Magnets & Temperature

The number that surprises engineers is β: coercivity falls four to six times faster than flux. A magnet at 120 °C hasn't just lost ~11% of its field — it has lost roughly half its resistance to demagnetization. That's why "how hot" and "how hard is the magnet being pushed" must always be asked together.

Within its safe envelope, an NdFeB magnet's flux simply tracks temperature — down as it heats, fully back as it cools. Representative flux and force levels versus a 20 °C baseline (α = −0.12 %/°C; force scales as B², per the pull force guide):

Two design consequences: performance targets must be met at the hottest operating point, not on the bench; and precision devices (sensors, instruments) see this drift as signal error — which is why thermally stable applications reach for SmCo or alnico (coefficients 3–6× better; see the material comparison) or compensate electronically.

Push past the safe envelope and losses stop coming back on their own:

• Irreversible (recoverable) loss: at high temperature, the demagnetization curve's "knee" rises into the magnet's operating region and portions of the magnet flip polarity locally. The flux is gone after cooling — but the material is undamaged, and full re-magnetization restores it. This is the classic over-temperature event.
• Permanent (structural) loss: prolonged very high temperature or corrosion changes the material itself — oxidation, metallurgical changes near Tc. No pulse brings it back.
• Long-term aging: a properly specified magnet held within its envelope loses only fractions of a percent per decade — "magnets wearing out" is almost always one of the events above in disguise, as the demagnetization guide details.

The rating on the grades chart is a material capability measured under favorable conditions. What your specific magnet tolerates depends on its permeance coefficient (Pc) — set by geometry and the magnetic circuit:

• Thin, wide magnets in open air have low Pc — they self-demagnetize hard, sit near the knee, and can suffer irreversible loss below their class rating. A 20 mm × 1 mm disc rated "80 °C" may lose flux at 60 °C in free air.
• Long magnets and closed circuits have high Pc — the same material in a pot assembly or against a steel return path can safely exceed the headline rating.
• The rule: for any application above ~60 °C, evaluate temperature and geometry together. Give your supplier the drawing, the circuit (open air / steel backing / closed), and the real thermal profile, and ask for the load-line check — it's a standard calculation, and it's free compared to field failures. (This is the "operating conditions" section of the RFQ guide earning its keep.)

• Establish the real thermal profile: continuous maximum, short excursions (reflow, sterilization, cure ovens), and self-heating in motors (rotor temperatures run well above ambient).
• Pick the class with margin after the load-line check — then take the strongest N-number available in that class (availability matrix in the grades chart).
• Don't buy letters you don't need: temperature classes are the most expensive characters in the magnet business — the premium structure is laid out in Magnet Pricing Explained.
• Above ~180–200 °C, compare materials, not just classes: SmCo delivers more flux than EH/AH NdFeB at the top of the range with none of the heavy-rare-earth premium volatility, and works to 350 °C.

• NdFeB gets slightly stronger cold — the coefficients run in your favor down to about −40 °C with no special precautions for most grades. At deep cryogenic temperatures, spin-reorientation physics (below ~135 K) reduces useful magnetization — cryogenic systems typically use SmCo instead.
• Ferrite is the cold-weather hazard: its coercivity falls as temperature drops, so ferrite magnets under demagnetizing load (motors, sensors with opposing fields) are checked for demag at −40 °C, not at max temperature — the inverse of NdFeB intuition, flagged in the material comparison.
• The cold-side risks for NdFeB assemblies are mechanical: CTE mismatch working adhesive joints and embrittled polymers — bonding-side issues covered in the bonding guide.

• Cycling within the envelope is benign magnetically — flux rides up and down the reversible line. The wear happens in the joints and coatings (CTE mismatch), not the magnetism.
• Thermal stabilization (pre-aging): precision applications deliberately expose magnetized parts to a temperature slightly above the worst-case service condition before calibration. The small irreversible knock-down happens once, on the bench — and the magnet is then stable for every excursion below it. Standard practice for sensors, instruments, and calibrated assemblies; specify it on the drawing if your device is calibrated after magnet installation.
• Process excursions count: reflow soldering (~250 °C peak) exceeds every NdFeB class — parts must be magnetized after reflow, use SmCo, or be mounted downstream. Autoclave cycles (121–134 °C) require SH-class or better with a load-line check.