Dear Dulitha,

it is not very clear to me which MOs you'd swap and why. Most probably your complex has a very small HOMO-LUMO gap and you suspect that the these two FMO may be swapped in the initial MO guess. This is quite common in complexes of non-innocent ligands, such as 1,2-dithiolene ligands.

If this is the case, there are a number of way to improve your initial guess. The most typical procedure is to use the "guess=alter" keyword: you will want to add to your input file a last line describing which MOs are to be swapped. More details at https://gaussian.com/guess/.

Please not that on passing from G09 to G16 the initial guess is generally more accurate and the need for "altering" the MOs' order is much less stringent.

I hope this helps.

Massimiliano Arca Thank you very much your answer Sir. Actually I am using NWChem. My problem is I am working with Nb+ and It has 3 spin states, singlet, triplet and quintet . For the triplet and quintet states I am not getting the reasonable results. So I have to change the Initial MO order by swapping some MO orbitals which have more Nb contribution to metal-ligand bonds.

I see, I was referring to Gaussian. The unexpected results may be caused by an unreliable basis set. In general, I'd adopt a BS augmented with relativistic effective core potentials (RECPs), such as LANL08(f) (https://www.basissetexchange.org/basis/lanl08%28f%29/format/nwchem/?version=0&elements=41).

This said, someone more skilled than me with NwChem will be wanting to give you a specific hint. Good luck!