Mullikan Charges Confusing

[kmkumar]

Dear All,

I am getting confused with charges computed by Turbomole.
Say for example for Pt, the electronic configuration is
4d10, 5s2, 5p6, 4f14, 5d9, 6s1.

I used def2-SVP and ECP(60 e)
that gives (2xS + 6xP + 10xd = 18 e-) electrons for the computation.
followed by ridft-proper using pop mo x-y (pop lall mo x-y)

Now, i have charges for Pt in PtCl4.

       Total          s              p              d               f
Pt    0.23841   2.81348   6.44446   8.47284   0.03081

Also for Pt in PtO2.
Pt    0.46431   3.22856   6.30666   7.92173   0.07874

Pt in PtCl4 and PtO2 have same oxidation state (+4).

1) Why do we get different numbers.
Assuming there is no backbonding from O to Pt.

2) Why do we get the occupation numbers more than
2 in S orbital and 6 in p orbital.
I can understand Mullikan charges are complicated,
but i used only small basis sets.

I checked also with lall option and no help.
Can anyone help me in understanding where i am
going wrong in understanding the concept.

Thanks in advance
KM

[antti_karttunen]

Hi,

That's how Mulliken charges usually are, confusing (and often plain wrong). I recommend to carry out a Natural population analysis instead. This is also possible in Turbomole ($pop nbo AO). Mulliken charges are very basis set dependent and often they cannot give even qualitatively correct description. See for example http://jcp.aip.org/resource/1/jcpsa6/v83/i2/p735_s1 for an analysis of Mulliken (and NPA) charges.

In your Pt case, you get more than two s electrons because you have both 5s and 6s electrons in the valence space. And the 6p orbitals seem to take some density, as well (but again, these are just Mulliken populations!). Also, trying to connect partial charges to the formal oxidation state is a bit dangerous effort. Perhaps it is possible to get some qualitative trends, but in general I would avoid making too strong conclusions...

Antti

[uwe]

Hi,

another option is to use population analysis based on occupation numbers (paboon). But, as Antti wrote, for anything but 'simple' main group chemistry all population analysis will be at least questionable. Paboon and NBO NPA can be easily performed using the graphical user interface TmoleX, they can be found in the start job panel as 'single point properties'.

Uwe