Chemistry - What is natural bond orbital theory used for?

Solution 1:

Yes, along with a family of related methods. I honestly don't fully understand the mechanics behind it, but the orbitals obtained with the NBO method can be used for a treatment called Natural Resonance Theory, which gives you an idea of what sort of Lewis-structure resonance components there are and in what weightings, as well as bond indices, and a measure of covalency.

There are three papers from the developers of the technique, covering theory, usage, and some more examples. (I'd probably start with the usage paper.)

Also, along with NBO, there are the Natural Atomic Orbitals which are often lumped in because they're also computed by the NBO program, and Natural Population Analysis which you can use the Natural Atomic Orbitals to do. These give you an idea of what sort of atomic orbital character is contributing to your molecular orbitals, and also atomic charges and net spins.

Solution 2:

The entire premise of NBO theory is to rotate canonical molecular orbitals into orbitals that have maximal chemical meaning, in the sense that is defined in the NBO papers. I am not aware of any purpose of NBOs and their corresponding natural charges beyond interpretation.

---

Solution 3:

(Fairness in reporting: I am a hard-core Quantum Theory of Atoms in Molecules partisan)

NBO is one of many attempts the apply rotations among molecular orbitals, each of which can spread over the entire molecule, (i.e. the wave function) so that they maximally-resemble localized atomic orbitals or "hybrid atomic orbitals". Then one may be able to talk about "pieces" of a overall state wave function, and draw connections to the historical descriptions of bonding.

But, the whole process of wave function localization/transformation is puzzling, because it leaves the observable--the electron density--just the way that it was, at least it ought to.

I believe that its motivation is good-intentioned. However, Schroedinger himself cautioned strongly against attempts to analyze the wave function directly. For example, since the wave function may be complex, what does one do with a localized wave function that has real and imaginary components? Why not just analyze the electron density (which is real and observable) directly?