Every magnet has at least two temperature limits, and they are not the same number. The Curie temperature is the point at which the material loses its ferromagnetism altogether. The maximum operating temperature is the point at which the magnet starts losing flux fast enough to stop being useful. In sintered NdFeB the first sits somewhere between 310 and 380 degrees Celsius, while the second ranges from about 80 degrees for a base grade to 230 degrees for the highest classes. The gap between those numbers is where most temperature-related failures live.
What the Curie temperature actually does
Heat agitates the magnetic domains in a material. Up to a point the agitation is reversible: cool the magnet and the domains settle back. At the Curie temperature the thermal energy exceeds the exchange interaction that keeps the domains aligned in the first place, and the material stops being ferromagnetic. Cool it back down and it does not simply recover as the same magnet; it can be re-magnetized, but the metallurgy has been through a structural transition and the result is a different part. Crossing the Curie temperature is a one-way door.
What the maximum operating temperature actually does
Well below Curie, a magnet loses flux reversibly with temperature, described by the remanence temperature coefficient, roughly minus 0.11 to minus 0.12 percent per degree Celsius for standard NdFeB. It also loses coercivity, described by a separate coefficient. Whether that loss is acceptable depends on the working point: a magnet sitting high on its demagnetization curve loses flux that largely returns on cooling, while one sitting low can be pushed past the knee, at which point the loss becomes permanent at a temperature far below Curie. The maximum operating temperature is the point where the reversible loss, plus the risk of irreversible loss, makes the magnet unsuitable. It is a property of the material and the circuit together, not of the material alone.
Why the same grade has different operating limits
This is the part that surprises buyers. An N42 disc working against a thick steel plate might be rated to 80 degrees. The same N42 in a long, thin geometry that keeps its working point high can be rated considerably warmer. The grade did not change; the circuit did. That is why responsible suppliers quote an operating temperature with a stated working point, or with a permeance coefficient, and why a bare number on a quotation deserves a follow-up question.
Choosing the class
Work out the hottest spot the magnet will see, not the ambient temperature and not the average. A motor magnet sitting next to windings sees the winding temperature, and a magnet in an enclosed cavity sees the cavity's steady-state rise. Then choose the class that covers that number with margin: M for around 100 degrees, H for around 120, SH for around 150, UH for around 180, EH for around 200, AH for around 230. Where nothing survives the duty, the material itself has to change, and SmCo with Curie points far above NdFeB's becomes the answer.
Practical checks
If you are unsure how hot your magnet actually runs, measure it: a thermocouple on the magnet's back iron during the worst duty cycle tells you more than any calculation. If you are comparing two suppliers, ask both to state the operating temperature with the working point they assumed. And if your application sees a one-off over-temperature event, ask specifically about irreversible loss at that temperature, because a magnet that recovers is a very different proposition from one that does not.
Frequently asked questions
If the Curie temperature is 350 degrees, why does the datasheet say 80?Because at 80 degrees the flux loss becomes large enough to matter for most designs, and past that point the magnet risks being pushed below the knee of its curve. The Curie temperature describes the material; the operating temperature describes the magnet in your circuit.
Is a magnet damaged permanently every time it gets warm?No. Below the knee, losses are reversible and the magnet recovers on cooling. The damage comes from crossing the knee, or from reaching Curie.
Does a higher class mean lower strength?Usually slightly, at room temperature, because the techniques that raise coercivity tend to cost a little remanence. At temperature, the higher class delivers more.
What about ferrite and SmCo?Ferrite has a much higher Curie temperature and a higher operating ceiling than base NdFeB, at the cost of far less field. SmCo goes higher still, with Curie points well above NdFeB's.
Not sure which class your application needs? Send us the duty cycle and the measured temperatures and we will match the grade.