If a container of pure water is rotated fast enough at high rpm, would $\mathrm{D_2 O}$ separation be feasible?

Another way to ask is: it practically and physically possible with current technology to spin the water container at sufficient rpm (revolutions per minute) to accomplish heavy water separation?

  • $\begingroup$ I suspect vibrations will work against the natural tendency of heavy water to separate, but I don't see a reason why such method would not work in a vibration-free environment e.g: space $\endgroup$
    – lurscher
    Aug 17, 2017 at 16:42
  • $\begingroup$ @lurscher I think vibrations can be handled with precise engineering and maybe with elastic compensation layers. More problem what I can see, is the ordinary diffusion. There is a good trick, how can it be calculated, but unfortunately this comment is too short for me to write there. $\endgroup$
    – peterh
    Aug 18, 2017 at 1:13
  • 1
    $\begingroup$ @peterh even with diffusion you would get differential concentrations at different radius of the centrifuge (much like enriched uranium centrifuges) and you could get a separation pipeline by concatenating such centrifuges resulting in increasingly concentrated heavy water $\endgroup$
    – lurscher
    Aug 18, 2017 at 13:18
  • 1
    $\begingroup$ @lurscher Exactly. And, with a little trick, the barometric height formula can be used to calculate the exact dependence of the separation on the length, rotation speed and the component densities. Unfortunately, the results don't show very easy configurations from an engineering perspective, but yes it is possible. $\endgroup$
    – peterh
    Aug 18, 2017 at 13:28
  • 1
    $\begingroup$ Wouldn't it be easier to do in the gas phase? That's how uranium is separated (and the mass difference between UF6 (the compound used in gas centrifuges) with U-235 and U-238 is <1% compared to >10% for D2O vs H2O.) We know this works for uranium. $\endgroup$
    – matt_black
    Feb 8, 2023 at 17:14

2 Answers 2


Using a standard (not continuous) centrifuge, I would expect results ot be fairly dissapointing, as on "power down", I would anticipate a lot of the gains to be undone by diffusion.


If it were possible to use a continous centrifuge - and then cascade them in such a way that each stage was only expected to provide a modest enrichment, I would see that it would almost HAVE to work.

With a stop/start style centrifuge, if you were able to modify it, such that longer tubes were possible and you worked on the basis of just "discarding about 90%" each time, I see not reson why you could not end up with "Fairly heavy water".

Of course once you have "heavyish" water, you can start to play around with different freezing temperatures of H2O - v - D20 at both normal and reduced / increased pressures. Also electrolysis and then Girdler/Sulphide - even if just a single stage column?

As is often the case, the answer to your questions is "sort of". Certainly you could increase the percentage of D20 - but as a method og producing high purity D2O, I think it would be flawed.


I am prety sure it would work, but separation factor is pretty low(1.1111…) because D2O is only 10% heavier. Its just inneficient.

  • $\begingroup$ How was that separation factor obtained? It would be instructive to see the equations that determine how feasible this is. $\endgroup$ Oct 9, 2018 at 21:14
  • 3
    $\begingroup$ Man, a 10% separation in one step would be amazingly better than any other method... $\endgroup$
    – Jon Custer
    Oct 9, 2018 at 21:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.