Bonded magnets, type by type

Short versionEverything on this page is measured against a plain sintered N42. It is the yardstick because it is the material you would use if shape and cost did not matter — and they always matter.

1 · The referenceStart with the magnet you already know

The black curve is the second-quadrant demagnetization curve of a sintered NdFeB magnet, mid-grade, room temperature. Remanence around 1.3 T, coercivity near 900 kA/m, energy product a little over 300 kJ/m³ (about 42 MGOe). If the axes look unfamiliar, the B-H curve page builds them up from scratch; here I am going to assume you can read one.

It stays on the figure the whole way down. Every bonded curve gets drawn underneath it, to scale, so you can see exactly what you are paying.

1 · The referenceWhat sintering costs you

While a sintered magnet is the strongest thing you can buy, almost everything about making it into a part works against you. It is a ceramic-like brittle block that has to be ground to shape with diamond wheels. Thin sections chip. Rings are turned from solid, which throws away most of the material. Every cut edge needs a coating, or it rusts. And the magnetization direction is fixed at pressing — you get the pattern the die gave you and no other.

So the honest question a designer asks is not "what is the strongest magnet?" It is "what is the strongest magnet I can get in THIS shape, at THIS price, in THIS quantity?" That question is where bonded magnets live.

Kitchen analogySintered is a cast-iron skillet: heavy, strong, and it comes in the shapes it comes in. Bonded is a silicone mold: you get exactly the shape you drew, and it will never be cast iron.

Short versionPlastic has no magnetism. Whatever fraction of the part is plastic is a fraction of the flux you do not get. Remanence scales with the magnet loading; energy product scales with the SQUARE of it.

2 · DilutionThe one line of math

Most bonded NdFeB starts as melt-spun flake: molten alloy quenched on a spinning wheel so fast that the grains come out tiny and pointed every which way. That flake is isotropic — it has no preferred direction — and even fully dense it would only reach a remanence around 0.8 T, the amber curve. That is the first price, and it is paid before any plastic shows up.

Then you dilute it. Keep scrolling and watch the curve shrink as the loading drops from 100% flake to the roughly 80% by volume you get with epoxy and a press:

where f is the volume fraction of magnet in the part. It is not exact — packing, alignment and squareness all lean on it — but it is close enough that you can do bonded-magnet arithmetic on a napkin.

2 · DilutionWhy the square matters

The square is the part people forget. Going from 80% loading to 60% costs you a quarter of your remanence and nearly half of your energy product. Every bonded-magnet process below is, at bottom, a fight to get more powder into the part while still being able to form it. The processes that flow easily hold less powder. The processes that pack tightly do not flow at all. Pick your poison.

In practiceThe spindle motor in the hard drives I got my start erasing almost certainly ran on a compression-bonded NdFeB ring — a thin, multipole ring is the textbook bonded-magnet part. The magnet I was working so hard to erase was a different thing entirely: the nanometers-thick coating on the platter. I did not know the difference at the time, and I suspect nobody in the room did either.

Short versionIsotropic NdFeB flake, a couple of percent epoxy, pressed cold and cured. The highest loading of any bonded process, so the strongest of the common bonded magnets — and the one whose parts most look like a magnet rather than a plastic part.

3 · Compression-bonded NdFeBThe workhorse

The green curve is a typical compression-bonded grade: remanence about 0.7 T, intrinsic coercivity around 700–750 kA/m, energy product 80 kJ/m³ or so — call it 10 MGOe.¹ Roughly half the remanence of the sintered reference and a quarter of the energy product. That is the price of the whole category, and it has been about the same since the 1990s.

What you get for it: the part comes out of the die at final size. No grinding, no diamond wheels, tolerances a sintered magnet cannot hold without finishing. And because the flake is isotropic, you can magnetize it in ANY pattern after the fact — axial, radial, twelve poles around a ring, a Halbach array on a flat plate. The sintered magnet fixed its direction at pressing. This one waits for you to decide.

3 · Compression-bonded NdFeBPluses and minuses

• Strongest of the common bonded types — the most powder per part
• Near-net shape, tight tolerances, thin walls and rings that sintered cannot do economically
• Isotropic: magnetize in any direction or pattern, after molding
• Roughly straight second-quadrant B curve, so it shrugs off demagnetizing fields a sintered magnet of the same B_r would not
• Half the flux and a quarter of the energy of sintered, for a part that contains 80% of the same rare earth by volume — the rare earth is not cheaper per tesla, it is dearer
• Still brittle: it is a pressed powder compact, not a plastic part, and it chips
• Needs a coating; the flake rusts just like sintered does
• Working temperature capped by the epoxy and by the flake's own coercivity loss — datasheets top out somewhere around 150–160 °C for the best grades²
• Shapes are limited to what a press can push straight down: no undercuts, no overmolding

Used in: spindle and stepper motor rings, small brushless motors, sensor targets, encoder wheels, anything that needs many poles on one small part.

Short versionThe same flake in a thermoplastic — nylon, or PPS if it has to run hot — molded on an ordinary injection press. Less powder, so less magnet, but you can mold it straight onto a shaft, put threads in it, and make shapes no press could.

4 · Injection-molded NdFeBA plastic part that happens to be a magnet

Here is where dilution bites. To flow through a gate and fill a mold, the compound has to be maybe 55–65% powder by volume, not 80%. The new curve shows it: remanence around 0.5 T, energy product in the 40–50 kJ/m³ range, roughly 5–6 MGOe.³ About half of compression-bonded, about an eighth of sintered.

In return you get the entire injection-molding toolbox. Overmold onto a steel shaft or a stamped rotor. Mold in the gear teeth, the snap fits, the mounting bosses. Hold true plastic-part tolerances. And the part is tough, in the way a glass-filled nylon part is tough — drop it and it bounces. A compression-bonded ring dropped on concrete is a story you tell later.

4 · Injection-molded NdFeBPluses and minuses

• Any shape a mold can make, including undercuts, integral features and overmolding onto inserts
• Tough, not brittle; handles like a plastic part because it is one
• Isotropic, so free magnetizing patterns, same as compression
• Cheap per part at volume — it is a molding cycle, not a press-and-cure-and-coat sequence
• Thermoplastic binder shields the flake somewhat, so corrosion resistance is better than compression-bonded (though not immune)
• Weakest of the NdFeB bonded routes — roughly 40% of the sintered remanence
• Working temperature set by the plastic: nylon grades run cooler than PPS grades, and both run cooler than a sintered H-grade
• The abrasive compound eats the tooling; hardened screws, barrels and molds are the cost of entry
• Mold tooling is expensive, so the economics only work above a few tens of thousands of parts

Used in: pump rotors, automotive sensors, small gearmotors, water-meter and flow-meter magnets, anything overmolded onto a shaft.

Short versionPush the compound through a die instead of into a mold and you get a continuous profile: strips, tubes, thin-wall rings cut to length. Rigid in nylon; flexible in nitrile rubber. Same flake, similar loading to injection, and one shape family the others cannot touch.

5 · Extruded and flexible NdFeBAnything, as long as it is long

The curve lands near the injection-molded one, sometimes a touch above it because a rigid extrusion can carry a little more powder than a compound that has to fill a complex mold: remanence in the 0.5–0.6 T range, energy product perhaps 50–60 kJ/m³.⁴ Flexible NdFeB in a rubber binder sits below that, closer to 0.5 T, because rubber will only take so much powder before it stops being rubber.

What extrusion buys you is thin, long, and uniform. A ring magnet with a half-millimeter wall, two meters of it, sliced. A flat strip for a linear encoder. The cross-section can be anything the die can hold — and nothing else, because the whole point is that it does not change along its length.

5 · Extruded and flexible NdFeBPluses and minuses

• Thin walls and long lengths that neither pressing nor molding can produce
• Very low tooling cost — an extrusion die is cheap next to a mold
• Flexible grades bend around a curved surface and can be adhesive-backed
• Isotropic, free magnetizing patterns
• Constant cross-section only; no features along the length
• Injection-molded-level strength at best; flexible grades are weaker still
• Rubber binders are the least heat-tolerant of the lot; keep the flexible grades cool
• Cut ends expose flake, so ends need attention if the environment is wet

Used in: long thin-wall motor rings, linear-encoder strips, magnetic seals and latches, flexible NdFeB in places a ferrite strip is not strong enough.

In practiceA confession about the word. I spent a long stretch of my career on magnet bonding — adhesion, how you glue a sintered magnet to a rotor so it stays there at speed — and for years the phrase "bonded magnet" made me think of that. It has nothing to do with it. In a bonded magnet the glue is on the inside, holding powder to powder. Two fields, one word, and nobody is going to fix it now.

Short versionSwap the flake for strontium ferrite powder and you get the cheapest magnet on earth, in sheet, strip and molded form. Look at where the curve lands. Then look at what a refrigerator door costs.

6 · Bonded ferriteThe fridge magnet, taken seriously

The curve is nearly on the floor: remanence 0.2–0.3 T, energy product 5–15 kJ/m³, from under one MGOe to about two. That is a fifth of the remanence of the sintered reference and a twentieth of its energy, and well under half of even a compression-bonded NdFeB. And it is still, by volume shipped, the most common permanent magnet on the planet, because the powder costs about what sand costs and it does not rust.

Two forms matter. Calendered sheet — ferrite in a rubber binder, rolled flat — is the fridge magnet, the car door sign, the shower door seal. The rolling can partly align the platelets, which is why "anisotropic" flexible ferrite exists and is noticeably stronger than the isotropic kind. Injection-molded ferrite is the same idea in nylon, and here you CAN align it: apply a field while the compound is molten in the mold and the platelets line up. Anisotropic injection-molded ferrite reaches the top of that low range and makes an entirely respectable small motor.

6 · Bonded ferritePluses and minuses

• Cheapest magnet material there is, by a wide margin
• Does not corrode — no coating, ever
• Coercivity RISES with temperature, the opposite of every rare-earth magnet; the risk with ferrite is the cold, not the heat
• Easy to magnetize, so multipole patterns are cheap to put in
• Flexible sheet can be die-cut, printed on, and stuck to things
• Weak. Whatever you were going to do with it, do the flux arithmetic first
• Low remanence means big, and big means heavy — the part grows to make up for the material
• Anisotropic grades are direction-locked at molding, like sintered

Used in: fridge magnets and signage, door seals, toys, small DC motors, sensor and reed-switch actuators, anything where the magnet must be nearly free.

Short versionMake the powder anisotropic instead of isotropic, align it in a field while you press, and the ceiling roughly doubles. This is the strongest bonded magnet you can buy — and it gives back the one freedom the isotropic ones had.

7 · Anisotropic bonded NdFeBThe ceiling

The isotropic flake threw away a big chunk of the material's potential before any plastic arrived. What if the powder itself were aligned? The route that works is HDDR — hydrogenation, disproportionation, desorption, recombination — a hydrogen treatment that breaks a cast NdFeB alloy into sub-micron grains and, done slowly enough, has them recombine sharing the parent grain's orientation. Each powder particle is then a tiny oriented magnet, and you can line them up in a field while you compress the part.⁵

The curve jumps: remanence near 1 T, energy product approaching 200 kJ/m³ — the best commercial grades claim around 25 MGOe, with working temperatures to 150 °C.⁶ That is more than half the sintered reference, from a part that never saw a sintering furnace or a grinding wheel.

7 · Anisotropic bonded NdFeBPluses and minuses

• Strongest bonded magnet available — roughly twice the energy of compression-bonded isotropic
• Still near-net shape, still no grinding, still thin rings and small motors
• Some grades are injection-moldable with field alignment, giving a molded anisotropic part
• Direction is locked in at forming, like sintered: the free-magnetizing-pattern trick is GONE, or at least much harder
• A very short list of powder suppliers — this is a Japanese specialty, and the supply picture is narrower than for isotropic flake
• Costs more per kilogram than isotropic flake, so the flux-per-dollar argument is not automatic
• Corrosion and coating story is the same as any NdFeB

Used in: small high-performance motors where a sintered ring is too expensive to make and an isotropic ring is too weak — power tools, automotive actuators, e-bike hubs.

In practiceA hallmark of overly ambitious pitch decks in this field is the claim of a "new bonded magnet" that is really an isotropic bonded magnet with an unfamiliar binder. The curve in the deck is the tell. If it is not above the green compression-bonded line on this figure, it is not new; it is repackaged. If it IS above that line, ask how the powder was aligned, and by whom.

Short versionSamarium-iron-nitride has the intrinsic properties to rival NdFeB. It also decomposes before it will sinter, so it exists only as a bonded magnet. Better than NdFeB flake in isotropic form, better at heat, and it does not need a coating.

8 · SmFeNThe chemistry that can only be bonded

Sm₂Fe₁₇N₃ is a lovely magnet on paper — high saturation, very high anisotropy, a Curie temperature well above NdFeB's. The catch is that the nitrogen leaves if you get it anywhere near sintering temperature, so you cannot make a dense block of it. What you CAN do is make a fine powder and bond it, and the curve shows the result for the isotropic compression-molded kind: remanence around 0.75–0.8 T, energy product around 95–105 kJ/m³ — meaningfully above the isotropic NdFeB workhorse.⁷

Anisotropic SmFeN bonded magnets go higher still, and there is a steady stream of research on hybrid SmFeN/NdFeB compounds that pack the two powder sizes together to push loading up. It is the most interesting corner of the bonded world right now, if you like that sort of thing, and I do.

• Higher energy than isotropic NdFeB flake, in the same processes
• Better temperature coefficient of remanence than NdFeB bonded, and better long-term thermal stability
• Good corrosion resistance — the supplier datasheets say no coating in ordinary environments
• No heavy rare earth: no dysprosium or terbium in the powder
• Harder to magnetize than NdFeB bonded — plan for it
• Fine powder, so it loads less easily, and the injection-molded grades give up more than NdFeB does
• Few suppliers, mostly in Japan; volumes are small next to NdFeB flake
• Cannot be sintered, ever. This is the bonded family's ceiling, not a path to a sintered rival

Short versionSamarium-cobalt powder in epoxy. Weak for its price, but it keeps its coercivity when everything else on this page has given up, and it does not rust. Nobody chooses it for fun.

9 · Bonded SmCoThe one you pick when the environment picks for you

The curve sits down with the injection-molded NdFeB — remanence around 0.5–0.6 T, energy product in the 50–80 kJ/m³ range depending on grade and process.⁸ Look at the coercivity, though: the intrinsic curve runs far out to the left, and it is still out there at temperatures that have long since cooked the NdFeB binder. Working temperatures are set by the epoxy, not the powder, and with a high-temperature binder that is a real number.

SmCo also does not corrode in any ordinary sense, so the part needs no coating, and it is very hard to demagnetize. Those three things — heat, no rust, no demagnetization — are the whole case. Against them: cobalt and samarium are expensive, the powder is abrasive and brittle, and the parts are correspondingly fragile. This is the bonded magnet for the sensor next to the exhaust manifold, not for the toy.

• Best hot-side performance of any bonded magnet; coercivity holds up
• No coating needed
• Near-net shape, multipole magnetizing, all the bonded conveniences
• Expensive raw material for modest flux
• Brittle; handle with care
• Small market, few suppliers, long lead times

Short versionThree things the curves do not show and the datasheets bury: the binder sets the ceiling, the flake still rusts, and an isotropic bonded magnet is harder to magnetize than you expect.

10 · Heat and fine printWhat the room-temperature figure hides

Heat. Every rare-earth bonded magnet loses remanence and coercivity as it warms, and NdFeB flake loses coercivity fastest. But the flake's own limits are rarely what stops you. The binder is. Epoxy softens; nylon softens sooner; rubber sooner still; PPS holds on longest. The "maximum operating temperature" on a bonded-magnet datasheet is a statement about the plastic at least as much as about the magnet, and it is a rule of thumb, not a law — a magnet running close to it in a demagnetizing field will lose more than the number implies.

Rust. Compression-bonded NdFeB is a compact of rare-earth flake with a little epoxy between the particles. Water gets in. It gets a coating, same as sintered — epoxy, or a parylene layer, or e-coat — and the coating is part of the cost. Injection-molded parts fare better because the thermoplastic wraps each particle. Ferrite and SmCo do not care, and SmFeN mostly does not.

Magnetizing. Isotropic NdFeB flake takes a large field to saturate, larger than ferrite by a wide margin, and a multipole pattern on a small ring means a fixture with many fine poles trying to deliver that field into a small volume. Getting every pole on a sixteen-pole ring to the same strength is a competence, and the shops that have it are not interchangeable with the shops that do not. Ask to see the pole-to-pole scatter data before you assume it is fine.

10 · Heat and fine printThe cousins that are not bonded

Two things get filed under "bonded" that are not, and it is worth knowing the difference. Hot-pressed isotropic NdFeB is melt-spun flake compacted hot, with no binder, to full density — the amber curve from the dilution section, made real. Die-upset (hot-deformed) NdFeB takes that hot-pressed puck and squashes it so the grains align, giving a fully dense anisotropic magnet, including radially oriented rings that sintering cannot make. Neither has a gram of plastic in it. They share the powder with bonded magnets, not the process, and their curves belong on the sintered side of this figure, not the bonded side.

And 3D-printed bonded magnets — flake in a filament or a binder-jet powder bed — are genuinely a thing, in the sense that national labs have made them and a few companies will sell you one. They are not yet a category a designer specifies, so they get this sentence and no curve. I will date-stamp a revision here when that changes.

Short versionDecide shape first, temperature second, strength third, cost last. If you go in the other order you will specify a sintered magnet you cannot make.

11 · Picking oneThe decision, in order

1. Can the part be a block? If yes, and you need the flux, it is sintered and this page was a nice read. If it needs to be a thin ring, a multipole anything, or overmolded onto something, keep going.

2. How hot, and how wet? Over about 150 °C, or in a corrosive place with no coating allowed, you are looking at bonded SmCo, or a PPS-bonded SmFeN, or you are looking at the wrong page. Ferrite if the flux budget allows and it never gets cold.

3. How much flux, really? Do the dilution arithmetic against the figure. Anisotropic HDDR if you must have the most and the direction can be fixed. Compression-bonded isotropic if you need the free magnetizing pattern. Injection-molded if you need the shape more than the last 30% of flux.

4. How many? Compression tooling is cheap and injection tooling is not, so quantity breaks the tie between them. Extrusion is cheapest of all and only makes one shape.

The figure now has every family on it. The gap between the black line and everything else is the honest size of the bonded-magnet tax. Whether it is worth paying is a question about your part, not about the magnet — which is, I think, the useful thing to have learned.