Suppose we have two isotopes of the same element. For example, isotope A could have a nuclear spin I = 0, isotope B could have nuclear spin I = 3/2. Let us say we have a compound where this element is participating, and in this compound we have a mixture of the two isotopes A, B. Up to this point, this is a rather universal case: with the exception of isotopically purified samples, all compounds contain mixtures of isotopes for all of their elements.
Let us now cool down a sample of our compound, below liquid He temperatures, and, for the purposes of this question, let us say we can cool it down arbitrarily low temperatures. Now the different isotopes A, B will start to behave differently, presenting different physical and chemical properties, if only because their quantum eigenstates are different, and differently populated at low temperature. This is on top of their different masses, of course.
The question is:
How could one separate this mixture of isotopes A, B, presenting different nuclear spins, at sub-Kelvin temperatures?
In particular, if more than one purification procedure is possible, can one define one such procedures which is based on the different magnetic properties of isotope A and B?