Ranking and Recommending Permanent Magnet Materials by Cost-Effectiveness Across Operating Temperatures

Cost-effectiveness across operating temperature is set mainly by usable energy after thermal derating, per unit of delivered cost, not by the room-temperature figure on the datasheet. For envelope-constrained designs the usual order is sintered NdFeB with little or no heavy rare earth at low magnet temperature, higher-coercivity or grain-boundary-diffused NdFeB as the knee rises, and SmCo once that knee cannot be cleared at acceptable mass. Ferrite leads on cost only when a larger, heavier magnet is acceptable. Crossovers move with circuit permeance. Two materials can rank in opposite order in one application if that permeance differs.

The controlling temperature is the magnet's, not the ambient

Rank on magnet temperature, not ambient air. The magnet often runs hotter than the air around it; in a rotor can, a gearbox-adjacent housing, or a poor thermal path the offset is design-specific and can be large. Ambient air as the input is the usual cause of an undersized grade.

Three effects compound:

• Reversible loss. Remanence falls with temperature: about −0.11 to −0.12 %/K for sintered NdFeB, about −0.03 to −0.045 %/K for SmCo, and near −0.20 %/K for ferrite. The loss returns on cooling. It costs torque headroom, not the part.
• Irreversible loss. Intrinsic coercivity of NdFeB falls much faster with temperature than remanence, so the knee moves up into a fixed load line. Crossing that knee means induction does not fully recover. SmCo's coercivity falls more slowly, so it can hold a load line NdFeB has lost.
• Long-term aging. Time at temperature accelerates magnetic aging. The allowance should grow with temperature and dwell. Aging is separate from the reversible coefficient.

Cost-effectiveness drops when knee margin forces a higher coercivity grade. On NdFeB that coercivity is often bought with dysprosium and terbium; microstructure and grain-boundary diffusion reduce how much is required.

Why the cost curve is not monotonic in grade

Dysprosium and terbium raise intrinsic coercivity and allowable temperature, but they depress remanence and raise price per kilogram. Too little coercivity means a thicker magnet or demagnetization risk; too much means paying for unused coercivity while giving away energy product. Stop where the knee at temperature clears the peak demagnetizing field with a defined margin.

Grain-boundary diffusion of Dy/Tb changes the ranking, not only the price. The heavy rare earth is concentrated where coercivity is generated, so a diffusion grade can match a homogeneous heavy-rare-earth grade at lower total heavy-rare-earth content and with less remanence loss. Diffusion depth is finite: a very thick section may not carry the surface coercivity through the volume. Where the depth covers the section, this route is often the better cost per usable energy.

Ranking by temperature band

Bands below are magnet temperature under a typical motor or actuator load line. Apply the allowance in Design windows before locking a suffix. Classes N, M, H, SH, UH and EH are common industry catalogue labels, conventionally near 80, 100, 120, 150, 180 and 200 °C. They do not transfer to a thinner magnet or a harsher load line.

Where a larger rotor is acceptable, ferrite near 150 °C can still be the cheapest route to a given flux: energy product is low, but unit material cost is far below NdFeB. The table assumes mass and envelope are constrained. State that constraint first.

A single relation that decides the crossover

In the linear range, well below the Curie point,

with ${\alpha }_{Br}$ in K${}^{-1}$ and negative. For fixed permeance and recoil permeability near ${\mu }_0$ — a better fit for NdFeB and SmCo than for ferrite — operating-point energy density scales about as $B_r^2$. A 12% drop in $B_r$ is then about a 23% drop in that energy term: an illustration of the square, not a measured loss.

Energy density per unit material cost then ranks as

with $\rho$ the density and $p$ the price per kilogram. Dividing $B_r^2$ by price per kilogram alone mixes a volume quantity with a mass price.

Heavy rare earth lowers $B_r(T)$ and raises $p$. Extra thickness to clear the knee is not a cut in $B_r$; it adds mass, so it sits outside this index. Below the crossover, NdFeB leads on price. Above it, non-diffused grades lose on both remanence and price, which moves the ranking toward diffusion grades and then to SmCo. Recompute when temperature, load line, or heavy-rare-earth price changes: a grade that leads at 140 °C on a thick magnet can fall behind on a thinner one in the same housing.

Design windows and margins worth holding

• Hold on the order of 20–30 K between calculated magnet temperature and the catalogue maximum, and treat that maximum as the vendor's ceiling under a stated permeance and irreversible-loss limit, not a guarantee on every load line.
• Check the knee at maximum magnet temperature and at the worst demagnetizing field — short circuit, reverse field from an adjacent coil, or assembly magnetization — not at rated current.
• For NdFeB, choose the coating with the grade. Hot, condensing, or salt-laden service can make corrosion the reliability limit.
• For SmCo, budget for brittleness in handling and grinding. Yield loss often moves program cost more than alloy price.

Failure modes that invalidate the ranking

Unanalyzed load line. A grade safe at 150 °C on a thick, high-permeance magnet can fall below the knee at 120 °C in a thin section. Ratings do not transfer across geometries.

Self-heating. Eddy currents in a high-speed rotor magnet or its can can put the magnet well above an ambient-based estimate. Recompute at the true temperature; the choice often moves from NdFeB toward SmCo. Illustration: ambient data at 110 °C, can losses near 165 °C, and the SH window is gone.

Cold ferrite. Intrinsic coercivity of ferrite falls as temperature falls. A circuit sized only for hot running can lose induction on a cold start, so the low-temperature band does not automatically belong to ferrite.

Non-magnetic cost. On small parts, coating, grinding, handling and inspection can exceed alloy cost. Alloy price then misses program cost.

Excursions. A stall, brake or fault that crosses the knee sets a new remanence. Repeating that same peak does not subtract the loss again. Further loss needs a worse peak, aging, or corrosion. Specify peak field, peak temperature and duration.

Supply. Heavy-rare-earth price moves independently of the magnetics. A modest mass increase that avoids dysprosium and terbium is often worth taking for that reason as well.

How AIC Engineering supports the temperature-based selection

AIC Engineering (骏材磁应用) selects from the circuit inward: magnetic-circuit and magnet-application structure design, permanent-magnet transmission systems, and special motor magnet assemblies, including multi-pole rings, radial rings, Halbach arrays and linear-motor magnet sets. Temperature-sensitive work pairs grade and diffusion-route choice with permanent-magnet quality inspection. Rapid prototyping is available in 3–7 days, so thermal and load-line assumptions can be checked on hardware before a program commitment. Global supply and regional delivery support keeps a selected grade consistent where site qualifications differ.

Near a crossover, the geometry check needs magnet temperature, worst-case demagnetizing field and envelope.

Next steps

1. Measure or simulate magnet temperature under worst-case duty, not ambient.
2. Compute the load line and the knee at that temperature for the actual thickness and permeance. If the candidate is ferrite, include a cold check.
3. Compare cost per usable energy for the two or three candidates that survive, then confirm the choice on a prototype.
4. Contact AIC Engineering via https://www.aicmagnetics.com for a magnetic-circuit review and rapid prototyping against that temperature, load line and envelope.

References

1. Parker, R. J. Advances in Permanent Magnetism. Wiley,
2. Load lines, permeance coefficients and the demagnetization knee.
3. Coey, J. M. D. Magnetism and Magnetic Materials. Cambridge University Press,
4. Temperature dependence of anisotropy and coercivity.
5. Sugimoto, S. "Current status and recent topics of rare-earth permanent magnets." Journal of Physics D: Applied Physics 44, 064001 (2011).
6. Nakamura, H. "The current and future status of rare earth permanent magnets." Scripta Materialia 154 (2018). Grain-boundary diffusion and heavy-rare-earth reduction.
7. Fidler, J., and Schrefl, T. "Overview of Nd–Fe–B magnets and coercivity." Journal of Applied Physics 79 (1996).
8. European Commission, Joint Research Centre. Raw Materials Scoreboard (series). Rare-earth supply risk.