Element Collection

Element Collection

Wednesday, August 10, 2011

Neodymium Metal from Hard Drive Magnets, Part I

Note: There's been a lot of work done on this subject on Science Madness, and the actual procedure I used is a bit different (specifically Step 4 & 6). Once I actually complete this experiment, I will update or replace this thread with the final method.

This is the first in a many-part series of posts detailing my efforts to isolate pure neodymium metal from hard drive magnets. This has been an extremely exciting project that I've really enjoyed undertaking, and the journey is just as fun as the final result. I'll be making a video on this eventually, but it's been such a long process that I thought people might be interested in reading some occasional updates on it in the meantime.

This first post will be about the theory, and will detail my reaction scheme.

Modern hard drive magnets are made of an alloy of three elements: Neodymium, Iron, and Boron (hence their other name, "NIB" magnets). This alloy approximately has the formula Nd2Fe14B. In order to get the neodymium separated, we must go through a number of different steps. You can't just melt the magnet because it's an alloy, so any physical separation techniques like that won't work. So, we need to use chemistry! The following steps are my proposed experimental outline.

1) Demagnetize the magnet by heating it to it's Curie temperature, the temperature at which a magnet loses its magnetism.
This can be done by holding the magnet in a propane torch for a few minutes. It's striking how quickly this happens - once the right temperature is reached, the magnet just falls off the tongs I used to hold it. Generally I heat it a bit more after that, because it does remain very slightly magnetic because of uneven heating. This step isn't strictly necessary, but it makes the next one a whole lot easier.

2) Remove the protective outer nickel plating from the magnet, and break it into pieces.These magnets are coated with a protective layer of nickel to protect it from the air, as the Fe and Nd metals would oxidize. We need to remove this coating for the next step. I've found it's easier to do this after breaking the magnet into a few pieces with pliers or a few gentle taps from a hammer. This gives you edges of the coating to grab onto.

3) Dissolve the magnet pieces in sulfuric acid, and filter off the unreacted boron.
This is where the chemistry begins. Dissolving the magnet in acid proceeds according to the following equation:
Nd2Fe14B + 17H2SO4 == Nd2(SO4)3 + 14FeSO4 + 17H2 + B
Nickel is inert towards sulfuric acid at normal temperatures and concentrations, so that's why it needs to be removed. From this equation, you can see that the boron is left unreacted. In this way, we can separate out one of the three components of the alloy. The Nd3+ ion is lavender in color, so if your solution turns purple you're on the right track. If not, it's probably one of the older style magnets (like samarium cobalt) and doesn't contain Nd.

4) Separate the neodymium sulfate from the iron sulfate.
Here we can exploit a very interesting property of neodymium sulfate - it's inverse solubility profile. Nearly all salts dissolve better in hot solution than in cold solution (think of dissolving table salt in water: it goes much quicker if you heat it up). Neodymium sulfate, however, is the opposite: it actually dissolves better in cold solution than in hot. From Wikipedia's Solubility Table, we see that 13g of Nd2(SO4)3 can dissolve in 100g of water at 0C, but only 1.2g can dissolve at 90C! Since this is the opposite behavior of the other salt in the solution, FeSO4, we can easily separate the two by heating. Neodymium sulfate precipitates out as pink or purple crystals when the solution is heated to near boiling, and the green iron sulfate remains in solution. This is a bit more complicated in practice, and I'll go into detail about that in upcoming posts.

5) Convert the neodymium sulfate to neodymium fluoride.
CAUTION: This step is extremely dangerous, and I do not recommend you try this yourself.
Soluble fluoride salts can be quite dangerous, and the fluoride ion is an insidious chemical. Burns from HF acid may not be immediately visible or painful, and symptoms of exposure can be delayed for up to 24 hours. I won't go into it here, but there's a wealth of information on the dangers of ammonium bifluoride online. I'll be working at maximum safety precautions for this step. Don't try this yourself at home.
With that out of the way, I'll be reacting the recovered neodymium sulfate with ammonium bifluoride:
Nd2(SO4)3 + 3NH4HF2 == 2NdF3 (s) + 3NH4HSO4
I'm not 100% sure this is the correct reaction, but the point is we end up with insoluble neodymium fluoride, which can be filtered off and collected. This must be done in a plastic container, because ammonium bifluoride (specifically, the hydrofluoric acid that's released) can eat through glass.

6) React the neodymium fluoride with magnesium powder in a thermite-like reaction.
This is where we finally arrive at neodymium metal:
2NdF3 + 3Mg == 2Nd + 3MgF2
This is why it was necessary to convert the sulfate to the fluoride - because elemental neodymium is such a reactive metal. Other compounds, like the oxides that are used in more traditional thermites, won't work because of neodymium's strong attachment to them. Using the fluoride here provides a high enough reaction enthalpy (i.e. heat) so that the reaction proceeds to completion and the products are in the molten state. The Nd metal will sink to the bottom of the mix, and the magnesium fluoride slag will provide a convenient crust to protect the desired product from air.

7) Recover the Nd metal from the reaction products.
If all goes well, the neodymium should have sank to the bottom of the reaction vessel and solidified in a single lump. After cooling, the whole mess of product will be placed in mineral oil and mechanically separated out. This should be done under oil because elemental neodymium will very quickly tarnish in air, similar to sodium.

So, those are the steps I'll be taking towards isolating neodymium metal for my element collection. As you can see from the process, it would be much cheaper (and faster) to simply buy some but that's not the point. The real goal here is to explore the chemistry, and try to recover something exotic out of something as ordinary as a magnet. Subsequent posts in this series will detail my efforts towards this goal, and I hope you find them as interesting as I have! Stay tuned!


  1. It is quite interesting, but I am not sure it will work. Neodymium fluoride is insoluble in water but it isn't crystalline precipitate. It looks more like jelly and it's hard to filter it and remove water occluded into precipitate.
    Next step (reaction with magnesium) is used some times to obtain this kind of metals but it have to be done into oxygen- and nitrogen-free atmosphere. In other case - the metal will oxidize again to form oxide or nitride.
    From hard drive magnets I obtained once large amount of neodymium compounds (magnet contains 26.5% of neodymium by weight) but I was using different way.
    I dissolved magnet in hydrochloric acid with hydrogen peroxide (to oxidize iron to iron(III)), filtered and precipitated neodymium as oxalate salt using ammonium oxalate. Iron remains in solution forming green coordination compound. Oxalate salt after filtration and washing can be changed to oxide by simple heating in crucible and oxide can be used to prepare many interesting salts.
    Sorry for language errors, English is not my native tongue.

  2. @Agricola: You were exactly right about the NdF3. I just did that part of the reaction last night, and it formed a very fine, gelatinous precipitate that took a while to settle out. I'll try and filter it tonight and see what happens. As for the 'thermite' - the hope is that the reaction products will be in the liquid phase, and the neodymium will sink to the bottom. That way, MgF2 forms a convenient crust on top to protect the Nd from the air. We'll see how that goes when I get to that step.
    Your oxalate route is very interesting, and I'd heard it mentioned before. I'd definitely like to try it to compare to my posted method above. To end at Nd metal I really do need to use the fluoride, so I wonder if NH4HF2 will still react with Nd-oxalate. Thanks for commenting!

    1. the problem is that the oxalate reaction will be very slow. and all oxalates are insoluble in water

  3. For the Cobalt/samarium, how hard would it be to isolate that from the magnet? have u thought about every testing on it?

  4. @Dan: Very thankful to you for your information. and I have a small doubt on this theory that how can we remove Iron from that and in which state it is present in the magnet? Hope for the reply ...
    thank u

  5. Dear you
    Currently I'm studying to recover Nd metal from NdF3. I find that CaO can react with NdF3 to form Nd2O3. I burned mixture CaO and NdF3 at other some temperature range 800-1000oC. But performance is not hight . I continue to study to find out burn condition is the best.
    I want to reference your idea. I wish that you will help me in this problem
    Thank you

  6. Hi I am wondering whether this worked in the end for you? Do you have a video?

    1. Wow, I posted this idea over 3 years ago! Believe it or not this project is still ongoing. I'm actually closing in on the end, though. It turns out step 6 isn't thermodynamically favorable, so instead I'm trying Li metal as the reductant. This requires an inert atmosphere crucible, which has been difficult for me to build. There's also a possible alternative to this step that I'm exploring, so we'll see how everything pans out. If I succeed, a video will definitely be made!

  7. Hi,
    It may be a bit late, but there is a procedure in Brauer's Handbook of Preparative Inorganic Chemistry (p.1143) that uses the chloride instead of the fluoride. It seems interesting, and it's probably safer, too.
    - Chris

    1. Thank you chris. I have been investigating that too. Expect I receive negative input from the community on that idea.

    2. Andrew, the only reason you receive 'negative replies' is because you appear to refuse to listen to us. We've told you countless times to read the relevant threads, but you still ask questions as if you haven't. Having 'no equipment' is a pretty big hindrance to conducting such a complex experiment, too. I'd recommend starting with a simpler isolation first.

      Chloride would definitely be safer, but it takes more energy to carry out the reaction (meaning higher heat needed). The fluoride is more energetically favorable, and has a better chance of yielding good quality metal. This is, believe it or not, already mentioned in the thread. If I am able to isolate Nd from its fluoride, it would be worth it to try the chloride too and compare results.