Selecting ultra-thin NdFeB magnets for consumer electronics like smartphones and smartwatches presents a unique engineering challenge. As the thickness-to-diameter ratio drops, severe self-demagnetization occurs, rendering high remanence grades ineffective and making intrinsic…
How to Choose Micro Magnets for Consumer Electronics? Size and Demagnetization Limits of Ultra-Thin NdFeB
Thickness-to-diameter ratio (t/D) and the coercivity left at the hottest operating point decide whether an ultra-thin NdFeB magnet works in a phone, earbud or smartwatch. As t/D falls toward 0.1, the self-demagnetizing field reaches about 0.9 Br, and it rises further as the disc gets thinner. Selection order therefore runs from geometry, to the resulting internal field, to a grade whose intrinsic coercivity (HcJ) still clears that field at maximum temperature with margin. Remanence (Br) comes last. Thickness tolerance matters in the same way. On a 3 mm diameter, 0.3 mm thick disc, with a Hall sensor 0.5 mm from the face, ±0.05 mm thickness moves the on-axis field by roughly ±15%. A low reading is often a dimensional reading.
The rest of this article sets out the model behind those statements, the design window it produces, and the cases where an ultra-thin sintered magnet is the wrong choice.
The Thin-Magnet Decision in Phones and Wearables
Smartphones, tablets, laptops, smartwatches and wireless earbuds all use micro magnets: Hall-sensor position and lid detection, charging-case retention, haptic and acoustic parts, and accessory alignment. Hall-sensing disc magnets are among the most common of these parts. Flexible SmFeN magnets also appear in curved mounting positions and custom die-cut assemblies.
The design engineer usually controls three things: the footprint (set by the industrial design), the thickness (set by the stack-up), and the grade. The procurement engineer needs to know whether the drawing can be produced, how much the field varies from part to part, and whether a cheaper grade would survive. Both questions come back to the same physics.
Options for Thin Magnetic Elements Compared
The real choice is between ways of building a thin magnetic element, so the table compares design routes. Grade lists come later.
| Design route | Self-demagnetization at low t/D | Field delivered per unit thickness | Temperature ceiling | Mechanical risk | Relative cost | Best fit | |||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Ultra-thin sintered NdFeB disc, standard-coercivity grade | High; internal field approaches Br | High; sintered (BH)max 35–52 MGOe | 80–200 °C depending on grade | Brittle; chipping at edges and in press-fit | ' in math mode at position 1:̲</td>
<td>Room-…" style="color:#cc0000"></td>
<td>Room-temperature parts with t/D comfortably above the limit</td>
</tr>
<tr>
<td>Ultra-thin sintered NdFeB disc, high-HcJ grade</td>
<td>Same geometry; larger coercivity margin</td>
<td>Same 35–52 MGOe band; Br set by the grade</td>
<td>80–200 °C depending on grade</td>
<td>Brittle; chipping at edges and in press-fit</td>
<td></span></div> |
Thin parts close to warm components or exposed to opposing fields | |||||||||||||||||||||||||||||
| Thin sintered NdFeB on a soft-steel back plate or cup | Reduced; effective t/D increases | Higher on the open face, near zero on the back | Set by the magnet grade, within 80–200 °C | Bonding joint is an added failure point | ' in math mode at position 1:̲ plus assembly<…" style="color:#cc0000">plus assembly</td>
<td>Single-sided sensing or retention where stack height is fixed</td>
</tr>
<tr>
<td>Thin SmCo disc</td>
<td>High; set by t/D, as for any thin disc</td>
<td>Set by the SmCo grade</td>
<td>High-temperature duty</td>
<td>Rigid sintered disc</td>
<td>—</td>
<td>Hot zones where SmCo is the temperature selection</td>
</tr>
<tr>
<td>Flexible / <a href="/en/glossary#bonded-ndfeb">bonded NdFeB</a> or ferrite sheet</td>
<td>Lower Br means a lower demagnetizing field</td>
<td>(BH)max 1.5–2.5 MGOe (ferrite) or up to 9 MGOe (NdFeB), against 35–52 MGOe for sintered NdFeB</td>
<td>80–120 °C, limited by the binder</td>
<td>Bendable, rollable, cuttable</td>
<td></td>
<td>Curved surfaces and large-area thin retention where field strength is secondary</td>
</tr>
</tbody>
</table>
</div>
<p>*Sintered NdFeB temperature span, flexible-magnet (BH)max, binder temperature, bendability and the marks follow AIC Engineering's published product specification data.*
The Demagnetization Model Behind Thin NdFeB DiscsTwo well-established relations cover most micro-magnet decisions. The first gives the field the magnet produces on itself. The second gives the field it delivers to the sensor. Self-demagnetizing field and permeance coefficientFor a disc magnetized through its thickness, the magnetic surface charge on the two faces sets up an internal field that opposes the magnetization. For a uniformly magnetized cylinder of diameter D and thickness t, superposing the fields of the two charged faces gives the demagnetizing factor on axis at the geometric centre: The last expression assumes recoil permeability close to 1, a reasonable first approximation for sintered NdFeB. Values computed from this relation:
For an ultra-thin disc, the demagnetizing field is almost the full remanence, expressed as a field. The face centres and the rim see higher local demagnetization than the body centre, so irreversible loss usually starts at the edges. Finite-element analysis, or the cylinder demagnetizing factors published by Chen, Brug and Goldfarb, will refine the numbers. The trend does not change. The relation also shows that a higher-Br grade does not help a thin magnet survive. Going from N42 toward N52 raises the demagnetizing field in proportion to Br, and higher-Br grades usually carry lower HcJ. Once t/D drops below about 0.2, coercivity is the property worth paying for. Field delivered to a Hall sensorOn the axis of a disc magnet, at distance z from its face, the standard expression for a rigid, uniformly magnetized cylinder of radius R is: The expression treats magnetization as uniform and equal to Br/μ0. It applies once the grade check shows the part stays above the knee; partial demagnetization lowers the delivered field. Worked example, computed from this formula: D = 3 mm (R = 1.5 mm), t = 0.3 mm, sensor at z = 0.5 mm, Br = 1.3 T. The result is B_z ≈ 0.077 Br ≈ 100 mT. At t = 0.25 mm the field falls to about 85 mT, and at t = 0.35 mm it rises to about 115 mT. A ±0.05 mm thickness tolerance therefore produces roughly ±15% field variation before Br scatter, mounting-gap variation and temperature are included. Two consequences follow for the sensor design. The Hall IC's switching thresholds or analog range must accept that spread. Alternatively, the parts must be sorted by measured flux, and that requirement should appear on the drawing and in the quotation. Worked demagnetization checkTake the same 3 × 0.3 mm disc (t/D = 0.1). The internal field is about 0.90 × 1.3 T ≈ 1.17 T, or roughly 930 kA/m (11.7 kOe). Because the knee of the J–H curve sits below HcJ, and because neighbouring magnets or assembly fixtures can add an opposing field, a practical guideline is to keep HcJ at maximum temperature at least about 20% above the computed internal field. Here that means about 1,120 kA/m (14 kOe) at the hot point. Published temperature coefficients of HcJ for sintered NdFeB are typically about −0.5 to −0.6 %/°C from a 20 °C reference, and they vary by grade, so the datasheet coefficient should be used. For a 60 °C hot point the rise is 40 °C, and room-temperature HcJ then needs to be about 1,400–1,470 kA/m (about 17.6–18.5 kOe). Under common grade nomenclature in MMPA 0100-00 and IEC 60404-8-1, N- and M-class minima, and the usual H-class minimum of 17 kOe, sit below that estimate. Specify H only when the datasheet HcJ clears the calculated value; SH, whose usual minimum is 20 kOe, covers this example. The thinner the part and the warmer its location, the further up the coercivity ladder the design has to go. Where the knee lies well below HcJ, take a larger allowance from the J–H curve. A soft-steel back plate mirrors the magnet. For an infinite, unsaturated plate, the open-face field matches a disc of thickness 2t. On axis at the plate, the internal field equals the centre field of that 2t disc: effective t/D = 0.2, about 11% below the bare t/D = 0.1 disc. Toward the open face the relief is smaller. Real cups, finite plates and saturation reduce it further. Grade selection still comes first. Recommended Design Window for Ultra-Thin NdFeBThese ranges come from the model above. Treat them as starting points to confirm by simulation and by prototype measurement.
Thickness tolerance belongs inside the window. A part designed at the lower edge of t/D will, at its minimum thickness, be both weaker and more exposed to demagnetization. Run the demagnetization check at minimum thickness and maximum temperature together, because worst cases add up in a stack-up. Where Ultra-Thin NdFeB Fails or Should Not Be UsedMost field failures of micro magnets come from a handful of mechanisms, and several appear only after assembly. Thermal excursions are the most common cause. Heat from a charging coil, a nearby processor, or adhesive curing can push the operating point past the knee, and the loss is irreversible. A disc that measured correctly at incoming inspection can arrive weak at final test. Map the worst thermal event in the build as well as the steady temperature in use. Opposing external fields add to the self-demagnetizing field. Phones and accessories pack alignment arrays, speakers, haptic actuators and sensor magnets into a small volume. Pairing fixtures and handling with other magnets on the line can also expose a part to a strong opposing field for a moment. If an adjacent magnet's field opposes the thin part, add that field to N·Br in the margin check. Surface condition counts for more as thickness drops. Grinding leaves a thin damaged layer with reduced coercivity, and in a sub-millimetre part that layer is a larger share of the volume. Nickel, epoxy or zinc plating adds thickness on every face, so drawings should state whether a dimension applies before or after coating. In wearables, sweat and cleaning agents attack coating defects, and a corroded edge loses material and starts cracks. Mechanical stress is a further risk. Sintered NdFeB is brittle, and thin discs crack under press-fit, snap-fit or uneven adhesive cure shrinkage. Use bonded retention with a defined bond-line thickness, or flexible magnet sheet where the part has to conform to a curve. Some applications need only light retention over a large or curved area. Where the binder-limited 80–120 °C ceiling is acceptable, flexible material is the simpler choice, and the lower cost usually decides it. How AIC Engineering Delivers Within This WindowThe AIC Engineering team applies the checks above in engineering and production. Magnetic-circuit design starts from the stack-up and sensor position. The team computes the operating point at minimum thickness and maximum temperature, assesses whether a back plate or a change in diameter would recover margin, and recommends material from NdFeB, SmCo or ferrite according to temperature, environment, cost and space. The grade cross-reference applies the same Br and HcJ rules to find the closest equivalents among MMPA, VAC, Shin-Etsu, NEOMAX, TDK and Arnold grades, so a design already qualified on another supplier's grade can be matched on those two properties. Hall-IC matching is handled as a single package. For two-pole Hall-sensing discs, magnetization angle deviation is held within ±3°. Every disc is measured automatically and sorted into 1°, 2° or 3° grades, which gives automated sensor lines consistent angle from batch to batch. Disc bodies are offered in NdFeB, SmCo or ferrite, ground to ±0.05 mm, with nickel, epoxy or zinc coating. The ±15% field spread in the worked example is the effect of that tolerance band on the stated geometry. Where the sensor needs a tighter field window, flux-based sorting should be agreed at the drawing stage. For inspection, sintered NdFeB is 100% flux-tested at the end of production. Dimensions and tolerances are verified on coordinate measuring machines, appearance is fully inspected, and process deviations are handled through nonconforming-part isolation and SPC records. Manufacturing is certified to ISO 9001 and IATF 16949. Rapid prototyping in 3–7 days lets an engineer test two or three thickness and grade combinations in the real housing before committing to tooling. Standard MOQ is 50 pieces, with lower quantities available for samples, and production lead time is typically 2–6 weeks depending on design complexity and material supply. Engineering support is free: send a grade, a datasheet or the application conditions, and the team will confirm grade, coating and magnetic circuit. Next Steps for Your Micro-Magnet Design
References
|