How Can Magnet Material Selection Reduce Motor Demagnetization Risk?

Reducing motor demagnetization risk depends on coercivity, at the temperature where that grade is weakest, still exceeding the reverse field at the highest current that occurs there.

Irreversible demagnetization in permanent magnet motors is a combined-stress problem. For sintered NdFeB and SmCo it is usually high magnet temperature, peak or fault current, and the weakest local region of the magnet, such as a trailing edge or a corner. For hard ferrite the coercivity weak point is low temperature, again with high current. Room-temperature datasheet values do not represent either case. The useful question is not "which magnet is strongest?" It is which material and grade keep enough coercivity margin across the real duty cycle, without paying for margin the motor will never use.

What this means for your design: specify the worst-case operating point first, then choose the material. Picking a grade from a catalog and checking afterwards often ends in a field failure or in a heavy-rare-earth grade the duty cycle did not require.

Magnet material families differ mainly in how their coercivity behaves with temperature, and that difference decides demagnetization risk more than remanence does.

The table compares material routes on the axes that matter for demagnetization. Where a temperature coefficient is given, it is a commonly cited order of magnitude for the family, not a guaranteed grade value. Actual coefficients vary by grade, temperature interval, and supplier. Verify them on the specific datasheet and on measured demagnetization curves before a grade is frozen.

What this means for procurement: a high-coercivity NdFeB grade is not automatically the safest purchase. Ferrite is weakest at the cold end of the range. SmCo is considered when thermal coercivity stability matters more than flux density. The failure condition in the specification decides the family.

The permanent magnet operating point calculation shows how geometry and material together set demagnetization margin.

A first-order magnetic circuit screen can rank options. It is not a sign-off model. Assume a magnet of length $l_m$ and area $A_m$, an air gap $g$ of area $A_g$, a linear recoil line with relative recoil permeability ${\mu }_{rec}$, and negligible iron reluctance and leakage. Those assumptions fit a simple surface-magnet screen better than an interior-magnet rotor.

With no armature MMF, Ampère's law ($H_m{l}_m+H_{g}g=0$) and flux continuity without leakage ($B_m{A}_m=B_g{A}_g$) give the no-load point:

$P_c$ is a dimensionless geometric permeance coefficient. The load-line slope is $|B_m/H_m|={\mu }_0{P}_c$. A $P_c$ of 4 is the same number often called a permeance coefficient of 4 in Gaussian datasheet practice. It is not $|B/H|$ in tesla and A/m.

Take $H$ as positive along the magnetization direction, so no-load $H_m$ is negative. Datasheet $H_{cj}$ and knee field $H_k$ are positive magnitudes.

A demagnetizing armature MMF $F_a$ (positive when it opposes magnetization), referred to the magnet length as $H_a=F_a/l_m$, moves the point along the recoil line:

The added demagnetizing field is $H_a/(1+{\mu }_{rec}/P_c)$, not the full $H_a$. Subtracting $F_a/l_m$ from the no-load field overstates the average reverse field. If a local field is already taken from finite-element analysis, compare that local magnitude with $H_k(T)$ directly. Do not subtract it again from the no-load load line.

The closed-form point is valid only while it stays on the linear recoil line, above the knee. The screen is

$T_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}}$ is the temperature at which that grade's knee is lowest, and the current in $H_a$ must be a current that can actually occur at that temperature. $m$ is a fractional margin agreed for the project. No universal value of $m$ is used here. It has to cover tolerance on $B_r$ and $H_{cj}$, thermal-model error, and assembly variation.

To first order, near a reference temperature $T_0$:

$\beta$ is the coefficient in %/°C. It is negative for NdFeB and SmCo and positive for hard-ferrite $H_{cj}$. $H_k$ lies below $H_{cj}$ by the squareness of that grade. Do not assume a squareness ratio.

What this means for your design: a thicker magnet or a smaller gap raises $P_c$ and moves the average operating point away from the knee. Added thickness costs mass and magnet volume. It can cost less than a heavy-rare-earth grade, but only a priced comparison at the same margin can show that.

Illustrative / hypothetical scenario. These figures are not a measured grade and not a recommended design point. Take $B_r=1.2$ T, ${\mu }_{rec}=1.05$, $P_c=4$, and ${\mu }_0=4\pi \times {10}^{-7}$ H/m. Then $B_m\approx 0.95$ T and no-load $|H_m|\approx 189$ kA/m. Take $H_a=F_a/l_m=300$ kA/m. Here $B_r/({\mu }_0{P}_c)\approx 239$ kA/m, so

Adding 300 kA/m directly to the no-load 189 kA/m gives 489 kA/m. That sum is not the load-line result.

A hypothetical magnet with $H_{cj}=1600$ kA/m at 20 °C and $\beta =-0.55$ %/°C extrapolates linearly to

$USD 1600\times\left[1 + \frac{-0.55}{100}\times 130\right] \approx 456\ \mathrm{kA/m}$$

at 150 °C. Average $|H_m|$ then sits under that extrapolated $H_{cj}$ by less than 30 kA/m. That is not a pass. $H_k$ is lower than $H_{cj}$, a constant $\beta$ is not a datasheet curve, and edge fields exceed the average. The same three levers remain: a higher-$H_{cj}$ grade, a higher $P_c$, or a lower peak current at the critical temperature. This arithmetic alone does not justify any of the three.

Use the formulas to discard options that already fail the average-field screen. Sign off with measured demagnetization curves at temperature and with finite-element analysis of local fields.

The recommended design window keeps worst-case internal field below the knee at the critical temperature, with an explicit margin.

The following practices are an engineering window, not a checklist taken from one standard:

• Define the worst case explicitly. For NdFeB and SmCo use maximum magnet temperature, which is not the winding temperature, with the highest current that can occur at that temperature. That current may be a three-phase short circuit or a field-weakening transient, not rated current. For ferrite, pair minimum magnet temperature with the highest current that can occur there, including cold start.
• **Use knee field, not $H_{cj}$, as the limit. Use measured demagnetization curves at the critical temperature where they exist. Squareness varies by grade and supplier.
• Hold a margin agreed with your reliability assessment.** No universal percentage applies. The margin should cover tolerance on $B_r$ and $H_{cj}$, thermal-model uncertainty, and assembly variation.
• Check local fields, not only the average load line. Pay particular attention to trailing pole edges, flux-barrier corners in IPM rotors, and thin magnet sections.
• Check cold conditions for ferrite. Cold-start current is the coercivity case for ferrite, even when the hot flux case looks acceptable.
• Trade geometry against grade on cost per unit of margin. Compare magnet thickness, segmentation, and rotor topology with a grade change. A modest increase in $P_c$ can remove the need for the highest heavy-rare-earth content. Confirm that on the specific bill of materials.

What this means for cost: do not assume either "always buy the highest $H_{cj}$" or "always add thickness." Price the margin both ways.

Irreversible demagnetization in permanent magnet motors occurs when the design operates outside the boundaries assumed during selection.

Treat the material choice as unproven, or the screen as invalid, in these situations:

• Magnet temperature outside the value used in selection. Blocked cooling, or a thermal model that used winding temperature in place of magnet temperature, can collapse NdFeB margin quickly because the coercivity coefficient is steep. Ignoring a cold soak is the matching error for ferrite.
• Fault current that was not bounded. Inverter faults or short circuits can produce MMF well above the screen current. Cold windings can draw more fault current than hot windings; for NdFeB the rise in $H_{cj}$ when cold often offsets this, but that offset should be checked, not assumed. If the fault field cannot be bounded, material selection alone does not close the risk. Current limiting, protective control, or a mechanical redesign has to bound it.
• Air gap larger than assumed. A larger gap lowers $P_c$ and moves the average operating point toward the knee. Coating build and assembly stack-up count as part of $g$ if they were omitted from the model.
• IPM or other high-leakage geometry judged only by the gap formula. The no-leakage, no-iron-reluctance screen does not bound local reverse field at flux barriers, bridges, or magnet corners.
• An operating point already below the knee. The linear recoil formulas do not describe the flux left after an excursion past $H_k$. A no-load voltage check after the event is the practical witness that the loss was irreversible.
• Eddy-current heating in the magnets. High-frequency harmonics or PWM ripple can heat a solid magnet above the temperature used for the grade. Segmentation may be required; that follows from the loss calculation, not from the load line alone.
• Corrosion or coating damage on sintered NdFeB. Loss of magnetic section reduces flux and can release particles into the machine. Specify the coating and the environment with the grade.
• Low-temperature ferrite operation. Ferrite coercivity falls in the cold. A current that is safe at room temperature can be unsafe at cold start.
• Grade substitution during procurement. A grade called similar, with a different $H_{cj}$, squareness, or temperature coefficient, invalidates the analysis. Lock the approved curve, not only a catalog name.
• **Linear $\beta$ used as if it were a measured curve. Over a span of more than about 100 °C a constant percent-per-degree coefficient can mis-rank grades. The extrapolation above is a screen, not a material property.
• Torque that is a safety function.** If flux loss can disable braking, actuation, or propulsion, the margin belongs in that machine's safety case. Which standard applies depends on the product; the magnet grade does not decide it.

AIC Engineering supports this design window through magnetic circuit analysis, rotor magnet assemblies, prototyping, and magnet inspection.

The AIC Engineering team supports motor developers at the stages where demagnetization margin is set:

• Magnetic circuit and structure design. Load-line screening and finite-element evaluation of candidate materials, magnet thickness, and segmentation against the worst-case temperature and current defined for the program.
• Special motor permanent magnet assemblies. Multipole rings, radial rings, Halbach arrays, and linear motor magnet assemblies, where geometry changes the local reverse field the grade must survive.
• Rapid prototyping in 3–7 days. Samples so a geometry-versus-grade trade can be checked on the bench before the grade is locked.
• Permanent magnet quality inspection. Checks that delivered parts match the magnetic properties assumed in the analysis. Inspection scope and test conditions are agreed per project. Incoming magnet inspection does not replace the machine-level demagnetization review.

Next steps to reduce motor demagnetization risk in your design

1. Document magnet temperature and peak or fault current for every mode that can demagnetize the grade, including hot overload for NdFeB and SmCo and cold start for ferrite.
2. Run the permeance coefficient screen above for two or three candidate materials and geometries. Use $H_a=F_a/l_m$ in the load-line formula. Do not add armature MMF directly to the no-load field. Then confirm local fields with finite-element analysis and with measured $H_k(T)$.
3. Require temperature-dependent demagnetization curves, not room-temperature datasheets only, and lock the approved grade and curve in the procurement specification.
4. Contact AIC Engineering for custom magnetic circuit design and rapid prototyping. Visit https://www.aicmagnetics.com to request a consultation. The team can review your operating point, demagnetization margin, and magnet assembly options against that duty cycle.

References

1. Hendershot, J. R., & Miller, T. J. E. Design of Brushless Permanent-Magnet Machines. Motor Design Books,
2. Gieras, J. F. Permanent Magnet Motor Technology: Design and Applications, 3rd ed. CRC Press,
3. Coey, J. M. D. Magnetism and Magnetic Materials. Cambridge University Press,
4. Sagawa, M., Fujimura, S., Togawa, N., Yamamoto, H., & Matsuura, Y. "New material for permanent magnets on a base of Nd and Fe." Journal of Applied Physics, 55(6),
5. Hirota, K., Nakamura, H., Minowa, T., & Honshima, M. "Coercivity enhancement by the grain boundary diffusion process to Nd–Fe–B sintered magnets." IEEE Transactions on Magnetics, 42(10),
6. IEC 60404-5. Magnetic materials – Part 5: Permanent magnet (magnetically hard) materials – Methods of measurement of magnetic properties. International Electrotechnical Commission.
7. IEC 60404-8-1. Magnetic materials – Part 8-1: Specifications for individual materials – Magnetically hard materials. International Electrotechnical Commission.
8. MMPA Standard No. 0100-00. Standard Specifications for Permanent Magnet Materials. Magnetic Materials Producers Association.