This is inspired by the recent question:

Does drilling a hole into a hard drive suffice to make its data unrecoverable?

In particular, it seems drilling holes itself might not be enough to stop a serious attacker with access to hyper-sophisticated methods. Others in the comments suggested various highly-dangerous methods as alternatives like applying thermite, acid and other destructive substances which, while they might work, also carry a significant amount of potential physical risk to the user that an untrained user might thus easily be able to hurt hirself with rather seriously (thermite could char a limb, acid eat flesh, not to mention the fumes that would doubtless be released from any chemical reactions like this would probably contain things not good for one's lungs as I'm not sure what all goes into the making of a drive platter.) and/or cause some unwanted property damage. This, then, makes them rather tricky and potentially a liability to use in at least some cases.

Thus this makes me wonder about methods that might be more accessible to those with less training and more simple means, such as, e.g. a home DIY person or something like that (although some may have the power to handle things like thermite, it is not always and most people don't bother with that stuff. To handle it safely I'd imagine it'd be at least advisable to have a thermite welding course under your belt.).

This makes me wonder about the possibility of (and I've wondered about it long in the past) data destruction by the means of simply removing the magnetic oxide recording layer. After all, I believe the disk substrates - usually aluminum and glass - are what are called "paramagnetic" materials. Paramagnetic materials are ones that only acquire any magnetization when a magnetic field is applied . If the field is removed, there is no remaining magnetization at all. Zero magnetization = zero data, not even to a hyper-powered attacker with access to world-class facilities, it's physically impossible. Is this a legit conclusion to make?

As if that's all that needs to be done then it seems there may be two methods that can achieve that. One is to just employ a simple power sander, and sand the drive until there is a color change in the platter, indicating you have gotten through the oxide and down to the substrate. (Not sure if this would be advisable on glass platters though, as I'd wonder if it'd cause them to shatter and then you get a chest full of glass. But for aluminum platters, they should hold up, eh?)

But another more intriguing method is one I've often wondered about and that is the possibility of a deliberate head crash, which could be done by opening the drive and touching the inside thoroughly with your hand to get grease and fingerprints all over the disk on purpose, or even just letting it stand out in the unclean air of a normal room for a few hours, perhaps. No dangerous tools are required for this. There are some interesting pictures that I saw a long time ago here:


where it shows how that a platter was wiped so clean by a head crash that even the glass substrate was exposed (scroll down the page - you can't miss it), indicating zero remaining oxide layer on that part. This apparently occurred with only one drive. Is that type of totalizing damage possible to reliably achieve in any hard drive if you abuse it right, or not, and furthermore, carry to completion so as to render the entire platter clean? (E.g. if you drop the heads then you let it run with some kind of constant seeking going on all the time to ensure those crashed heads smear wild all over the disk for a period of, say, hours or days continuous.) As I'm not sure even the best attackers could recover anything from those glass-clear areas given what I said above about paramagnetism, and that sludge on the inside of the drive is obviously highly entropized which is a good thing insofar as security is concerned. But on the other hand, it could only be a fluke with that one type of drive in that one circumstance at that one time, and this may be otherwise utterly unfeasible or at least far, far too unreliable to be counted on.

Is any of this reasoning sound, or is it 100% full of BS? Also, I'm not advocating any of these methods, rather I'm asking about them out of curiosity.

  • machanical method exist to manually extract the platten data, if available, but that assume the attacker would have access to a selfcontained room and the gear to do that. Anyhow I vtc as opinion based as each recuperation case is a guess even at those facility.
    – yagmoth555
    Aug 19, 2017 at 14:50
  • Of course you don;t need to use a cheap item like a belt sander. Expose the drives to a sufficiently large magnetic field (as found inside a good superconducting magnet) for a short time will be enough to totally rearrange all the magnetic poles. Ask your local hospital if you can use spare time on the MR scanner to wipe your disks, they so won't mind the odd bit of disk drive getting sucked into the works.
    – matt
    Aug 21, 2017 at 15:28
  • @matt : The problem with that is I'd be worried if the drive broke as you suggest when the field is turned on. If pieces of metal get into the machine it could damage it and/or render it a hazard for the next patient it would receive. A sander is rather simpler, and safer, than that. (But as said though apparently this is not necessary anyways, so the question was founded on wrong premise I didn't realize; a simple software wipe may be enough.) Aug 24, 2017 at 7:49

3 Answers 3


Why do you care? The same question you mention already has the answer at your question too:

We can be fairly sure this is possible, because the SANS paper linked from the linked SF article demonstrates that you can't recover data from a platter with an MFM after a single overwriting pass, and such a test would be completely meaningless if you couldn't recover non-overwritten data using the same procedure.

If you've read (I'm assuming you've read at least the posts you've linked before writing such a long question, unless you are a rep hunter) that a single pass full disk rewrite is enough to make anything unrecoverable, why would you want to destroy the platters?

If you think about sanding off, are you sure it would be a safe enough deletion? What if someone finds the powdered magnetic coat in your trash and patiently rebuilds the platters back?

Why don't we build a huge spacecraft and burn old drives right into the Sun?

  • But if this is the case, why were people still mentioning in the comments about punching and melting disks as something they still did, if there is no need to now anymore? That's why I posted the question. Aug 19, 2017 at 15:06
  • I don't know, like I don't know why mankind is still using oil based engines, or why there are still hundreads of wars. What I know is that it's very unlikely that the guys at the SANS Institute underestimate a threat.
    – Marco
    Aug 19, 2017 at 15:07

Why do you care? Have some phantasy.

  • Open the hard disc.
  • Remove platters.
  • Put them into a blender. Get one from BlendTec - they literally blend anything.
  • Press button, watch TOTAL destruction.

Finished. They demonstrate how to annihilate buckyballs - so a little platter won't be a problem.


Hard disc platters are made of glass or aluminium and are non magnetic. If you remove the magnetic coating with a sufficiently fine abrasive then the platters will be non magnetic and whilst the dust will still contain magnetised domains, reconstructing it will be impossible.

  • Impossibile?! You say! I was in the Metro yesterday afternoon, and I heard from two guys that data would still be recoverable even if disks are burnt right into the Sun, as the resulting atoms will still be magnetised! Definitely no drilling nor sanding nor any other type of destruction would ever be more effective than a single pass all zeroes rewrite of the entire drive. And in my professional career I've seen people willing to suicide due to just a shift+canc, so please quit writing such bu****it.
    – Marco
    Aug 20, 2017 at 1:20

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .