Can You Weld a Magnet? The Real Answer, Temperatures, and What Works Instead

Saturday afternoon, garage door open, the project almost done: you’ve built a little steel mounting plate for a bench sensor, the beads are clean, the fit is right, and there’s one step left — a chunky neodymium magnet that needs to sit flat on the plate so the whole thing clips onto the machine frame. Your welder is still hot. The magnet is right there on the bench. Twenty seconds of tacking seems so much faster than drilling and tapping two holes, and nobody is watching. So you touch the electrode to the edge of the magnet, the arc bites, and something inside your head says that was a bad idea before the puddle even solidifies.

When the part cools, the magnet is a dull grey lump. It won’t pick up a screw. You tap it against the bench — nothing. The bracket goes in the scrap pile, the magnet goes in the bin, and the job that was twenty seconds from finished now needs a replacement part ordered because nobody stocks that grade locally. This happens in real shops more often than anyone admits, because the question “can you weld a magnet?” sounds like it should have a simple yes-or-no answer, and the internet is full of both.

Here’s the honest version, and we’ll spend the rest of this article backing it up with real numbers: you can absolutely put an arc on a magnet — and you will almost certainly destroy it. The heat of welding exceeds every permanent magnet’s Curie temperature by an order of magnitude, so magnetism dies long before the metal melts. On top of that, the shrinkage of a cooling weld bead cracks the magnet body, because magnet materials are hard, brittle, and have a fraction of steel’s tensile strength. And as if the magnet’s destruction weren’t enough, the magnet fights back — its field bends your arc and wrecks the weld you were trying to make. Three separate failures, one careless tack.

Key Takeaways: You cannot weld a permanent magnet and keep it functioning as a magnet. A welding arc runs in the thousands of degrees while common magnets lose their magnetism at 310–460 °C (neodymium and ferrite) and start degrading at working temperatures as low as 60–80 °C. Welding heat also cracks the magnet body, since neodymium has roughly 78 MPa tensile strength against steel’s 350–400 MPa and almost no ductility. The workarounds that actually work are: weld the steel case instead of the magnet, fit the magnet after all hot work is finished, bolt it, epoxy it, or braze the casing — and if a magnet must sit near your arc, manage heat with sinks, spacing, short tacks, and constant temperature checks. Never forget that a magnet near the joint also causes magnetic arc blow, which produces the lack of fusion, undercut, and porosity you’ll otherwise blame on your technique.

The Short Answer: Can You Weld a Magnet?

Physically yes, practically no — and the practical no is the one that matters. Strike an arc on a neodymium magnet and you will melt something, fuse something, and produce a joint that looks like a weld. Metallurgically, nothing prevents you from depositing filler across the boundary between a magnet and a steel plate. But the moment the arc touches it, two irreversible things begin: the temperature of the magnet races past its maximum working temperature and toward its Curie point, and the weld pool begins solidifying and contracting against a material that cannot stretch. When you’re done, you have a cracked, demagnetized lump welded to a plate, with a heat-affected zone full of the defects that happen when an arc behaves unpredictably near a magnetic field.

The physics is not close, either. A welding arc core sits somewhere around 6,000 to 10,000 °C. The heat-affected zone several millimeters away from the puddle is still well above 500 °C. Steel filler metal melts near 1,450 °C. Against those numbers, standard neodymium stops being a magnet at about 310 °C, ferrite around 450–460 °C, and even the famously heat-tolerant alnico tops out near 750–860 °C for its Curie point. Every one of them is destroyed as a magnet before your puddle even finishes solidifying.

That’s why the honest answer has to be a no for anyone asking “can I weld this magnet into my part and have it still work?” — the same kind of honest no you get when people ask whether they can fuse two metals that form brittle intermetallics, as covered in our guide on welding aluminum to steel. The fact that the arc will happily run does not mean the result will do what you need it to do. What you can do is redesign the join so the magnet never sees the heat — and that’s the second half of this article.

Two Different Things Called “Welding Magnets”

Worth thirty seconds, because it explains half the confusing search results you’ll find on this topic.

Welding a magnet means joining a permanent magnet — a neodymium block, a ferrite disc, an alnico bar — to a metal part using a welding process. This is what most people mean, and the answer is a careful no. Welding magnets (plural, the tool category) means the magnetic squares, arrow clamps, ground clamps, and pipe jigs sold in every welding supply catalogue. These are built to survive near-arc duty, and they’re genuinely useful tools. Their design also explains precisely why your own magnet won’t survive — because the manufacturers specifically choose magnet materials with Curie temperatures and working temperatures high enough to live next to a puddle, and enclose them in steel bodies that take the heat.

If you’re shopping for gear rather than trying to save a magnet, skip ahead to the section on magnetic tools further down. If you’re holding a magnet and a stinger, keep reading — the physics decides your job for you.

What Actually Happens to a Magnet Under an Arc

You’ve probably read that magnets work “because atoms have unpaired electrons that spin in the same direction.” It’s a half-truth, and more importantly, it doesn’t help you predict anything about welding. The useful mental model is the one materials engineers actually work with: domains.

A permanent magnet isn’t one solid magnetic object. It’s a crowd of microscopic regions called magnetic domains, each internally aligned, all pointed the same way after magnetization. Magnetizing a magnet means lining those domains up with a field strong enough to flip them. Demagnetizing it means knocking them out of line — and here’s the part that surprises people: every individual atom stays perfectly magnetic the whole time. The magnet’s external field collapses anyway, because the domains no longer agree with each other. The sum of trillions of tiny magnets pointing in random directions is zero.

Why Heat Is the Enemy of Alignment

In ferromagnetic materials — iron, nickel, cobalt, and their alloys — what holds the domain alignment together is an exchange interaction, a quantum mechanical coupling between neighboring atoms that costs real energy to break. Heat is precisely the energy that breaks it. Warm a magnet gently and the domains wobble and disorder; you lose a predictable fraction of pull, and it comes straight back when the part cools. This is reversible loss, and it’s why a magnet feels weaker on a hot day and stronger on a cold one.

Push past a certain point and something more fundamental happens: thermal energy overwhelms the exchange interaction completely, the crystal structure reorganizes into a phase that cannot sustain long-range alignment, and the material stops being ferromagnetic at all. It becomes paramagnetic — weakly attracted, essentially useless as a permanent magnet. That threshold is the Curie temperature, named after Pierre Curie, who identified the behavior in the 1890s. It’s a fixed property of the material, not of the shape or size of the magnet, which is why it answers this entire welding question in one number. You can read the details on Curie temperature at Wikipedia, but the practical takeaway is simple: above this temperature, there is no magnet, only a piece of metal that used to be one.

The trick is that different magnet materials hit that point at wildly different temperatures — which is why the alnico core inside a professional welding clamp shrugs off the heat of a nearby arc, while a neodymium magnet sitting two feet away quietly dies. Material selection is the whole game, and it’s also why “can you weld a magnet?” doesn’t have a single answer until you specify which magnet.

The Grain Structure Matters Too

It’s not only about the temperature number. Sintered neodymium is a composite of a hard magnetic phase grains embedded in a neodymium-rich boundary phase. That boundary phase softens at relatively low temperatures, and repeated heating coarsens the grain structure — a metallurgical change that survives cooling. Ferrite is a ceramic: oxygen bonded into a crystal lattice, hard and utterly without ductility. Alnico is a metal alloy of aluminum, nickel, cobalt, iron and copper that behaves like a proper metallurgical material — it can be cast, machined, and, critically, it tolerates thermal cycling the way alloys do. Samarium cobalt is a rare-earth intermetallic that’s more thermally stable than neodymium but notoriously brittle. Each of these microstructures reacts differently to the sudden, violent heat of an arc, and none of them reacts well.

The Temperatures That Settle the Argument

Two numbers matter for every magnet material. The maximum working temperature is where you start losing strength permanently — above it, the magnet doesn’t fully recover on cooling. The Curie temperature is where the material cannot be a magnet at all, no matter how hard you try to magnetize it afterward. Here’s the data for the common types, compared against what a welding arc actually does.

Now put those numbers next to the process. Arc core temperatures of 6,000 to 10,000 °C. A heat-affected zone that’s still above 500 °C several millimeters from the fusion line. Molten steel at 1,450 °C. The comparison isn’t close: every common permanent magnet material has a Curie temperature roughly an order of magnitude below welding heat. There is no grade of neodymium, no brand name, no specialty electrode, and no clever tacking technique that changes that arithmetic.

One detail catches people out constantly: the strongest neodymium grades are the worst at temperature. N50, N52 and N50M are typically rated to only about 60 °C — below the standard N-series 80 °C — because that extra pull force is bought at the cost of coercivity margin. If you buy the strongest magnet on the shelf, you’ve bought the one least able to survive getting warm. And since a typical bench magnet project uses whatever strong little block was cheapest, most failed welding attempts involve the most heat-sensitive material in the catalogue.

For context on why the whole industry leans on these numbers rather than guesswork, bodies like AWS standards, ASM International reference data, and the metallurgy research summarized by TWI all treat maximum service temperature as a hard design constraint, not a suggestion. If a design review would never let a component run above its rated temperature, a magnet has no business sitting in a welding fixture’s heat-affected zone.

It’s Not Only the Heat: The Magnet Cracks First

Here’s the part most articles skip entirely. Assume for a moment that magnetism somehow survived the arc — say you got lucky, or the magnet is alnico, or you found some miracle cooling scheme. The magnet would still be broken, because permanent magnet materials are mechanically hostile to the welding process.

Take a standard sintered neodymium magnet. Density around 7.5 g/cm³. Vickers hardness near 570 HV — far harder than most weld metals, so a soft filler tends to smear rather than wet and flow. Compressive strength around 780 MPa, which sounds reassuring until you look at the number that actually matters in a joint. Tensile strength is roughly 78 MPa. Mild steel sits around 350 to 400 MPa. A neodymium magnet has about a tenth of steel’s tensile strength and essentially zero capacity to elongate before fracture — no yield, no plastic deformation, just a snap.

A weld bead contracts as it cools — that shrinkage is a real tensile load acting across the joint. Steel yields a little and absorbs it. A magnet cannot, so it fractures, often with an audible tick, sometimes while your clamp is still tight on it. Ferrite is worse: it’s a true ceramic, and ceramics fail catastrophically. Samarium cobalt is described by its own trade as prone to chipping. And neodymium fragments are sharp enough to embed in skin — “it exploded into pieces” is a genuinely common workshop outcome when someone clamps one too hard, taps it with a hammer, or, yes, tacks it.

This is also why experienced welders treat brittle materials with such caution generally. If you’ve ever watched a careful procedure for welding cast iron to steel, you know the drill: preheat slowly, control the cooling rate, keep the bead short, because a material that can’t stretch will find the shrinkage stress and split along the weakest path. A magnet is cast iron’s impatient cousin — no ductility, no tolerance for thermal shock, and a heat rating measured in tens of degrees instead of hundreds.

So the honest answer has two halves, and both have to fail for the job to fail — which they do, together, every time. The heat destroys the magnetism, and the weld’s own shrinkage destroys the body. Welding a magnet isn’t a heat problem with a clever workaround; it’s a heat problem and a fracture problem arriving at the same moment. And there’s a third problem waiting behind them, which we’ll get to after we sort out what “demagnetized” actually means.

Three Kinds of Magnet Loss (Only Two Are Recoverable)

“The magnet got demagnetised” covers three physically different situations, and knowing which one you have decides whether the part is scrap or repairable. This distinction matters far beyond welding — it applies to any magnet that’s seen a hot engine bay, a summer car dashboard, an industrial oven nearby, or a careless pass with a heat gun.

1. Reversible Loss

Strength falls with temperature along a predictable coefficient and returns completely on cooling. Standard neodymium loses about 0.12% of its output per degree Celsius above room temperature — so a magnet sitting 20 °C above ambient has given up roughly 2.4% of its pull and will get every bit of it back once it cools. Nothing is damaged, nothing needs fixing, and this is the only category where you can honestly say nothing happened. Every magnet in the world does this all the time.

2. Irreversible but Recoverable Loss

Exceed the maximum working temperature and the magnet’s operating point gets driven past the knee of its demagnetisation curve. Some domains flip and stay flipped even after the part cools. The useful detail: the material’s intrinsic coercivity has survived, so re-magnetising can restore it. From the outside the magnet just looks weaker — same size, same weight, less pull. A side benefit is that the magnet is now thermally stabilised at that temperature; repeating the same exposure won’t cost anything further. This is the category you land in when you hold a neodymium magnet at 100 °C for a while with a heat gun. It’s also the category a careful tack-weld might leave you in if you got very lucky with heat sinking.

3. Irreversible and Irrecoverable Loss

Go past the Curie point, or hold a high temperature long enough for the microstructure to change, and the material undergoes a metallurgical transformation — oxidation, grain coarsening, a different phase on cooling. Re-magnetising will not bring it back, because there’s no longer a structure capable of holding alignment. This is where a properly heated neodymium magnet ends up, and it’s why a cooked magnet is scrap. Cracking belongs here too: the fragments may each show local magnetism when you hold a compass near them, but the assembly no longer works as one magnet, and no amount of magnetising fixes a broken part.

The test is simple enough for any shop. Let the part cool completely, then check it against a known-good magnet of the same size — a paperclip test, a gauss meter reading, or a pull-force check on a scale. If it came back, you had reversible loss. If it’s weak, try re-magnetising before you throw it out. If it’s dead and it got really hot, you’re in category three and it’s over.

Welding a Magnet to Steel: The Scenarios People Actually Attempt

The abstract question hides a dozen concrete jobs, and each one fails slightly differently. It’s worth walking through them, because the fix for each isn’t quite the same.

Tacking a magnet flush onto a steel bracket

The classic mistake, and the one from the story at the top of this article. The magnet is in direct contact with the steel, the arc strikes at the interface, and heat conducts into the magnet faster than you can believe — neodymium’s thermal conductivity is decent (around 8–10 W/m·K), so a magnet touching the weld zone heats through in seconds. The magnet reaches its working temperature almost instantly, then its Curie point if the arc lingers, while the weld pool shrinks against a body with 78 MPa tensile strength. The magnetism goes, the body cracks, and your bead probably has a cold-lap or lack-of-fusion defect on the magnet side anyway because filler won’t wet a hard, oxide-covered rare-earth surface the way it wets clean steel.

Butt-welding a magnet inside a steel channel or pocket

This sounds more controlled — the magnet is shielded by steel on three sides, and you’re only closing one seam. It isn’t, because steel is an excellent conductor of heat and the channel walls conduct it straight to the magnet. The pocket also traps heat: where an open bead can radiate and convect away, an enclosed channel holds temperature in. Add the fact that the root of a channel joint is the hardest place to control penetration, and you get a magnet that’s been cooked from both sides and a root pass with the classic lack-of-penetration profile. If your channel welds are developing strange root defects, our guide to porosity in welding and the related defect patterns are worth a look before you blame the rod.

Welding a magnet that’s already bonded to a part

Sometimes the magnet arrived glued in — a sensor module, a purchased assembly — and the job requires welding the assembly’s steel tab to a frame. Now you have two heat-sensitive materials stacked: the adhesive softens somewhere around 80–150 °C for most epoxies and loses all load capacity, and the magnet behind it cooks anyway. The correct move is disassembly: pop the magnet out (or unbolt the module), weld the steel, then refit. If the magnet won’t come out, you’ve inherited a design problem, not a welding problem.

Welding near a magnet that must stay magnetised

Some assemblies need the field intact — Hall-effect sensors, magnetic holders, motor housings, latches. Here the magnet doesn’t have to be welded to, just near. Even so, you’re in the danger zone: the heat-affected zone extends millimeters from the fusion line, and heat conducts along steel far faster than people expect. A magnet 15 mm from a fast root pass can still see 150–200 °C if you’re not managing the cycle. Section by section below covers what to do about that, but the underlying rule never changes: track the magnet’s working temperature, not the arc’s.

Can You Weld a Magnetic Jig or Clamp Body Without Ruining It?

This is the other half of the search intent, and it deserves a real answer rather than a hand-wave, because plenty of fabricators have wanted to modify, repair, or extend a magnetic fixture.

If the magnet inside is alnico, you have genuine margin — but not unlimited margin. Professional pipe-welding angles and heat-resistant holders use alnico cores specifically rated to about 450 °C service temperature, with Curie points in the 750–860 °C range. That’s an enormous gap between “working” and “dead,” and it’s exactly why these products exist. Welding a bracket onto the steel body of such a clamp, with the core inside, is survivable if you keep the heat brief and sink the heat away from the core. Manufacturers of alnico holders ship them with a removable steel yoke closing the magnetic circuit — part to protect the magnet in transit, part to give you a way to take the magnet out before hot work. Use that feature. Take the core out, weld the body, put the core back. It costs thirty seconds.

If the magnet inside is neodymium or ferrite, don’t. Even switchable magnetic ground clamps and rare-earth butt-weld holders that use NdFeB are designed around the assumption that the magnet never gets hot — the steel body, the fin geometry (modern designs drop clamp body temperature noticeably), and the switching mechanism all exist to keep the magnet at ambient. Add an arc and you exceed its 80–150 °C rating in a single pass. The clamp will still look fine afterward. It just won’t hold anything.

What about welding a plain steel fixture that a magnet will sit on later? Perfectly fine, and this is the right way to build magnetic jigs. Weld the fixture complete, let it cool, clean the spatter off the mating face, then add the magnets — bolted, epoxied, or dropped into machined pockets. Fabricators who build their own magnetic squares and holding fixtures universally arrive at this sequence, because it’s the only one that works. If you’re earlier in your shop journey, our welding tips for beginners collection covers the fixture and fit-up fundamentals that make this kind of shop-built toolmaking practical.

And the reverse case — welding a part while it’s held by a magnetic clamp — is fine, with one caveat. The clamp isn’t being welded; it’s holding. The caveat is arc blow, which is the subject of the next section, and which costs more bad welds in real shops than almost any other invisible cause.

What Actually Works Instead

The design problem is solvable even though the welding problem isn’t. These approaches are ordered roughly by how comfortable an engineer sleeps afterward.

Weld the case, not the magnet

Most real magnet assemblies are already dressed in steel. A soft-iron yoke or pot shell closes the magnetic circuit and concentrates the field where it’s wanted, and the magnet sits inside, protected. Weld the yoke. Industrial magnetic chucks, sensor housings, lifting fixtures, and motor assemblies are all attached this way, and it works because you’re welding ordinary low-carbon steel with a known procedure while the magnet sits behind a thermal buffer. Heat still conducts inward, so keep heat input low, use several small passes instead of one big one, and watch the total cycle against the magnet’s working temperature rather than the arc’s.

Weld the steel first, magnetise last

The design-stage answer, and the one that eliminates the problem instead of managing it. Build and weld the whole steel assembly, stress relieve it if needed, machine it, coat it — only then fit the magnet or magnetise it in place. Any bracket that will need welding later should be bolted so the magnet can come out before hot work begins. It costs a little more at design time and nothing forever after. This is also how professionals assemble dissimilar-metal structures and coated assemblies in general: sequence the hot operations first and the sensitive ones last.

Mechanical fastening

Bolts, rivets, screws, threaded inserts, press fits, machined pockets with a retaining lip. No heat reaches the magnet, the joint is reversible, and maintenance is trivial — when the magnet eventually fails from age or impact, you replace it in five minutes. If the joint doesn’t need to be hermetic or extremely strong, this is simply the right answer, and it’s the one every production designer reaches for first.

Adhesive bonding

Structural two-part epoxy joins magnet to metal with zero heat into the magnet. Keep the bond line thin — thick adhesive creeps under sustained load, and a magnet that creeps is a magnet that has moved out of calibration. Surface-prep properly: degrease both faces, scuff the steel, and use an epoxy rated for the temperature the assembly will actually see. Design the bond to fail before the magnet does — the magnet is a consumable part and should be replaceable. Cyanoacrylate works for light, indoor, room-temperature duty; for anything structural or warm, a two-part epoxy or a magnet-rated adhesive is the only honest choice.

Brazing and soldering — with honest limits

Common silver brazes run roughly 450 to 700 °C, which is far kinder than an arc, but 450 °C still exceeds every neodymium working rating by a wide margin. Brazing genuinely works for ferrite and alnico assemblies — this is a standard industrial technique for joining ferrite magnets to steel poles. For NdFeB it trades total destruction for partial permanent loss, which is arguably worse because the part looks like it survived. A low-liquidus braze around 400 °C is the best case, and even then you’re over the limit. Soldering with an iron at 330–400 °C has the same problem: it’s over the 80 °C neodymium limit, so solder to the steel casing (the normal method) or directly to ferrite, whose 250 °C working rating comfortably clears a low-temperature solder.

Re-magnetising afterwards

Possible, rarely practical. Restoring a magnet needs a field strong enough to overcome its coercivity, applied in the correct direction, after full cooling — pulsed coils or a magnetising fixture, not another permanent magnet. Sticking a strong magnet against a weak one will not reliably do the job; the field geometry is wrong and the strength is usually insufficient. If the part survived the heat without cracking and you have access to a magnetiser, it can come back. In most workshops, replacing the magnet costs less than the magnetising service, and you’ll never be certain how much of the original strength you got back.

The pattern in that table is unmistakable: every method that works puts distance — physical or thermal — between the magnet and the heat. And notice that none of the surviving entries is a fusion-welding process applied to the magnet itself. That’s not a coincidence; it’s the physics we covered above, written out as a design rule.

How a Magnet Wrecks Your Weld, Not Just Itself

This is the effect that catches experienced fabricators, and it runs in the opposite direction from the original question. A welding arc is not just heat and light — the welding current flowing through the electrode and workpiece creates its own magnetic field around the arc, and that field is what helps hold the arc column on the shortest path between electrode and work. Introduce an external field — from a permanent magnet sitting on your joint, from residual magnetism in the plate, from a magnetic clamp — and you disturb that balance. The arc gets pushed off its path. That’s magnetic arc blow, and it’s one of the most misdiagnosed problems in the shop.

The defects it produces are recognisable once you know the signature: lack of root penetration, lack of fusion, undercut on one side of the bead, and porosity that seems to appear from nowhere. Industry literature also lists irregular bead width, crooked and wavy beads, and excessive spatter. Small amounts of arc blow are sometimes even handy for shaping a bead profile — pipe welders use controlled blow deliberately — but once it’s producing defects it has to be managed.

How much field is too much?

The numbers come from steel-industry and pipeline practice. Meaningful arc disturbance starts at roughly 40 Gauss in the field around the workpiece, with some highly permeable materials reacting at 20 Gauss. For scale, the Earth’s own field is about half a Gauss — so we’re not talking about subtle effects. In pipe welding the geometry concentrates the field brutally: a few tens of Gauss measured in open air can reach 1,000 Gauss in the root gap, because the gap acts as a magnetic concentrator. If you’ve ever watched a root pass wander sideways for no visible reason, this is usually what happened.

A permanent magnet sitting on your joint is a far stronger source than residual plate magnetism — a typical rare-earth clamp produces hundreds or thousands of Gauss at its face. Treat it as a second magnetising source feeding the problem directly at the worst possible location.

The fixes that actually work

Move the magnet. The simplest option available, and usually sufficient. If it’s holding the work, hold the work another way for the affected segment. Switch to AC. The field direction reverses with the current and induced eddy currents oppose external fields, so AC is far less susceptible to arc blow than DC — this is one of AC’s genuine remaining advantages, and it’s why some operators flip polarity for problem joints. Our explainer on MIG polarity walks through what DCEN and DCEP each mean for heat and penetration if you want the fuller picture. Move the earth connection. Return current through the work creates its own field; for back blow, position the connection away from your welding direction, and for forward blow, towards it. Incline the torch against the direction of the blow — a temporary, local fix, but effective. Degauss the steel with an AC field slowly reduced to zero, remembering that line-frequency degaussers have poor skin depth and may clean the surface while leaving magnetism underneath. Check first with a cheap handheld gauss meter: thirty seconds of measurement beats an hour of blaming your technique.

Porosity from arc blow is especially treacherous because the bubbles look exactly like dirty metal, wet rod, or a bad gas setting — and many welders go chasing shielding-gas problems (see our MIG gas settings guide) when the real cause is an invisible field bending the arc and blowing the puddle open. If your porosity patterns look strange, or the defects cluster on one side of the bead near where a magnet or clamp sits, check the field before you change anything else. And if you’re still building foundational skills, the defect-recognition material in our MIG versus TIG comparison explains how each process behaves when the arc gets disturbed.

Welding Magnetic Steel vs. Welding an Actual Magnet

Here’s a confusion that deserves its own section, because it trips up beginners constantly: steel is ferromagnetic — it’s attracted to magnets, and magnetised steel behaves like a weak magnet — but steel is not a permanent magnet. Those are two completely different things, and welding them is two completely different problem sets.

When you weld mild steel, nothing about its “magnetism” matters, because mild steel in its normal state has no net field. Its domains are randomly oriented; it’s attracted to a magnet because the magnet temporarily induces alignment in those domains, and it releases when the magnet moves away. Running a bead on it is an ordinary metallurgical operation: melt, mix, solidify, cool. The result is another piece of steel with the same useful properties. If the steel happens to be residual-magnetised from storage or from a previous welding operation, your only concern is arc blow — covered above — not the survival of anything.

When you weld a permanent magnet, you’re attacking a component whose entire function is a fragile alignment of domains held in place by an energy balance that heat destroys. The magnet doesn’t just tolerate the process poorly; the process is fundamentally incompatible with the thing that makes the part what it is. You’re not joining two pieces of material — you’re destroying the property of one of them while simultaneously trying to bond it.

This distinction also explains why welders rarely worry about welding “magnetic” materials in general. Nickel alloys, for instance, are used in many corrosive or high-temperature services, and our notes on nickel-based welding wire cover their joining behavior — but nickel’s usefulness in those applications doesn’t depend on it holding a field. The same goes for every structural alloy in the shop. Weld strength in ordinary steel comes from fusion, filler chemistry, and joint design — you can see what contributes to a good number in our guide to how strong a weld actually is — and none of those factors has anything to do with magnetism.

So the honest summary: if your part is steel, weld it and stop worrying. If your part is a magnet, stop and redesign the join. If your part is steel with a magnet in it, take the magnet out first — or weld the steel before the magnet ever goes in.

If the Magnet Must Stay: Buying Time on the Clock

Sometimes the drawing says the magnet stays, the customer says the magnet stays, and the schedule says you start now. You can’t redesign the part today — you can only slow the heat down. None of this makes welding a magnet a good idea, and none of it makes the magnet “safe” — but it’s what a careful fabricator does while the redesign argument happens in the background, and it has saved plenty of jobs from the scrap pile.

Heat-sink the magnet

Clamp a copper or aluminium backing plate against the magnet side of the joint, and pack wet rags or a clay-based heat shield around everything you can reach. Copper pulls heat away from the joint faster than almost anything else you can bolt on — thermal conductivity around 400 W/m·K against steel’s roughly 50 — so it acts as a thermal short circuit that keeps energy from soaking into the magnet. Aluminium works nearly as well and is cheaper and easier to shape. The goal is simple to state and hard to do: keep the magnet below its working temperature for the entire cycle, including the minutes after you stop welding, when heat already in the plate keeps migrating inward.

Water cooling, where the fixture allows it

For fixtures and jigs, active water cooling is the serious version of the same idea. A water-cooled backing bar or a copper channel plumbed to a small pump holds the joint side near ambient while you weld — industrial welding fixtures, particularly in production, use exactly this to control distortion. On a one-off shop job, a wet rag refreshed between passes, a spray bottle, or a piece of copper tubing coiled against the magnet side does a credible imitation. Two cautions: don’t quench the weld itself, because rapid cooling of the fusion zone creates hard, crack-prone microstructures in the steel (and for anyone who’s dealt with that, our walkthrough of cast iron welding technique explains why controlled cooling is everything with crack-sensitive material); and keep water out of the arc — steam and hydrogen both cause porosity.

Distance is free protection

Heat falls off with distance faster than most people expect, and a magnet that’s 30 mm from the puddle is in a completely different world from one that’s 3 mm away. Where the design allows, move the magnet as far from the joint as the geometry permits, and remember that steel conducts heat along its length — a magnet on the far end of a short bracket still gets warm. If you can add a thermal break — a slot, a washer-thick gap, an air space — you’ve cut the conduction path. And if the magnet can be removed for the duration, removed is the only distance that’s guaranteed to work.

Spot heating away from the magnet

Sometimes the job needs localized heat — a shrink-fit, a preheat, a tack that must go in near a magnet-loaded section. Plan the sequence so the hottest operations happen furthest from the magnet first, and finish the nearest segment last, when everything else has already shrunk and the total remaining heat input is smallest. Use the lowest-amperage process that will still make a sound weld: fine wire instead of thick (our MIG wire size chart shows how much current a given diameter pulls), short stitch beads instead of a continuous run, and pauses between passes long enough that the magnet side stays under its limit. A temperature stick or a cheap thermocouple tape stuck to the magnet side turns “I think it’s okay” into a number you can act on — set your alarm point well below the magnet’s rating, because heat is still soaking inward after you stop.

Stagger the tacks and sequence the joint

Four small tacks with cooling pauses beat one continuous bead every time, because total heat input is what cooks the magnet, and total heat input is heat per pass × passes × time — shorten all three. Tack the side opposite the magnet first, let it cool, then work progressively toward it. This discipline is the same heat-control philosophy behind every low-distortion procedure, and it’s covered in more detail across our flux-cored settings guide and TIG settings chart, both of which are really about managing energy input rather than just dial numbers.

Finally: check the result. A paperclip test against a known-good magnet of the same size, or a before-and-after gauss meter reading, tells you whether you got away with it. Expect irreversible loss if the magnet got properly warm, even if it still appears to work — and remember that the magnet is now thermally stabilised at whatever temperature it reached, which sounds like a silver lining until you realise the next job starts from a weaker baseline than the last one.

Choosing a Lower-Temperature Process

If something hot has to happen near a magnet, process selection is your main lever, because heat input is what you’re managing.

Brazing or soldering to the case — lowest heat, preferred wherever the design allows it. TIG — the most controllable of the arc processes: small passes, precise puddle placement, and you stop the moment the joint is made instead of feeding wire continuously; the MIG vs TIG breakdown covers why TIG wins on control while MIG wins on speed, and our companion piece on whether TIG is stronger than MIG addresses the strength question that usually comes next. MIG — productive, but higher heat input per pass; thin wire, low settings, and a short arc keep it civil, and our wire-feed welding guide covers the technique for staying gentle. Flux-cored — excellent for dirty or outdoor work, but generally hotter per pass; if it’s your only option, run the lowest viable settings per the amperage chart discipline and keep passes short. Stick — a heavy thermal footprint and the worst option on this list if a nearby magnet is the priority.

Whatever you pick, the discipline is identical: smallest possible bead, shortest practical arc, let the part cool between passes, and track the total heat cycle against the magnet’s working temperature rather than the arc’s. Where you need a general refresher on keeping those habits consistent, the fundamentals in our beginner welding tips apply as much to a fifth-year welder protecting a sensor as to a first-week student running coupons.

Safety: Magnets Are Not Innocent Parts

Welding safety guides — ours included, at welding safety tips — usually focus on fumes, radiation, and hot metal. Magnets add two hazards that catch people off guard, plus one material-specific danger worth knowing.

The kinetic energy problem

A large neodymium magnet carries no stored energy itself, but it creates stored energy the moment it’s near steel — potential energy in the form of attraction, released as kinetic energy when something gets between the magnets or between magnet and steel. Two 50 mm blocks slamming together can shear skin, crush fingertips, and chip each other into razor-sharp fragments that fly. Welders working around magnets learn to keep ferrous tools, offcuts, and hands out of the gap. The snap is fast enough that you cannot react to it — you can only prevent it.

The projectile hazard is worse near the arc. A magnet that jumps toward a steel part while your hands and face are close to the puddle doesn’t just pinch — it can pull a tool, a scrap, or the workpiece itself out of position mid-weld. And a cracked magnet fragment launched by that snap becomes a hardened, sharp-edged missile. Keep magnets secured or well clear of the joint zone while you’re welding, and treat any magnet that has started to chip as compromised — the chips are sharp, and the remaining body is now unpredictable.

The official framework for the rest of the hazards — hot work, fumes, electrical, fire — is set out by OSHA’s welding, cutting and brazing standards, and every shop should be familiar with them whether or not an inspector is coming.

Samarium cobalt: the fume hazard that isn’t thermal

If your magnet is samarium cobalt rather than neodymium or ferrite, the hazard conversation changes completely, because SmCo is cobalt-based and cobalt is a genuinely dangerous material to inhale.

Chronic cobalt exposure is associated with interstitial lung disease, respiratory sensitisation, allergic dermatitis, and cardiac effects. Combined with tungsten carbide it’s specifically linked to hard-metal lung disease. The exposure limits are low — OSHA’s permissible limit for cobalt compounds is 0.02 mg/m³ as an eight-hour average, and ACGIH recommends a tighter 0.005 mg/m³ with cobalt designated a respiratory sensitiser. Add the physical side: SmCo is brittle, its chips are sharp, grinding dust is an inhalation hazard, and powdered cobalt can burn. This is not a material to casually run an angle grinder over, and it is emphatically not a material to arc-weld.

The practical rule: don’t cut, grind, or weld samarium cobalt magnets. If a job genuinely requires handling them, use local exhaust ventilation at minimum, wear proper respiratory protection — our guide to the best welding respirators covers what actually fits and filters — wear gloves so the chips stay out of your hands, and treat any material removal as a serious exposure requiring containment and a written assessment rather than a casual bench operation.

General shop safety around magnet assemblies

One more: a magnet assembly that’s been welded, cracked, or cooked is a compromised assembly. Sharp edges, loose fragments, unpredictable pull. Deburr it, inspect it, and don’t put it in service as a lifting or holding device if there’s any doubt about the magnet inside. Safety-critical holding applications deserve a known, intact, correctly rated magnet — not one that’s been through an arc and hopes.

Symptoms, Causes, and What to Do

Myths Worth Dropping

“Magnets work because unpaired electrons all spin the same way.” Not usefully so. Domains are the operative concept — otherwise you can’t explain why a fully demagnetised magnet’s atoms are still individually magnetic while the block shows no field.

“Magnets cannot be welded under any circumstances.” Too absolute. Alnico-cored welding clamps rated to 450 °C service are a real product category, and welding the steel body of a fixture around an alnico core is done routinely. What is true: you can’t weld a neodymium or ferrite magnet and keep its magnetism, and welding any magnet risks cracking the body.

“All magnets die at the same temperature.” Working limits run from 60–80 °C for standard neodymium to 500–540 °C for alnico. The right question is never “will it survive” but “which material, and what is its limit?”

“A welding magnet will shield my magnet.” A magnetic clamp shields nothing. It holds work and magnetises work — placed near the root pass, it creates the exact arc blow you were trying to avoid.

“Magnets are just iron, cobalt and nickel.” Those three are the ferromagnetic elements, and in their plain forms they’re soft magnetic materials, not strong permanent magnets. Real permanent magnets are engineered alloys and ceramics — NdFeB, SmCo, alnico, ferrite — each with its own temperature story. Background on the family is on the magnet overview at Wikipedia, and the materials research institutions like NIST publish on these systems treat composition and microstructure as the whole ballgame.

“If it’s still stuck to the fridge, the weld didn’t hurt it.” Pull force is a poor inspection tool. A magnet can lose 30–40% of its strength and still hold a picture card. Check against a known-good reference of the same size, or measure it.

Frequently Asked Questions

Can you weld a magnet to a steel bracket?

Not if you need the magnet to keep working. The arc destroys the magnetism — standard neodymium is finished long before the puddle forms — and the weld’s shrinkage cracks the magnet body. Weld a steel case or yoke if the assembly has one, or redesign so the magnet is fitted after all welding is finished. Bolting or epoxying a magnet into a pocket in the finished bracket takes minutes and never fails this way.

What temperature destroys a neodymium magnet?

Standard N-grade neodymium starts losing strength permanently above 80 °C — about 60 °C for N50, N52 and N50M — and cannot be a magnet at all above its Curie point near 310 °C. High-temperature grades (SH, UH, EH, AH) reach 150–230 °C working temperatures with Curie points around 330–350 °C, which still leaves them nowhere near welding heat. For scale, a welding arc’s heat-affected zone is above 500 °C within millimeters of the fusion line.

Can a demagnetised magnet be re-magnetised?

Often yes, if it wasn’t heated past its Curie point, hasn’t cracked, and hasn’t corroded. The crystalline structure and coercivity survive, and a sufficiently strong field in the correct direction restores it. That takes pulsed coils or a magnetising fixture — another permanent magnet will not reliably do it. In most workshops, replacement is cheaper than the magnetising service, and you’ll know exactly what strength you’re getting.

Can you solder to a magnet?

Solder to the steel casing or yoke, which is the normal industrial method, or directly to ferrite — its 250 °C working rating clears a low-temperature solder comfortably. Direct soldering to neodymium is a bad idea: a soldering iron at 330–400 °C is far past the 80 °C limit and causes irreversible loss even if the joint itself holds. Prep matters too — solder doesn’t key into a rare-earth surface the way it does into tinned steel.

Can I use a magnet as a welding ground?

Not as the earth connection itself — a raw magnet isn’t designed to carry welding current, and the current path through it is neither reliable nor safe. A purpose-built magnetic ground clamp is a legitimate, genuinely good tool: it makes contact through a designed steel path, grips where a bolted clamp can’t, and saves you tacking temporary ground tabs on sheet or pipe. Watch for the arc blow issue when positioning it, and switch it off between welds so it doesn’t collect every filing on your bench.

Does a magnet near the weld damage the weld?

Yes, through magnetic arc blow. An external field deflects the arc off its shortest path, producing lack of fusion, lack of penetration, undercut, and porosity. Meaningful disturbance starts around 40 Gauss at the workpiece, and a root gap can concentrate a weak ambient field to over 1,000 Gauss. Check with a gauss meter, move the magnet, switch to AC, or reposition your earth point — in that order of simplicity.

Can you glue a cracked magnet back together?

Structurally, yes — two-part epoxy with proper surface prep will hold the pieces. Magnetically, expect partial recovery at best: the fragments retain local magnetism, so stacking them with poles aligned restores some of the original pull, usually less than all of it, because the crack introduces an air gap that the field hates. A crack under clamping load sees real shear, so glue is temporary. If the assembly is safety-critical, replace it.

Which magnet material survives the most heat?

Alnico — maximum working temperature around 500–540 °C, Curie point roughly 750–860 °C — which is exactly why alnico is the standard core for welding clamps and fixtures living near an arc. Samarium cobalt 2:17 is next among the rare-earth grades, good to roughly 250–350 °C depending on grade with a Curie point near 800 °C. If you’re choosing a magnet for a hot environment, alnico is the first thing to spec — and the last thing to weld.

Why does welding a magnet crack it?

Tensile strength and ductility, not just heat. Neodymium has roughly 78 MPa tensile against 350–400 MPa for mild steel, with effectively zero elongation before fracture. A weld bead contracts as it cools and the magnet can’t stretch to absorb the shrinkage, so it splits — often with an audible tick. High surface hardness (around 570 HV) and the thermal shock of arc-on-cold complete the picture. Ferrite, as a true ceramic, is worse still.

Can you weld near steel that is already magnetised?

You can, but expect arc blow and check the field first. Residual magnetism deflects the arc the same way a permanent magnet does — roughly 40 Gauss is where disturbance becomes meaningful, and a pipe root gap can multiply a modest field past 1,000 Gauss. Degauss the area, switch to AC, or reposition your earth and torch angle. If you’re welding aluminium in the same shop, note that the heat and current concerns stack with the oxide and porosity issues covered in our flux-core aluminium and MIG aluminium pages — different physics, same discipline about controlling the conditions at the arc.

Does welding affect magnetism through the joint, even if the magnet is far away?

Heat travels, and so does stress. A magnet 20–30 mm from a heavy pass can still see well over 100 °C in steel, because steel conducts heat quickly and the plate keeps feeding energy outward after the arc is gone. Move further, sink the heat, or take the magnet out. Distance you can measure beats distance you can estimate — thermocouple tape on the magnet side settles arguments in seconds.

The Bottom Line

Can you weld a magnet? You can put an arc to it — what you cannot do is keep a magnet. The arithmetic isn’t close: an arc runs at thousands of degrees, standard neodymium gives up permanently at 80 °C and stops being a magnet entirely around 310 °C, ferrite tops out near 450–460 °C, and even the best-performing alnico at 500–540 °C working temperature can’t touch welding heat for long. And if magnetism somehow survived, roughly 78 MPa of tensile strength against the weld’s own shrinkage would fracture the body anyway — while the magnet’s field bent your arc and put defects in the weld for good measure.

The productive way to think about it is that the design problem is solvable and the welding problem is not. Weld the steel case or yoke. Fit the magnet after all hot work is finished. Bolt it, bond it, or braze the casing where the design allows. Specify alnico when a fixture must live near an arc. Manage heat with sinks, short staggered tacks, water cooling, and honest distance when the magnet has to stay. And when a magnet sits close to your weld, remember it’s acting on your weld too — check the field with a gauss meter before you decide your technique has gone bad.

Back to that Saturday project: the fix took twenty minutes. Weld the plate complete, drill and tap two holes, drop the magnet into a shallow pocket with a thin epoxy film under it for damping, and screw it down. The bracket goes on the machine, the magnet still picks up every paperclip in the shop, and the only thing that got demagnetised was the temptation to take a shortcut. Keep the magnet out of the heat, and the heat out of the magnet — everything else in this article is just detail.