Why Magnets Lose Strength

Not in any way that matters over a product's life. Long-term ambient aging of sintered NdFeB held within its rated envelope is typically under 1% flux loss per decade — often much less. The magnetization is a property of the crystal structure, and the crystal structure doesn't fatigue from being used: holding, releasing, attracting, and repelling millions of cycles costs nothing.

So when a magnet is measurably weaker than it was, the productive question is never "how old is it?" — it's "what happened to it?" There are exactly five candidate answers, and each leaves distinctive evidence.

The dominant cause. Above the magnet's real thermal limit — which is set by grade class and geometry together, not the datasheet number alone — regions of the magnet flip polarity and the flux doesn't return on cooling. The full mechanism, including why thin open-circuit magnets fail below their rated class, is in the companion page Magnets & Temperature.

• Usual suspects: motor rotors running hotter than modeled, parts near process heat (soldering, welding, cure ovens, sterilization), enclosures in the sun, and under-specified temperature classes chosen on price.
• Evidence: loss correlates with the thermal event; hottest/thinnest regions lose most, so pole patterns scan distorted rather than uniformly reduced.
• Fix & prevention: re-magnetization restores it (the structure is intact) — but it will recur until the grade class, geometry/load line, or thermal environment changes. Class selection guidance: grades chart.

Any external field opposing the magnet's own magnetization pushes it toward demagnetization — and if the combined stress (field + temperature, since hot magnets have far less coercivity) crosses the knee, part of the magnet flips.

• Where it happens: motor fault currents and aggressive field-weakening control (the classic EV/industrial failure), magnets assembled in repulsion, strong magnets stored pole-to-like-pole, magnetizing/degaussing equipment nearby, welding cables draped over assemblies, and MRI environments.
• The compounding rule: field events that are survivable cold become fatal hot — β from the temperature guide is the multiplier. Motor demag analyses are always run at maximum rotor temperature for this reason.
• Evidence: loss follows the geometry of the offending field — one pole, one edge, or the segments nearest the stator — and correlates with an electrical or assembly event.
• Fix & prevention: re-magnetizable. Prevent with adequate Hcj class for worst-case field-at-temperature, assembly sequencing that avoids repulsion dwell, and separation of magnetized stock from field sources.

The one that can't be undone. When the coating is breached — scratch, chip, pinhole, or simply the wrong coating for the environment — moisture attacks the reactive Nd-rich grain-boundary phase, converting magnetic material to non-magnetic oxide and eventually liberating whole grains as powder.

• Evidence: visible — white/gray bloom, blistered or flaking plating, rust-colored staining, powder shed, dimensional swelling. If the magnet looks bad, this is your cause.
• Why the loss is permanent: oxidized material isn't demagnetized — it's gone. Re-magnetizing recovers nothing, and the corrosion keeps advancing under the remaining coating.
• Fix: replacement, with a corrected coating specification. The full selection logic is in the coatings guide; hydrogen environments are a special case where no coating saves NdFeB (use SmCo or ferrite).

First, the myth to retire: dropping a modern sintered NdFeB magnet does not meaningfully demagnetize it. The "don't drop your magnets" rule comes from alnico, whose low coercivity genuinely loses alignment from shock — a caution that transferred culturally but not physically to rare earth magnets.

• What impact actually does to NdFeB: chips and cracks. Lost fragments remove flux in proportion to lost volume; cracks disrupt the local flux path and — the bigger issue — breach the coating, handing the part to Cause 3.
• Machining and abrasion: grinding on a magnetized part removes material, overheats it locally, and strips coating — one event, three causes. (Also a fire hazard: see the safety guide.)
• Evidence: visible damage; flux loss roughly proportional to missing material, localized at the damage.
• Fix: a chipped-but-sealed magnet may serve on with reduced flux (define chip-acceptance limits on the drawing rather than debating each part); cracked or coating-breached parts should be replaced.

• Radiation: significant neutron and high-energy doses progressively demagnetize NdFeB — a real design constraint in accelerators, space systems, and nuclear instrumentation. SmCo is markedly more radiation-tolerant and is the standard choice in those industries.
• Hydrogen: NdFeB absorbs hydrogen and disintegrates — the same mechanism used deliberately in powder production. Not demagnetization but destruction; avoid the material in hydrogen service.
• Deep cryogenics: below ~135 K, NdFeB's spin reorientation reduces usable magnetization — a physics limit, recovered on warming; SmCo serves the cryogenic range.
• What is NOT on this list: ordinary vibration, magnetic "use," water on an intact coating, microwaves in any household sense, and time. If none of causes 1–5 fits, re-examine the measurement before the magnet — which brings us to diagnosis.

• Step 1 — Verify the loss is real. Most "weak magnet" reports are measurement artifacts: a gaussmeter probe at a slightly different spot, a pull test on a different surface, a warmer day. Use a repeatable method — helmholtz total-moment against a known-good reference is the gold standard (testing guide).
• Step 2 — Look at it. Corrosion and damage announce themselves visually. Intact, clean magnet → thermal or field cause.
• Step 3 — Map the loss. Uniform-ish reduction with a thermal history → heat. Loss localized to one pole/edge/region matching an external field geometry → field event. Pole scans make this vivid on multipole and radial parts.
• Step 4 — Reconstruct the event. Thermal excursions, electrical faults, assembly sequence changes, storage changes — the loss almost always has a date.

If the material is structurally intact — thermal and field losses, causes 1 and 2 — a saturation pulse in a magnetizing fixture restores the magnet to full strength. Practical notes:

• It's a factory/lab operation: a capacitor-discharge magnetizer and a fixture matching the part's orientation and pole pattern. Waving a bigger magnet at it does not saturate anything.
• Economics: re-magnetizing loose magnets is rarely cheaper than replacing them; it earns its keep on assembled rotors and bonded assemblies where replacement means teardown.
• It doesn't fix the cause: a re-magnetized rotor returned to the same over-temperature duty loses flux again on schedule. Pair the pulse with the grade, geometry, or thermal fix.
• Corrosion and lost material are not recoverable — replacement with a corrected spec is the fix.

Most of this list is free — it's specification discipline and housekeeping. The handling & safety guide covers the human side of the same practices.