually nucleates instead of being spread through the grain core. Heavy rare earth displaces NdPr one-for-one in the total. These are rules of thumb for showing the shape of the trade, not a costing model — real loadings vary by supplier, part geometry and how hard the grade is pushed within its class.

Raw material cost per kg of magnet

Standard (N)

No added heavy rare earth. Hci around 11–12 kOe, rated to about 80°C. Highest Br for the least cost.

Bulk-doped (M–AH)

Dysprosium or terbium alloyed through the whole grain. Hci climbs with the suffix; Br drops the whole way across. The old ceiling.

GBD

Dy/Tb at the boundary only. Roughly two-thirds less heavy rare earth for the same Hci — then Br and Hci that used to be mutually exclusive, together.

References

Most of this page is materials physics, which needs no citation. These are claims about specific numbers and specific history, so they get one each.

Y. Liu et al., "Advances in grain-boundary diffusion for high-performance permanent magnets", Materials Futures. Credits Nakamura and coworkers with proposing grain boundary diffusion for NdFeB; reviews the mechanism and its industrial adoption.
Hitachi Metals, "Sintered NdFeB magnet and method for manufacturing the same", US Patent 9,589,714. Describes a rare-earth-rich grain-boundary layer continuing to a depth of 2.5 mm from the treated surface — the figure used above for effective penetration.
C. Liu, "Most frequently asked questions about the coercivity of Nd-Fe-B permanent magnets", Frontiers in Physics. States the anisotropy field of Nd₂Fe₁₄B at roughly 7.5 T (μ₀Hₐ) and typical commercial coercivity at about 15–20% of that value — the basis for "Brown's paradox" above.
Shanghai Metals Market benchmarks as reported by Critical Minerals News and rare-earth-mining.com, early September 2026. Default price basis for the calculator: NdPr alloy $132/kg, Dy metal $239/kg, Tb metal $1,097/kg, all China domestic, VAT excluded. The Western/FOB preset uses indicative dealer quotes, which have run several times the domestic benchmark through 2026 — the two-market structure is real and it matters more to this calculation than any other input.

Work with Peculiar Materials

Technical and financial diligence on magnetics, magnet manufacturing, and the rare-earth supply chain — from oxide to finished magnet.

Get in touch