Subtitle: Why the same NdFeB magnet lasts ten years in one place and falls off in three months in another — bonding and assembly are a double-edged sword
Neodymium iron boron (NdFeB) magnets are, in themselves, a highly stable material — as long as they aren't corroded, aren't heated beyond their operating temperature, and aren't mechanically shocked, their magnetic properties can hold for decades without decay. But in the real world, magnet failures are rarely about the magnet itself.
On the front line we've seen too many such cases: a robot joint motor whose magnets delaminated entirely from the steel back-iron after three months, leaving the robot dead on the floor; a magnetic-drive pump whose inner-rotor magnets cracked after six months of operation, contaminating the medium with magnetic debris; a motor rotor working continuously in high-temperature conditions, silently losing 30% of its magnetization, with the customer not noticing until the torque fell short.
There's an empirical number in the industry: in cases of early magnet failure (within warranty), over 80% are not magnet-material problems but issues of adhesive process and assembly control. A magnet is a passive component — how it's bonded, how it's assembled, and at what temperature it works determines whether it lasts ten years or falls off in three months.
This article doesn't discuss magnet principles; it only talks about failure — breaking down the most common categories of quality failures, then explaining exactly what a professional factory does in bonding and assembly to keep those failures from leaving the plant.
Debonding is the most common failure mode of magnet assemblies. It manifests as the magnet falling off the steel back-iron or housing, or delamination between multiple magnet pieces.
There are five root causes:
First, wrong adhesive selection. This is the highest-frequency error. NdFeB magnets may work at 120–200°C, but many factories, for cost or convenience, use room-temperature structural adhesives (such as certain acrylate quick-bond glues rated at 80–100°C, which actually soften at a sustained 40°C). We've seen a case where a customer's robot-joint magnets delaminated entirely at just 40°C — traced back to a quick-bond glue with poor compatibility with the nickel-plated surface, whose bond completely lost strength under prolonged warmth.
Second, inadequate surface preparation. If the nickel-plated surface has oil, oxide layer, or passivation film, the adhesive simply won't stick firmly. A professional factory must degrease, sandblast, or plasma-treat the surface before gluing, and sometimes apply a primer to improve adhesion. Skip this step and the adhesive is just "floating" on the surface.
Third, insufficient curing. Epoxy needs sufficient temperature and time to fully cross-link. Many factories, to save time, cure at room temperature or with short low-temperature cycles, so the adhesive layer only cures 60–70%, and its strength and temperature resistance fall short of design values.
Fourth, out-of-control bond-line thickness. Too-thin adhesive leaves voids and bubbles; too-thick adhesive generates internal stress and curing-shrinkage cracks. Professional factories use dispensing shims or fixtures to control the bond line within 0.05–0.20 mm.
Fifth, thermal-cycle fatigue. Equipment start-stop and ambient temperature fluctuations make the adhesive layer expand and contract repeatedly. If the adhesive's glass-transition temperature (Tg) is too low or its toughness insufficient, micro-cracks appear after a few hundred cycles and eventually debonding occurs.
Heat demagnetization is more hidden than debonding because it usually comes with no obvious physical damage — the magnetic performance just silently drops.
There are four root causes:
First, exceeding the grade's operating temperature. This is the most fundamental. NdFeB is graded by temperature: N (80°C), M (100°C), H (120°C), SH (150°C), UH (180°C), EH (200°C), AH (230°C). If the actual working temperature is 160°C but the SH grade (ceiling 150°C) was used, the magnet crosses the "knee point" of the demagnetization curve and suffers irreversible demagnetization. Wrong grade selection is the most common cause of this failure.
Second, local overheating from eddy currents. Magnets working in alternating magnetic fields (such as high-speed motor rotors, linear-motor magnetic tracks), if not laminated, develop eddy currents inside the magnet due to skin effect, with local temperatures far above ambient. Even if the overall temperature is below the grade ceiling, internal hot spots may already be demagnetized.
Third, operating-point shift from assembly gaps. Excessive assembly gap or uneven air gap shifts the magnet's operating point into the dangerous zone of the demagnetization curve. It looks like temperature isn't the issue, but the operating point is already near the knee.
Fourth, superimposed reverse magnetic field. Wrong magnetization direction, reversed polarity of adjacent magnets, or external strong-field interference all superimpose a reverse field on the magnet, directly pushing local regions past the knee point.
The scary part of heat demagnetization: it's irreversible. Once a magnet loses 20% of its magnetization, it won't recover on its own — only full replacement works.
NdFeB fears two things most: corrosion and high temperature. The coating is its only barrier against corrosion.
Coating delamination manifests as nickel plating peeling, magnet edges powdering, rust spots appearing on the surface. The root causes concentrate on three points: poor pre-plating treatment (incomplete cleaning, weak bond between coating and substrate), insufficient coating thickness (Ni-Cu-Ni three layers total below 15 μm), and plating micro-cracks from impacts (cracks become corrosion channels).
A professional factory's Ni-Cu-Ni three-layer coating is usually controlled at 20–35 μm, with salt-spray testing on every batch. Thicker isn't always better — too thick means high internal stress and easy peeling; the balance between thickness and adhesion must be found.
NdFeB's flexural strength is only about 250 MPa — one-tenth of steel — essentially a "ceramic-state" metal. It doesn't fear compression but fears tension, impact, and stress concentration.
Common causes of cracking: excessive interference during press-fit, with stress concentration chipping the magnet edges; thermal shock (rapid cooling and heating, differences in expansion coefficients between coating and substrate); assembly impacts (striking the magnet directly with metal tools); uneven force on individual arc segments during multi-segment assembly.
The professional factory's approach: use FEA simulation to calculate interference and stress distribution before press-fit; use soft tooling and buffer spacers during press-fit; keep the assembly environment at constant temperature to avoid thermal shock; ban dropping and striking throughout.
This category shows no physical damage but "insufficient magnetic force" or "uneven flux density."
Manifestations: motor torque falls short of design (insufficient performance); sensor readings drift (uneven magnetization); local weak magnetization (incomplete magnetization). Root causes: grade confusion (using N42 where N52 was specified, a notch lower in remanence), insufficient magnetization field strength (underpowered magnetizer, magnet not saturated), wrong magnetization direction or pole count (fixture positioning deviation), pole-arc coefficient deviation (poor control of inter-pole transition zones in multi-pole magnetization).
This failure can't be seen by eye — it must be intercepted at the factory by scanning each piece's air-gap flux density and waveform with a flux mapper.
Keeping the above failures from leaving the plant relies on a systematic adhesive process. A professional factory does at least the following in bonding:
Adhesive selection matrix. Not "one glue for everything" but a selection table by temperature, medium, and mechanical load. High-temperature conditions use structural epoxies rated above 200°C (such as Henkel, 3M series); corrosive conditions use corrosion-resistant grades; applications needing quick positioning use thixotropic types; those needing high toughness use modified acrylates. Every adhesive is certified for DSC (glass-transition temperature), TGA (thermal decomposition temperature), and lap-shear strength.
Surface preparation. Degreasing (ultrasonic cleaning), sandblasting or plasma activation, primer when necessary. Plated and bare-magnet surfaces use different preparation processes — they can't be mixed.
Curing curve. Standard curing curve of stepwise heating, holding, and slow cooling — not "air-dry at room temperature for a day." Critical parts are oven-cured with temperature control, recording each furnace's temperature curve.
Bond-line thickness control. Dispensing shims, screen printing, or fixture positioning keep the bond line within the design range. Both over- and under-dispensing are flagged and intercepted.
Batch certification. Every batch of adhesive gets incoming sample lap-shear and heat-resistance verification; non-conforming batches don't go into production.
Assembly is where magnets are "installed correctly, installed securely, without damage." A professional factory's control points include:
Precision tooling. Integrated positioning, clamping, and pressure-holding fixtures ensure magnet-to-back-iron concentricity, parallelism, and gap consistency. The dispersion of manual placement is far higher than tooling.
Interference control. Press-fit interference is calibrated through FEA simulation plus measurement — not "press it in by feel." Too much interference causes cracking; too little causes loosening; it must land within the design window.
Force/torque curve monitoring. The press-fit process records force-displacement curves; abnormal curves (force spikes, insufficient displacement) trigger automatic alarms and interception.
Clean assembly environment. Constant temperature and humidity, dust-free, to prevent dust entering the adhesive layer and temperature-humidity fluctuations affecting curing.
Anti-impact system. Dedicated anti-static trays, soft tooling, no-drop no-strike operating rules. Magnets go from intake to finished product without touching the floor or bare hands.
No matter how good the process, inspection is the backstop. A professional factory's pre-shipment inspection of magnet assemblies includes at least:
Pull-out test. Destructive pull testing of the bond interface on samples to verify bond-line strength meets spec.
Thermal cycling. -40°C to +150°C (or actual condition ceiling) for 500–1000 cycles, verifying stability of adhesive and magnet under temperature fluctuation.
Salt-spray test. Neutral salt spray for 1000 hours, verifying coating corrosion resistance.
Flux distribution measurement. Scanning each piece's air-gap flux density and waveform with a flux mapper, intercepting uneven magnetization, weak magnetization, and wrong pole count.
Dynamic balancing. Rotating assemblies get G1.0 or G2.5 dynamic balancing, avoiding high-speed vibration causing loosening.
Dimensional and geometric. Optical/CMM measurement of key dimensions, concentricity, runout, flatness.
100% outgoing inspection. Critical safety parts (such as robot joints, pump inner rotors) are fully inspected piece-by-piece, not sampled.
Magnet failures are rarely "the magnet isn't good" — more often "bonded poorly, assembled poorly, inspected poorly." Debonding, heat demagnetization, coating delamination, cracking, uneven magnetization — each of these five failures has a clear root cause, and each can be intercepted in advance through systematic adhesive process, assembly control, and full-process inspection.
For buyers, selecting a magnet supplier can't be based only on the magnetic-property spec sheet and price quote — you must also look at whether it has this "invisible professionalism" in bonding and assembly. The same N52 magnet in different factories' hands can differ tenfold in lifespan.
For magnet-bonding process evaluation, high-temperature adhesive selection, or failure analysis of custom magnet assemblies, please feel free to reach out — FAIZEAL can provide technical support.