I saw somewhere that, "in PF5, the 3d orbitals of phosphorus end up in the anti-bonding molecular orbitals, and thus contribute less to the final molecule" and that "the hypervalency of some hypervalent molecules can be explained more efficiently by canonical forms involving ionic bonds, and some by involving d-orbitals".

I also saw somewhere that "the effect of d-orbitals on bonding is ALWAYS negligible in hypervalent molecules". (I said "negligible" since it cannot mathematically be zero)

My question is, is the first statement right or is the second one right? And, if the first one is right, is there a way to crudely estimate how much the d-orbitals affect bonds, for example a way based on the electronegativities/number of lone pairs(etc) of the subsituents attached to the hypervalent atom?

• If the antibonding MOs are not populated, then they don't describe the bonding. Empty orbitals don't exist. So there is no contradiction. – orthocresol Dec 2 '19 at 12:43

If we are talking about practical calculations, than yes, absolutely. d-Type basis function are a vital correction to p-orbitals. For $$\ce{PF5}$$ population analysis shows considerable projection of electron density on $$d$$-orbitals... but same applies to $$\ce{PF3}$$ [1]
1. Is bonding in $$\ce{PF5}$$ and the likes better described as extended on full octet.