I would like to know if a Mulliken population analysis to calculate spin densities is in general a valid choice. I see that it is made use of, for example here1. So up to-date researches apparently use it. On the other hand there seem to be more advanced methods (natural population analysis). Moreover Wikipedia states:

In principle, a complete basis set for a molecule can be spanned by placing a large set of functions on a single atom. In the Mulliken scheme, all the electrons would then be assigned to this atom.

So there is some ill definition for this method. This would not only account for charge analysis but also spin, if I am not mistaken.

Knowing that my basis set is reasonable; is there anything to consider that would make Mulliken spin density analysis doubtful?


  1. H-Bonding on spin centres enhances spin–spin coupling for organic diradicals by Francis Kirby B. Burnea, Yeonsig Nama and Jin Yong Lee, Journal of Materials Chemistry C, Issue 10, 2020 (link)
  • 5
    $\begingroup$ Great first question! Someone here may be able to provide an answer specific to spin densities, but I expect the spin densities from a population method have similar issues to the charge density. For some details on the pros/cons of different charge analysis methods, see mattermodeling.stackexchange.com/questions/1439/… $\endgroup$
    – Tyberius
    Jul 3, 2020 at 14:30
  • $\begingroup$ Mulliken analysis seems to be considered a method of random number generation because there is no basis set limit: mattermodeling.stackexchange.com/q/1439/5, so if you keep increasing the basis set (getting more and more accurate wavefunctions) you will keep getting different numbers, forever... even if your basis set is infinitely large. $\endgroup$ Jul 3, 2020 at 16:40


Your Answer

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