Same final Configurations and different energies

[kambiz]

Hi dear All,

I tired to absorption of a molecule on a surface in two different sites (let s say X1 and X2). the favorable adsorption site for both cases was X2.

I mean that the molecule that was firstly placed in X1 site, moved to X2 and the molecule that was placed in X2 site remained in that site.

For both cases the final configuration were exactly the same. However, their energies were totally different (about 0.5 e.v.). I checked everything and the only difference was because of their final MAG MOMs.

Anyone can please tell me why their final MAG MOMs are different from each others while their optimized configurations are the same. I used same input files for both cases.

Thanks

[alex]

You are trapped in two different local minima of the electronic structure. The spin state remains the same, after the minimizer found a stable one.
Take the one with the less favourable total energy and fix the spin state with NUPDOWN and check, if the energy is the same afterwards (should be the case).

Cheers,

alex

[kambiz]

Thanks Dear Alex,

As you said, it seems that the system has two different local minima for the electronic structure. However, I couldnt understand where is the role of initial magnetic moments on the energy.
Should I test every system in different spin multiplicity?

I really appreciate if you elaborate on my problem.

Tnx
Kambiz

[alex]

Hi Kambiz,

the initial magnetic moment gives a different start when no value given. E.g. all unpaired spin up. This might be very unreasonable for e.g. metal oxides.
MAGMOM gives a starting magnetic moment and the minimizer tends to go to lower magnetic moment or it stays there. I haven't seen it going up. NUPDOWN simply fixes it leaving the minimiser to decide where to put electrons. You may also combine both flags.

And well, in principle, yes, you have to test all spin states. After some calculations you'll see that there might be only some of the spin states worth to consider.

Cheers,

alex

[kambiz]

Thanks Dear Alex,

That really helped me and solved my problem,

Kambiz