The ability of a permanent magnet to support an external magnetic field is due to crystal anisotropy within the magnetic material that “locks” small magnetic domains in place. Once the initial magnetization is established, these positions remain the same until a force exceeding the locked magnetic domain is applied, and the energy required to interfere with the magnetic field produced by the permanent magnet varies for each material. Permanent magnets can generate extremely high coercivity (Hcj), maintaining domain alignment in the presence of high external magnetic fields.

Stability can be described as the repetitive magnetic properties of a material under specified conditions over the life of the magnet. Factors that affect magnet stability include time, temperature, changes in reluctance, adverse magnetic fields, radiation, shock, stress, and vibration.

Time has little effect on modern permanent magnets, which studies have shown change immediately after magnetization. These changes, known as “magnetic creep,” occur when less stable magnetic domains are affected by thermal or magnetic energy fluctuations, even in thermally stable environments. This variation decreases as the number of unstable regions decreases.

Rare earth magnets are unlikely to experience this effect because of their extremely high coercivity. A comparative study of longer time versus magnetic flux shows that newly magnetized permanent magnets lose a small amount of magnetic flux over time. For more than 100,000 hours, the loss of samarium cobalt material is basically zero, while the loss of low permeability Alnico material is less than 3%.

Temperature effects fall into three categories: reversible losses, irreversible but recoverable losses, and irreversible and irrecoverable losses.

Reversible Losses: These are the losses that recover when the magnet returns to its original temperature, permanent magnet stabilization cannot remove reversible losses. Reversible losses are described by the reversible temperature coefficient (Tc), as shown in the table below. Tc is expressed as a percentage per degree Celsius, these numbers vary by the specific grade of each material, but are representative of the material class as a whole. This is because the temperature coefficients of Br and Hcj are significantly different, so the demagnetization curve will have an “inflection point” at high temperature.

Irreversible but recoverable losses: These losses are defined as the partial demagnetization of a magnet due to exposure to high or low temperatures, these losses can only be recovered by re-magnetization, the magnetism cannot recover when the temperature returns to its original value. These losses occur when the operating point of the magnet is below the inflection point of the demagnetization curve. An effective permanent magnet design should have a magnetic circuit in which the magnet operates with a permeability higher than the inflection point of the demagnetization curve at the expected high temperature, which will prevent performance changes at high temperature.

Irreversible Irrecoverable Loss: Magnets exposed to extremely high temperatures undergo metallurgical changes that cannot be recovered by remagnetization. The following table shows the critical temperature for various materials, where: Tcurie is the Curie temperature at which the fundamental magnetic moment is randomized and the material is demagnetized; Tmax is the maximum practical operating temperature of the primary material in the general category.

The magnets are made temperature stable by partially demagnetizing the magnets by exposing them to high temperatures in a controlled manner. The slight decrease in flux density improves the stability of the magnet, since the less oriented domains are the first to lose their orientation. Such stable magnets will exhibit constant magnetic flux when exposed to equal or lower temperatures. Additionally, a stable batch of magnets will exhibit lower flux variation when compared to each other, since the top of the bell curve with normal variation characteristics will be closer to the batch’s flux value.